R: Difference between revisions

From 太極
Jump to navigation Jump to search
 
Line 1: Line 1:
== Install and upgrade R ==
= Install and upgrade R =
[[Install_R|Here]]
[[Install_R|Here]]


== Online Editor ==
== New release ==
* R 4.4.0
** [https://www.r-bloggers.com/2024/04/whats-new-in-r-4-4-0/ What’s new in R 4.4.0?]
** [https://www.r-bloggers.com/2024/05/cve-2024-27322-should-never-have-been-assigned-and-r-data-files-are-still-super-risky-even-in-r-4-4-0/ CVE-2024-27322 Should Never Have Been Assigned And R Data Files Are Still Super Risky Even In R 4.4.0], [https://www.ithome.com.tw/news/162626 程式開發語言R爆有程式碼執行漏洞,可用於供應鏈攻擊], [https://www.bleepingcomputer.com/news/security/r-language-flaw-allows-code-execution-via-rds-rdx-files/ R language flaw allows code execution via RDS/RDX files], [https://www.r-bloggers.com/2024/05/a-security-issue-with-r-serialization/ A security issue with R serialization] and the [https://cran.r-project.org/web/packages/RAppArmor/index.html RAppArmor] Package.
* R 4.3.0
** [https://www.jumpingrivers.com/blog/whats-new-r43/ What's new in R 4.3.0?]
** Extracting from a pipe. The underscore _ can be used to refer to the final value from a pipeline <code style="display:inline-block;">mtcars |> lm(mpg ~ disp, data = _) |> _$coef</code>. Previously we need to use [https://stackoverflow.com/a/56038303 this way] or [https://stackoverflow.com/a/60873298 this way]. If we want to apply some (anonymous) function to each element of a list, use '''map(), map_dbl()''' from the [https://purrr.tidyverse.org/ purrr].
* R 4.2.0
** Calling if() or while() with a condition of length greater than one gives an error rather than a warning.
** [https://twitter.com/henrikbengtsson/status/1501306369319735300 use underscore (_) as a placeholder on the right-hand side (RHS) of a forward pipe]. For example, '''mtcars |> subset(cyl == 4) |> lm(mpg ~ disp, data = _) '''
** [https://developer.r-project.org/Blog/public/2022/04/08/enhancements-to-html-documentation/ Enhancements to HTML Documentation]
** [https://www.jumpingrivers.com/blog/new-features-r420/ New features in R 4.2.0]
* R 4.1.0
** [https://developer.r-project.org/blosxom.cgi/R-devel/2021/01/13#n2021-01-13 pipe and shorthand for creating a function]
** [https://www.jumpingrivers.com/blog/new-features-r410-pipe-anonymous-functions/ New features in R 4.1.0] '''anonymous functions''' (lambda function)
* R 4.0.0
** [https://blog.revolutionanalytics.com/2020/04/r-400-is-released.html R 4.0.0 now available, and a look back at R's history]
** [https://www.infoworld.com/article/3540989/major-r-language-update-brings-big-changes.html R 4.0.0 brings numerous and significant changes to syntax, strings, reference counting, grid units, and more], [https://www.infoworld.com/article/3541368/how-to-run-r-40-in-docker-and-3-cool-new-r-40-features.html R 4.0: 3 new features]
**# factor is not default in data frame for character vector
**# palette() function has a new default set of colours, and [[R#New_palette_in_R_4.0.0|palette.colors() & palette.pals()]] are new
**# r"(YourString)" for ''raw'' character constants. See ?Quotes
* R 3.6.0
** [https://blog.revolutionanalytics.com/2019/05/whats-new-in-r-360.html What's new in R 3.6.0]
*** Changes to random number generation
*** More functions now support vectors with more than 2 billion elements
* R 3.5.0
** [https://community.rstudio.com/t/error-listing-packages-error-in-readrds-pfile-cannot-read-workspace-version-3-written-by-r-3-6-0/40570/2 The default serialization format for R changed in May 2018, such that new default format (version 3) for workspaces saved can no longer be read by versions of R older than 3.5]
 
= Online Editor =
We can run R on web browsers without installing it on local machines (similar to [/ideone.com Ideone.com] for C++. It does not require an account either (cf RStudio).  
We can run R on web browsers without installing it on local machines (similar to [/ideone.com Ideone.com] for C++. It does not require an account either (cf RStudio).  


=== [https://rdrr.io/snippets/ rdrr.io] ===
== [https://rdrr.io/snippets/ rdrr.io] ==
It can produce graphics too. The package I am testing ([https://www.rdocumentation.org/packages/cobs/versions/1.3-3/topics/cobs cobs]) is available too.
It can produce graphics too. The package I am testing ([https://www.rdocumentation.org/packages/cobs/versions/1.3-3/topics/cobs cobs]) is available too.


=== rstudio.cloud ===
== rstudio.cloud ==


=== [https://www.rdocumentation.org/ RDocumentation] ===
== [https://www.rdocumentation.org/ RDocumentation] ==
The interactive engine is based on [https://github.com/datacamp/datacamp-light DataCamp Light]
The interactive engine is based on [https://github.com/datacamp/datacamp-light DataCamp Light]


Line 21: Line 49:
The packages on RDocumentation may be outdated. For example, the current stringr on CRAN is v1.2.0 (2/18/2017) but RDocumentation has v1.1.0 (8/19/2016).
The packages on RDocumentation may be outdated. For example, the current stringr on CRAN is v1.2.0 (2/18/2017) but RDocumentation has v1.1.0 (8/19/2016).


== Web Applications ==
= Web Applications =
[[R_web|R web applications]]


See also CRAN Task View: [http://cran.r-project.org/web/views/WebTechnologies.html Web Technologies and Services]
= Creating local repository for CRAN and Bioconductor =
[[R_repository|R repository]]


=== TexLive ===
= Parallel Computing =
TexLive can be installed by 2 ways
See [[R_parallel|R parallel]].
* Ubuntu repository; does not include '''tlmgr''' utility for package manager.
* [http://tug.org/texlive/ Official website]


==== texlive-latex-extra ====
= Cloud Computing =
https://packages.debian.org/sid/texlive-latex-extra


For example, framed and titling packages are included.
== Install R on Amazon EC2 ==
http://randyzwitch.com/r-amazon-ec2/


==== tlmgr - TeX Live package manager ====
== Bioconductor on Amazon EC2 ==
https://www.tug.org/texlive/tlmgr.html
http://www.bioconductor.org/help/bioconductor-cloud-ami/


=== [https://yihui.name/tinytex/ TinyTex] ===
= Big Data Analysis =
https://github.com/yihui/tinytex
* [https://cran.r-project.org/web/views/HighPerformanceComputing.html CRAN Task View: High-Performance and Parallel Computing with R]
* [http://www.xmind.net/m/LKF2/ R for big data] in one picture
* [https://rstudio-pubs-static.s3.amazonaws.com/72295_692737b667614d369bd87cb0f51c9a4b.html Handling large data sets in R]
* [https://www.oreilly.com/library/view/big-data-analytics/9781786466457/#toc-start Big Data Analytics with R] by Simon Walkowiak
* [https://pbdr.org/publications.html pbdR]
** https://en.wikipedia.org/wiki/Programming_with_Big_Data_in_R
** [https://olcf.ornl.gov/wp-content/uploads/2016/01/pbdr.pdf Programming with Big Data in R - pbdR] George Ostrouchov and Mike Matheson Oak Ridge National Laboratory


=== [https://github.com/hadley/pkgdown pkgdown]: create a website for your package ===
== bigmemory, biganalytics, bigtabulate ==
[http://lbusettspatialr.blogspot.com/2017/08/building-website-with-pkgdown-short.html Building a website with pkgdown: a short guide]


=== Rmarkdown: create HTML5 web, slides and more ===
== ff, ffbase ==
[[Rmarkdown]]
* tapply does not work. [https://stackoverflow.com/questions/16470677/using-tapply-ave-functions-for-ff-vectors-in-r Using tapply, ave functions for ff vectors in R]
* [http://www.bnosac.be/index.php/blog/12-popularity-bigdata-large-data-packages-in-r-and-ffbase-user-presentation Popularity bigdata / large data packages in R and ffbase useR presentation]
* [http://www.bnosac.be/images/bnosac/blog/user2013_presentation_ffbase.pdf ffbase: statistical functions for large datasets] in useR 2013
* [https://www.rdocumentation.org/packages/ffbase/versions/0.12.7/topics/ffbase-package ffbase] package


=== [http://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol HTTP protocol] ===
== biglm ==


* http://en.wikipedia.org/wiki/File:Http_request_telnet_ubuntu.png
== data.table ==
* [http://en.wikipedia.org/wiki/Query_string Query string]
See [[Tidyverse#data.table|data.table]].
* How to capture http header? Use '''curl -i en.wikipedia.org'''.
* [http://trac.webkit.org/wiki/WebInspector Web Inspector]. Build-in in Chrome. Right click on any page and choose 'Inspect Element'.
* [http://en.wikipedia.org/wiki/Web_server Web server]
* [http://www.paulgriffiths.net/program/c/webserv.php Simple TCP/IP web server]
* [http://jmarshall.com/easy/http/ HTTP Made Really Easy]
* [http://www.manning.com/hethmon/ Illustrated Guide to HTTP]
* [http://www.ibm.com/developerworks/systems/library/es-nweb/ nweb: a tiny, safe Web server with 200 lines]
* [http://sourceforge.net/projects/tinyhttpd/ Tiny HTTPd]


An HTTP server is conceptually simple:
== disk.frame ==
[https://www.brodrigues.co/blog/2019-10-05-parallel_maxlik/ Split-apply-combine for Maximum Likelihood Estimation of a linear model]


# Open port 80 for listening
== Apache arrow ==
# When contact is made, gather a little information (get mainly - you can ignore the rest for now)
* https://arrow.apache.org/docs/r/
# Translate the request into a file request
* [https://www.infoworld.com/article/3637038/the-best-open-source-software-of-2021.html#slide17 The best open source software of 2021]
# Open the file and spit it back at the client


It gets more difficult depending on how much of HTTP you want to support - POST is a little more complicated, scripts, handling multiple requests, etc.
= Reproducible Research =
* http://cran.r-project.org/web/views/ReproducibleResearch.html
* [[Reproducible|Reproducible]]


==== Example in R ====
== Reproducible Environments ==
<syntaxhighlight lang='r'>
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
> co <- socketConnection(port=8080, server=TRUE, blocking=TRUE)
> # Now open a web browser and type http://localhost:8080/index.html
> readLines(co,1)
[1] "GET /index.html HTTP/1.1"
> readLines(co,1)
[1] "Host: localhost:8080"
> readLines(co,1)
[1] "User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0"
> readLines(co,1)
[1] "Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"
> readLines(co,1)
[1] "Accept-Language: en-US,en;q=0.5"
> readLines(co,1)
[1] "Accept-Encoding: gzip, deflate"
> readLines(co,1)
[1] "Connection: keep-alive"
> readLines(co,1)
[1] ""
</syntaxhighlight>


==== Example in C ([http://blog.abhijeetr.com/2010/04/very-simple-http-server-writen-in-c.html Very simple http server written in C], 187 lines) ====
== checkpoint package ==
* https://cran.r-project.org/web/packages/checkpoint/index.html
* [https://timogrossenbacher.ch/2017/07/a-truly-reproducible-r-workflow/ A (truly) reproducible R workflow]


Create a simple hello world html page and save it as <[http://en.wikipedia.org/wiki/List_of_Hello_world_program_examples#H index.html]> in the current directory (/home/brb/Downloads/)
== Some lessons in R coding ==
# don't use rand() and srand() in c. The result is platform dependent. My experience is Ubuntu/Debian/CentOS give the same result but they are different from macOS and Windows. Use [[Rcpp|Rcpp]] package and R's random number generator instead.
# don't use [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/list.files list.files()] directly. The result is platform dependent even different Linux OS. An extra [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/sort sorting] helps!


Launch the server program (assume we have done ''gcc http_server.c -o http_server'')
= Useful R packages =
<pre>
* [https://support.rstudio.com/hc/en-us/articles/201057987-Quick-list-of-useful-R-packages Quick list of useful R packages]
$ ./http_server -p 50002
* [https://github.com/qinwf/awesome-R awesome-R]
Server started at port no. 50002 with root directory as /home/brb/Downloads
* [https://stevenmortimer.com/one-r-package-a-day/ One R package a day]
</pre>


Secondly open a browser and type http://localhost:50002/index.html. The server will respond
== Rcpp ==
<pre>
http://cran.r-project.org/web/packages/Rcpp/index.html. See more [[Rcpp|here]].
GET /index.html HTTP/1.1
Host: localhost:50002
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive


file: /home/brb/Downloads/index.html
== RInside : embed R in C++ code ==
GET /favicon.ico HTTP/1.1
* http://dirk.eddelbuettel.com/code/rinside.html
Host: localhost:50002
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive


file: /home/brb/Downloads/favicon.ico
=== Ubuntu ===
GET /favicon.ico HTTP/1.1
With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on [http://www.youtube.com/watch?v=UQ8yKQcPTg0 Youtube]. I have tested this '''qtdensity''' example successfully using Qt 4.8.5.  
Host: localhost:50003
# Follow the instruction [[#cairoDevice|cairoDevice]] to install required libraries for cairoDevice package and then cairoDevice itself.
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
# Install [[Qt|Qt]]. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
# Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.  
Accept-Language: en-US,en;q=0.5
# Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the ''interactive'' program qtdensity appears on your desktop.
Accept-Encoding: gzip, deflate
Connection: keep-alive


file: /home/brb/Downloads/favicon.ico
[[:File:qtdensity.png]]
</pre>
The browser will show the page from <index.html> in server.


The only bad thing is the code does not close the port. For example, if I have use Ctrl+C to close the program and try to re-launch with the same port, it will complain '''socket() or bind(): Address already in use'''.
With RInside + [http://www.webtoolkit.eu/wt Wt web toolkit] installed, we can also create a web application. To demonstrate the example in ''examples/wt'' directory, we can do
 
 
==== Another Example in C (55 lines) ====
http://mwaidyanatha.blogspot.com/2011/05/writing-simple-web-server-in-c.html
 
The response is embedded in the C code.
 
If we test the server program by opening a browser and type "http://localhost:15000/", the server received the follwing 7 lines
<pre>
<pre>
GET / HTTP/1.1
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
Host: localhost:15000
make
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:23.0) Gecko/20100101 Firefox/23.0
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Connection: keep-alive
</pre>
</pre>
Then we can go to the browser's address bar and type ''http://localhost:8080'' to see how it works (a screenshot is in [http://dirk.eddelbuettel.com/blog/2011/11/30/ here]).


If we include a non-executable file's name in the url, we will be able to download that file. Try "http://localhost:15000/client.c".
=== Windows 7 ===
 
To make RInside works on Windows OS, try the following
If we use telnet program to test, wee need to type anything we want
# Make sure R is installed under '''C:\''' instead of '''C:\Program Files''' if we don't want to get an error like ''g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory''.
# Install RTools
# Instal RInside package from source (the binary version will give an [http://stackoverflow.com/questions/13137770/fatal-error-unable-to-open-the-base-package error ])
# Create a DOS batch file containing necessary paths in PATH environment variable
<pre>
<pre>
$ telnet localhost 15000
@echo off
Trying 127.0.0.1...
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
Connected to localhost.
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
Escape character is '^]'.
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
ThisCanBeAnything        <=== This is what I typed in the client and it is also shown on server
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
HTTP/1.1 200 OK          <=== From here is what I got from server
set R_HOME=C:\R\R-3.0.1
Content-length: 37Content-Type: text/html
echo Setting environment for using R
 
cmd
HTML_DATA_HERE_AS_YOU_MENTIONED_ABOVE <=== The html tags are not passed from server, interesting!
</pre>
Connection closed by foreign host.
In the Windows command prompt, run
$
<pre>
cd C:\R\R-3.0.1\library\RInside\examples\standard
make -f Makefile.win
</pre>
Now we can test by running any of executable files that '''make''' generates. For example, ''rinside_sample0''.
<pre>
rinside_sample0
</pre>
</pre>


See also more examples under [[C#Socket_Programming_Examples_using_C.2FC.2B.2B.2FQt|C page]].
As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See some discussions in
* http://stackoverflow.com/questions/12280707/using-rinside-with-qt-in-windows
* http://www.mail-archive.com/rcpp-devel@lists.r-forge.r-project.org/msg04377.html
So the Qt and Wt web tool applications on Windows may or may not be possible.


==== Others  ====
== GUI ==
* http://rosettacode.org/wiki/Hello_world/ (Different languages)
=== Qt and R ===
* http://kperisetla.blogspot.com/2012/07/simple-http-web-server-in-c.html (Windows web server)
* http://cran.r-project.org/web/packages/qtbase/index.html [https://stat.ethz.ch/pipermail/r-devel/2015-July/071495.html QtDesigner is such a tool, and its output is compatible with the qtbase R package]
* http://css.dzone.com/articles/web-server-c (handling HTTP GET request, handling content types(txt, html, jpg, zip. rar, pdf, php etc.), sending proper HTTP error codes, serving the files from a web root, change in web root in a config file, zero copy optimization using sendfile method and php file handling.)
* http://qtinterfaces.r-forge.r-project.org
* https://github.com/gtungatkar/Simple-HTTP-server
* https://github.com/davidmoreno/onion


=== shiny ===
== tkrplot ==
See [[Shiny|Shiny]].
On Ubuntu, we need to install tk packages, such as by
<pre>
sudo apt-get install tk-dev
</pre>


=== [https://www.rplumber.io/ plumber]: Turning your R code into a RESTful Web API ===
== reticulate - Interface to 'Python' ==
* https://github.com/trestletech/plumber
[[Python#R_and_Python:_reticulate_package|Python -> reticulate]]
* https://www.rstudio.com/resources/videos/plumber-turning-your-r-code-into-an-api/
* [https://blog.rstudio.com/2018/10/23/rstudio-1-2-preview-plumber-integration/ RStudio 1.2 Preview: Plumber Integration]
* [https://medium.com/@skyetetra/using-docker-to-deploy-an-r-plumber-api-863ccf91516d Using docker to deploy an R plumber API]


=== Docker ===
== Hadoop (eg ~100 terabytes) ==
* There are two major Docker images
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
** [https://hub.docker.com/_/r-base/?tab=tags Official] which supports version tags
** [https://hub.docker.com/r/rocker/r-base rocker project] which only has the latest tag
* [https://blog.ouseful.info/2016/05/03/using-docker-as-a-personal-productvity-tool-running-command-line-apps/ Using Docker as a Personal Productivity Tool – Running Command Line Apps Bundled in Docker Containers]
* [https://peerj.com/preprints/3181.pdf#page=8 Dockerized RStudio server] from Duke University. 110 containers were set up on a cloud server (4 cores, 28GB RAM, 400GB disk). Each container has its own port number. Each student is mapped to a single container. https://github.com/mccahill/docker-rstudio
* [http://sas-and-r.blogspot.com/2016/12/rstudio-in-cloud-with-amazon-lightsail.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+SASandR+%28SAS+and+R%29 RStudio in the cloud with Amazon Lightsail and docker]
* Mark McCahill (RStudio + Docker)
** http://sites.duke.edu/researchcomputing/files/2014/09/mccahill-DockerDays.pdf
** https://github.com/mccahill/docker-rstudio
** https://hub.docker.com/r/mccahill/rstudio/~/dockerfile/
* [https://github.com/Bioconductor-notebooks/BiocImageBuilder BiocImageBuilder]
** [https://github.com/Bioconductor-notebooks/Identification-of-Differentially-Expressed-Genes-for-Ectopic-Pregnancy/blob/master/CaseStudy1_EctopicPregnancy.ipynb Reproducible Bioconductor Workflow w/ browser-based interactive notebooks+Container].
** [http://biorxiv.org/content/early/2017/06/01/144816 Paper]
** Original [http://www.rna-seqblog.com/reproducible-bioconductor-workflows-using-browser-based-interactive-notebooks-and-containers/ post].
* [https://www.opencpu.org/posts/opencpu-with-docker/ Why Use Docker with R? A DevOps Perspective]
* [https://www.statworx.com/de/blog/running-your-r-script-in-docker/ Running your R script in Docker]. Goal: containerizing an R script to eventually execute it automatically each time the container is started, without any user interaction. An enhanced version of the instruction is at [https://github.com/arraytools/RinDocker this page].


=== [http://cran.r-project.org/web/packages/httpuv/index.html httpuv] ===
* RHadoop
http and WebSocket library.
* Hive
* [http://cran.r-project.org/web/packages/mapReduce/ MapReduce]. Introduction by [http://www.linuxjournal.com/content/introduction-mapreduce-hadoop-linux Linux Journal].
* http://www.techspritz.com/category/tutorials/hadoopmapredcue/ Single node or multinode cluster setup using Ubuntu with VirtualBox (Excellent)
* [http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-single-node-cluster/ Running Hadoop on Ubuntu Linux (Single-Node Cluster)]
* Ubuntu 12.04 http://www.youtube.com/watch?v=WN2tJk_oL6E and [https://www.dropbox.com/s/05aurcp42asuktp/Chiu%20Hadoop%20Pig%20Install%20Instructions.docx instruction]
* Linux Mint http://blog.hackedexistence.com/installing-hadoop-single-node-on-linux-mint
* http://www.r-bloggers.com/search/hadoop


See also the [https://cran.r-project.org/web/packages/servr/index.html servr] package which can start an HTTP server in R to serve static files, or dynamic documents that can be converted to HTML files (e.g., R Markdown) under a given directory.
=== [https://github.com/RevolutionAnalytics/RHadoop/wiki RHadoop] ===
 
* [http://www.rdatamining.com/tutorials/r-hadoop-setup-guide RDataMining.com] based on Mac.
=== [http://rapache.net/ RApache] ===
* Ubuntu 12.04 - [http://crishantha.com/wp/?p=1414 Crishantha.com], [http://nikhilshah123sh.blogspot.com/2014/03/setting-up-rhadoop-in-ubuntu-1204.html nikhilshah123sh.blogspot.com].[http://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/ Bighadoop.wordpress] contains an example.
* RapReduce in R by [https://github.com/RevolutionAnalytics/rmr2/blob/master/docs/tutorial.md RevolutionAnalytics] with a few examples.
* https://twitter.com/hashtag/rhadoop
* [http://bigd8ta.com/step-by-step-guide-to-setting-up-an-r-hadoop-system/ Bigd8ta.com] based on Ubuntu 14.04.


=== [http://cran.r-project.org/web/packages/gWidgetsWWW/index.html gWidgetsWWW] ===
=== Snowdoop: an alternative to MapReduce algorithm ===
 
* http://matloff.wordpress.com/2014/11/26/how-about-a-snowdoop-package/
* http://www.jstatsoft.org/v49/i10/paper
* http://matloff.wordpress.com/2014/12/26/snowdooppartools-update/comment-page-1/#comment-665
* [https://github.com/jverzani/gWidgetsWWW2 gWidgetsWWW2] gWidgetsWWW based on Rook
* [http://www.r-statistics.com/2012/11/comparing-shiny-with-gwidgetswww2-rapache/ Compare shiny with gWidgetsWWW2.rapache]
 
=== [http://cran.r-project.org/web/packages/Rook/index.html Rook] ===
 
Since R 2.13, the internal web server was exposed.
 
[https://docs.google.com/present/view?id=0AUTe_sntp1JtZGdnbjVicTlfMzFuZDQ5dmJxNw Tutorual from useR2012] and [https://github.com/rstats/RookTutorial Jeffrey Horner]
 
Here is another [http://www.rinfinance.com/agenda/2011/JeffHorner.pdf one] from http://www.rinfinance.com.
 
Rook is also supported by [rApache too. See http://rapache.net/manual.html.
 
Google group. https://groups.google.com/forum/?fromgroups#!forum/rrook
 
Advantage
* the web applications are created on desktop, whether it is Windows, Mac or Linux.
* No Apache is needed.
* create [http://jeffreyhorner.tumblr.com/post/4723187316/introducing-rook multiple applications] at the same time. This complements the limit of rApache.
 
----
 
4 lines of code [http://jeffreybreen.wordpress.com/2011/04/25/4-lines-of-r-to-get-you-started-using-the-rook-web-server-interface/ example].


== [http://cran.r-project.org/web/packages/XML/index.html XML] ==
On Ubuntu, we need to install libxml2-dev before we can install XML package.
<pre>
<pre>
library(Rook)
sudo apt-get update
s <- Rhttpd$new()
sudo apt-get install libxml2-dev
s$start(quiet=TRUE)
s$print()
s$browse(1)  # OR s$browse("RookTest")
</pre>
</pre>
Notice that after s$browse() command, the cursor will return to R because the command just a shortcut to open the web page http://127.0.0.1:10215/custom/RookTest.


[[File:Rook.png|100px]]
On CentOS,
[[File:Rook2.png|100px]]
[[File:Rookapprnorm.png|100px]]
 
We can add Rook '''application''' to the server; see ?Rhttpd.
<pre>
<pre>
s$add(
yum -y install libxml2 libxml2-devel
    app=system.file('exampleApps/helloworld.R',package='Rook'),name='hello'
</pre>
)
s$add(
    app=system.file('exampleApps/helloworldref.R',package='Rook'),name='helloref'
)
s$add(
    app=system.file('exampleApps/summary.R',package='Rook'),name='summary'
)


s$print()
=== XML ===
* http://giventhedata.blogspot.com/2012/06/r-and-web-for-beginners-part-ii-xml-in.html. It gave an example of extracting the XML-values from each XML-tag for all nodes and save them in a data frame using '''xmlSApply()'''.
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
* https://www.tutorialspoint.com/r/r_xml_files.htm
* https://www.datacamp.com/community/tutorials/r-data-import-tutorial#xml
* [http://www.stat.berkeley.edu/~statcur/Workshop2/Presentations/XML.pdf Extracting data from XML] PubMed and Zillow are used to illustrate. xmlTreeParse(),  xmlRoot(),  xmlName() and xmlSApply().
* https://yihui.name/en/2010/10/grabbing-tables-in-webpages-using-the-xml-package/
{{Pre}}
library(XML)


#Server started on 127.0.0.1:10221
# Read and parse HTML file
#[1] RookTest http://127.0.0.1:10221/custom/RookTest
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)
#[2] helloref http://127.0.0.1:10221/custom/helloref
#[3] summary  http://127.0.0.1:10221/custom/summary
#[4] hello    http://127.0.0.1:10221/custom/hello


# Stops the server but doesn't uninstall the app
# Extract all the paragraphs (HTML tag is p, starting at
## Not run:
# the root of the document). Unlist flattens the list to
s$stop()
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))


## End(Not run)
# Replace all by spaces
s$remove(all=TRUE)
doc.text = gsub('\n', ' ', doc.text)
rm(s)
 
# Join all the elements of the character vector into a single
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')
</pre>
</pre>
For example, the interface and the source code of ''summary'' app are given below


[[File:Rookappsummary.png|100px]]
This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.
{{Pre}}
> library(RCurl) # getURL()
> library(XML)  # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # <span class="css-truncate-target">2.5.3a</span>
> plain.text
[1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"   
[6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"


<nowiki>
> # try picard
app <- function(env) {
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
    req <- Rook::Request$new(env)
> doc = htmlParse(xData)
    res <- Rook::Response$new()
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
    res$write('Choose a CSV file:\n')
[1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
    res$write('<form method="POST" enctype="multipart/form-data">\n')
[10] "2.6.0"
    res$write('<input type="file" name="data">\n')
</pre>
    res$write('<input type="submit" name="Upload">\n</form>\n<br>')
This method can be used to monitor new tags/releases from some projects like [https://github.com/Ultimaker/Cura/releases Cura], BWA, Picard, [https://github.com/alexdobin/STAR/releases STAR]. But for some projects like [https://github.com/ncbi/sra-tools sratools] the '''class''' attribute in the '''span''' element ("css-truncate-target") can be different (such as "tag-name").


    if (!is.null(req$POST())){
=== xmlview ===
data <- req$POST()[['data']]
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/
res$write("<h3>Summary of Data</h3>");
res$write("<pre>")
res$write(paste(capture.output(summary(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
res$write("</pre>")
res$write("<h3>First few lines (head())</h3>");
res$write("<pre>")
res$write(paste(capture.output(head(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
res$write("</pre>")
    }
    res$finish()
}
</nowiki>


More example:
== RCurl ==
* http://lamages.blogspot.com/2012/08/rook-rocks-example-with-googlevis.html
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
* [http://www.road2stat.com/cn/r/rook.html Self-organizing map]
{{Pre}}
* Deploy Rook apps with rApache. [http://jeffreyhorner.tumblr.com/post/27861973339/deploy-rook-apps-with-rapache-part-i First one] and [http://jeffreyhorner.tumblr.com/post/33814488298/deploy-rook-apps-part-ii two].
# Test on Ubuntu 14.04
* [https://rud.is/b/2016/07/05/a-simple-prediction-web-service-using-the-new-firery-package/ A Simple Prediction Web Service Using the New fiery Package]
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</pre>


=== [https://code.google.com/p/sumo/ sumo] ===
=== Scrape google scholar results ===
Sumo is a fully-functional web application template that exposes an authenticated user's R session within java server pages. See the paper http://journal.r-project.org/archive/2012-1/RJournal_2012-1_Bergsma+Smith.pdf.
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R


=== [http://www.stat.ucla.edu/~jeroen/stockplot Stockplot] ===
No google ID is required


=== [http://www.rforge.net/FastRWeb/ FastRWeb] ===
Seems not work
http://cran.r-project.org/web/packages/FastRWeb/index.html
<pre>
 
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :
=== [http://sysbio.mrc-bsu.cam.ac.uk/Rwui/tutorial/Instructions.html Rwui] ===
  arguments imply differing number of rows: 2, 0
</pre>


=== [http://cran.r-project.org/web/packages/CGIwithR/index.html CGHWithR] and [http://cran.r-project.org/web/packages/WebDevelopR/ WebDevelopR] ===
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
CGHwithR is still working with old version of R although it is removed from CRAN. Its successor is WebDevelopR. Its The vignette (year 2013) provides a review of several available methods.
'''devtools''' package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the [[#remotes|remotes]] package which was suggested by the [https://cran.r-project.org/web/packages/BiocManager/index.html BiocManager] package.
{{Pre}}
# Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev


=== [http://www.rstudio.com/ide/docs/advanced/manipulate manipulate] from RStudio ===
# Ubuntu 16.04, 18.04
This is not a web application. But the '''manipulate''' package can be used to create interactive plot within R(Studio) environment easily. Its source is available at [https://github.com/rstudio/rstudio/tree/master/src/cpp/r/R/packages/manipulate here].
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev


Mathematica also has manipulate function for plotting; see [http://reference.wolfram.com/mathematica/tutorial/IntroductionToManipulate.html here].
# Ubuntu 20.04
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev
</pre>


=== [https://github.com/att/rcloud RCloud] ===
[https://github.com/wch/movies/issues/3 Lazy-load database XXX is corrupt. internal error -3]. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.
RCloud is an environment for collaboratively creating and sharing data analysis scripts. RCloud lets you mix analysis code in R, HTML5, Markdown, Python, and others. Much like Sage, iPython notebooks and Mathematica, RCloud provides a notebook interface that lets you easily record a session and annotate it with text, equations, and supporting images.


See also the [http://user2014.stat.ucla.edu/abstracts/talks/193_Harner.pdf Talk] in UseR 2014.
NB. According to the output of '''apt-cache show r-cran-devtools''', the binary package is very old though '''apt-cache show r-base''' and [https://cran.r-project.org/bin/linux/ubuntu/#supported-packages supported packages] like ''survival'' shows the latest version.


=== [https://github.com/cloudyr cloudyr] and [https://github.com/socialcopsdev/flyio flyio] - Input Output Files in R from Cloud or Local ===
=== [https://github.com/hadley/httr httr] ===
https://blog.socialcops.com/inside-sc/announcements/flyio-r-package-interact-data-cloud/ Announcing flyio, an R Package to Interact with Data in the Cloud]
httr imports curl, jsonlite, mime, openssl and R6 packages.


=== Dropbox access ===
When I tried to install httr package, I got an error and some message:
[https://cran.r-project.org/web/packages/rdrop2/index.html rdrop2] package
<pre>
Configuration failed because openssl was not found. Try installing:
* deb: libssl-dev (Debian, Ubuntu, etc)
* rpm: openssl-devel (Fedora, CentOS, RHEL)
* csw: libssl_dev (Solaris)
* brew: openssl (Mac OSX)
If openssl is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
--------------------------------------------------------------------
ERROR: configuration failed for package ‘openssl’
</pre>
It turns out after I run '''sudo apt-get install libssl-dev''' in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!


=== Web page scraping ===
Real example: see [http://stackoverflow.com/questions/27371372/httr-retrieving-data-with-post this post]. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).
http://www.slideshare.net/schamber/web-data-from-r#btnNext


==== [https://cran.r-project.org/web/packages/xml2/ xml2] package ====
Since httr package was used in many other packages, take a look at how others use it. For example, [https://github.com/ropensci/aRxiv aRxiv] package.
rvest package depends on xml2.


==== [https://cran.r-project.org/web/packages/purrr/index.html purrr]: Functional Programming Tools ====
[https://www.statsandr.com/blog/a-package-to-download-free-springer-books-during-covid-19-quarantine/ A package to download free Springer books during Covid-19 quarantine], [https://www.radmuzom.com/2020/05/03/an-update-to-an-adventure-in-downloading-books/ An update to "An adventure in downloading books"] (rvest package)
* https://purrr.tidyverse.org/
* [http://colinfay.me/purrr-cookbook/ purrr cookbook]
* Functional programming
** [http://www.youtube.com/watch?v=vLmaZxegahk Functional programming for beginners]
* [http://data.library.virginia.edu/getting-started-with-the-purrr-package-in-r/ Getting started with the purrr package in R], especially the [https://www.rdocumentation.org/packages/purrr/versions/0.2.5/topics/map map()] function.
* [http://staff.math.su.se/hoehle/blog/2019/01/04/mathgenius.html Purr yourself into a math genius]


==== [https://cran.r-project.org/web/packages/rvest/index.html rvest] ====
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
[http://blog.rstudio.org/2014/11/24/rvest-easy-web-scraping-with-r/ Easy web scraping with R]
curl is independent of RCurl package.


On Ubuntu, we need to install two packages first!
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
<syntaxhighlight lang='bash'>
* https://www.opencpu.org/posts/curl-release-0-8/
sudo apt-get install libcurl4-openssl-dev # OR libcurl4-gnutls-dev


sudo apt-get install libxml2-dev
{{Pre}}
</syntaxhighlight>
library(curl)
h <- new_handle()
handle_setform(h,
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)
</pre>


* https://github.com/hadley/rvest
=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
* [http://datascienceplus.com/visualizing-obesity-across-united-states-by-using-data-from-wikipedia/ Visualizing obesity across United States by using data from Wikipedia]
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
* [https://stat4701.github.io/edav/2015/04/02/rvest_tutorial/ rvest tutorial: scraping the web using R]
* https://renkun.me/pipeR-tutorial/Examples/rvest.html
* http://zevross.com/blog/2015/05/19/scrape-website-data-with-the-new-r-package-rvest/
* [https://datascienceplus.com/google-scholar-scraping-with-rvest/ Google scholar scraping with rvest package]


==== Animate ====
== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ==
* [https://guyabel.com/post/football-kits/ Animating Changes in Football Kits using R]: rvest, tidyverse, xml2, purrr & magick
Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).
* [https://guyabel.com/post/animated-directional-chord-diagrams/ Animated Directional Chord Diagrams] tweenr & magick
* [http://smarterpoland.pl/index.php/2019/01/x-mas-trees-with-gganimate-ggplot-plotly-and-friends/ x-mas tRees with gganimate, ggplot, plotly and friends]


==== [https://cran.r-project.org/web/packages/V8/index.html V8]: Embedded JavaScript Engine for R ====
Example:
[https://rud.is/b/2017/07/25/r%E2%81%B6-general-attys-distributions/ R⁶ — General (Attys) Distributions]: V8, rvest, ggbeeswarm, hrbrthemes and tidyverse packages are used.
{{Pre}}
# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')
</pre>


==== [http://cran.r-project.org/web/packages/pubmed.mineR/index.html pubmed.mineR] ====
== DirichletMultinomial ==
Text mining of PubMed Abstracts (http://www.ncbi.nlm.nih.gov/pubmed). The algorithms are designed for two formats (text and XML) from PubMed.
On Ubuntu, we do
<pre>
sudo apt-get install libgsl0-dev
</pre>


[https://github.com/jtleek/swfdr R code for scraping the P-values from pubmed, calculating the Science-wise False Discovery Rate, et al] (Jeff Leek)
== Create GUI ==
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===


=== These R packages import sports, weather, stock data and more ===
== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
https://www.computerworld.com/article/3109890/data-analytics/these-r-packages-import-sports-weather-stock-data-and-more.html
[http://statistical-research.com/simulating-random-multivariate-correlated-data-categorical-variables/?utm_source=rss&utm_medium=rss&utm_campaign=simulating-random-multivariate-correlated-data-categorical-variables here]


=== Diving Into Dynamic Website Content with splashr ===
== json ==
https://rud.is/b/2017/02/09/diving-into-dynamic-website-content-with-splashr/
[[R_web#json|R web -> json]]


=== Send email ===
== Map ==
==== [https://github.com/rpremraj/mailR/ mailR] ====
=== [https://rstudio.github.io/leaflet/ leaflet] ===
Easiest. Require rJava package (not trivial to install, see [[#RJava|rJava]]). mailR is an interface to Apache Commons Email to send emails from within R. See also [http://unamatematicaseltigre.blogspot.com/2016/12/how-to-send-bulk-email-to-your-students.html send bulk email]
* rstudio.github.io/leaflet/#installation-and-use
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/


Before we use the mailR package, we have followed [https://support.google.com/accounts/answer/6010255?hl=en here] to have '''Allow less secure apps: 'ON' '''; or you might get an error ''Error: EmailException (Java): Sending the email to the following server failed : smtp.gmail.com:465''. Once we turn on this option, we may get an email for the notification of this change. Note that the recipient can be other than a gmail.
=== choroplethr ===
<syntaxhighlight lang='rsplus'>
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
> send.mail(from = "sender@gmail.com",
* http://www.arilamstein.com/blog/2015/06/25/learn-to-map-census-data-in-r/
          to = c("[email protected]", "Recipient 2 <[email protected]>"),
* http://www.arilamstein.com/blog/2015/09/10/user-question-how-to-add-a-state-border-to-a-zip-code-map/
          replyTo = c("Reply to someone else <someone.else@gmail.com>")
          subject = "Subject of the email",
          body = "Body of the email",
          smtp = list(host.name = "smtp.gmail.com", port = 465, user.name = "gmail_username", passwd = "password", ssl = TRUE),
          authenticate = TRUE,
          send = TRUE)
[1] "Java-Object{org.apache.commons.mail.SimpleEmail@7791a895}"
</syntaxhighlight>


==== [https://cran.r-project.org/web/packages/gmailr/index.html gmailr] ====
=== ggplot2 ===
More complicated. gmailr provides access the Google's gmail.com RESTful API. [https://cran.r-project.org/web/packages/gmailr/vignettes/sending_messages.html Vignette] and an example on [http://stackoverflow.com/questions/30144876/send-html-message-using-gmailr here]. Note that it does not use a password; it uses a '''json''' file for oauth authentication downloaded from https://console.cloud.google.com/. See also https://github.com/jimhester/gmailr/issues/1.
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]
<syntaxhighlight lang='rsplus'>
library(gmailr)
gmail_auth('mysecret.json', scope = 'compose')


test_email <- mime() %>%
== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
  to("to@gmail.com") %>%
See an example from [[R#RJSONIO|RJSONIO]] above.
  from("[email protected]") %>%
  subject("This is a subject") %>%
  html_body("<html><body>I wish <b>this</b> was bold</body></html>")
send_message(test_email)
</syntaxhighlight>


==== [https://cran.r-project.org/web/packages/sendmailR/index.html sendmailR] ====  
== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
sendmailR provides a simple SMTP client. It is not clear how to use the package (i.e. where to enter the password).
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.


=== [http://www.ncbi.nlm.nih.gov/geo/ GEO (Gene Expression Omnibus)] ===
== gtrendsR - Google Trends ==
See [[GEO#R_packages|this internal link]].
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=ngs%20next%20generation%20sequencing ngs next generation sequencing] from 2004-04-01
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=dna%20sequencing dna sequencing] from 2004-01-01.
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=rna%20sequencing rna sequencing] from 2004-01-01. It can be seen RNA sequencing >> DNA sequencing.
* [http://www.kdnuggets.com/2017/09/python-vs-r-data-science-machine-learning.html?utm_content=buffere1df7&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer Python vs R – Who Is Really Ahead in Data Science, Machine Learning?] and [https://stackoverflow.blog/2017/09/06/incredible-growth-python/ The Incredible Growth of Python] by [https://twitter.com/drob?lang=en David Robinson]


=== Interactive html output ===
== quantmod ==
==== [http://cran.r-project.org/web/packages/sendplot/index.html sendplot] ====
[http://www.thertrader.com/2015/12/13/maintaining-a-database-of-price-files-in-r/ Maintaining a database of price files in R]. It consists of 3 steps.
==== [http://cran.r-project.org/web/packages/RIGHT/index.html RIGHT] ====
The supported plot types include scatterplot, barplot, box plot, line plot and pie plot.


In addition to tooltip boxes, the package can create a [http://righthelp.github.io/tutorial/interactivity table showing all information about selected nodes].
# Initial data downloading
# Update existing data
# Create a batch file


==== [http://cran.r-project.org/web/packages/d3Network/index.html d3Network] ====
== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
* http://christophergandrud.github.io/d3Network/ (old)
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
* https://christophergandrud.github.io/networkD3/ (new)
* https://www.r-project.org/conferences/useR-2013/Tutorials/kuhn/user_caret_2up.pdf
<source lang="rsplus">
* https://github.com/cran/caret source code mirrored on github
library(d3Network)
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/
* [https://daviddalpiaz.github.io/r4sl/the-caret-package.html Chapter 21 of "R for Statistical Learning"]


Source <- c("A", "A", "A", "A", "B", "B", "C", "C", "D")
== Tool for connecting Excel with R ==
Target <- c("B", "C", "D", "J", "E", "F", "G", "H", "I")
* https://bert-toolkit.com/
NetworkData <- data.frame(Source, Target)
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html


d3SimpleNetwork(NetworkData, height = 800, width = 1024, file="tmp.html")
== write.table ==
</source>
=== Output a named vector ===
<pre>
vec <- c(a = 1, b = 2, c = 3)
write.csv(vec, file = "my_file.csv", quote = F)
x = read.csv("my_file.csv", row.names = 1)
vec2 <- x[, 1]
names(vec2) <- rownames(x)
all.equal(vec, vec2)


==== [http://cran.r-project.org/web/packages/htmlwidgets/ htmlwidgets for R] ====
# one liner: row names of a 'matrix' become the names of a vector
Embed widgets in R Markdown documents and Shiny web applications.  
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
all.equal(vec, vec3)
</pre>


* Official website http://www.htmlwidgets.org/.
=== Avoid leading empty column to header ===
* [http://deanattali.com/blog/htmlwidgets-tips/ How to write a useful htmlwidgets in R: tips and walk-through a real example]
[https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
<pre>
write.table(a, 'a.txt', col.names=NA)
# Or better by
write.table(data.frame("SeqId"=rownames(a), a), "a.txt", row.names=FALSE)
</pre>


==== [http://cran.r-project.org/web/packages/networkD3/index.html networkD3] ====
=== Add blank field AND column names in write.table ===
This is a port of Christopher Gandrud's [http://christophergandrud.github.io/d3Network/ d3Network] package to the htmlwidgets framework.
* '''write.table'''(, row.names = TRUE) will miss one element on the 1st row when "row.names = TRUE" which is enabled by default.
** Suppose x is (n x 2)
** write.table(x, sep="\t") will generate a file with 2 element on the 1st row
** read.table(file) will return an object with a size (n x 2)
** read.delim(file) and read.delim2(file) will also be correct
* Note that '''write.csv'''() does not have this issue that write.table() has
** Suppose x is (n x 2)
** Suppose we use write.csv(x, file). The csv file will be ((n+1) x 3) b/c the header row.
** If we use read.csv(file), the object is (n x 3). So we need to use '''read.csv(file, row.names = 1)'''
* adding blank field AND column names in write.table(); [https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
:<syntaxhighlight lang="rsplus">
write.table(a, 'a.txt', col.names=NA)
</syntaxhighlight>
* '''readr::write_tsv'''() does not include row names in the output file


==== [http://cran.r-project.org/web/packages/scatterD3/index.html scatterD3] ====
=== read.delim(, row.names=1) and write.table(, row.names=TRUE) ===
scatterD3 is an HTML R widget for interactive scatter plots visualization. It is based on the htmlwidgets R package and on the d3.js javascript library.
[https://www.statology.org/read-delim-in-r/ How to Use read.delim Function in R]


==== [https://github.com/bwlewis/rthreejs rthreejs] - Create interactive 3D scatter plots, network plots, and globes ====
Case 1: no row.names
[http://bwlewis.github.io/rthreejs/ Examples]
<pre>
 
write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=FALSE)
==== d3heatmap ====
my_df <- read.delim('my_data.txt')  # the rownames will be 1, 2, 3, ...
See [[Heatmap#d3heatmap|R]]
</pre>
Case 2: with row.names. '''Note:''' if we open the text file in Excel, we'll see the 1st row is missing one header at the end. It is actually missing the column name for the 1st column.
<pre>
write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=TRUE)
my_df <- read.delim('my_data.txt')  # it will automatically assign the rownames
</pre>


==== [https://cran.r-project.org/web/packages/svgPanZoom/index.html svgPanZoom] ====
== Read/Write Excel files package ==
This 'htmlwidget' provides pan and zoom interactivity to R graphics, including 'base', 'lattice', and 'ggplot2'. The interactivity is provided through the 'svg-pan-zoom.js' library.
* http://www.milanor.net/blog/?p=779
 
* [https://www.displayr.com/how-to-read-an-excel-file-into-r/?utm_medium=Feed&utm_source=Syndication flipAPI]. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
==== DT: An R interface to the DataTables library ====
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
* http://blog.rstudio.org/2015/06/24/dt-an-r-interface-to-the-datatables-library/
** [https://stackoverflow.com/a/17976604 Export both Image and Data from R to an Excel spreadsheet]
 
* [http://cran.r-project.org/web/packages/openxlsx/index.html openxlsx]: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
==== plotly ====
** It can't be used to open .xls or .xlm files.
* [http://moderndata.plot.ly/power-curves-r-plotly-ggplot2/ Power curves] and ggplot2.
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
* [http://moderndata.plot.ly/time-series-charts-by-the-economist-in-r-using-plotly/ TIME SERIES CHARTS BY THE ECONOMIST IN R USING PLOTLY] & [https://moderndata.plot.ly/interactive-r-visualizations-with-d3-ggplot2-rstudio/ FIVE INTERACTIVE R VISUALIZATIONS WITH D3, GGPLOT2, & RSTUDIO]
** [https://fabiomarroni.wordpress.com/2018/08/07/use-r-to-write-multiple-tables-to-a-single-excel-file/ Use R to write multiple tables to a single Excel file]
* [http://moderndata.plot.ly/filled-chord-diagram-in-r-using-plotly/ Filled chord diagram]
* [https://github.com/hadley/readxl readxl]: it does not depend on anything although it can only read but not write Excel files. 
* [https://moderndata.plot.ly/dashboards-in-r-with-shiny-plotly/ DASHBOARDS IN R WITH SHINY & PLOTLY]
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
* [https://plot.ly/r/shiny-tutorial/ Plotly Graphs in Shiny],
** [https://github.com/rstudio/webinars/tree/master/36-readxl readxl webinar].
** [https://plot.ly/r/shiny-gallery/ Gallery]
** One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a '''tbl_df''', a '''tbl''' and a '''data.frame.'''
** [https://plot.ly/r/shinyapp-UN-simple/ Single time series]
** For writing to Excel formats, use writexl or openxlsx package.
** [https://plot.ly/r/shinyapp-UN-advanced/ Multiple time series]
:<syntaxhighlight lang='rsplus'>
* [https://www.r-exercises.com/2017/09/28/how-to-plot-basic-charts-with-plotly/ How to plot basic charts with plotly]
library(readxl)
* [https://www.displayr.com/how-to-add-trend-lines-in-r-using-plotly/?utm_medium=Feed&utm_source=Syndication How to add Trend Lines in R Using Plotly]
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE,
 
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf,
=== Amazon ===
    guess_max = min(1000, n_max), progress = readxl_progress(),
[https://github.com/56north/Rmazon Download product information and reviews from Amazon.com]
    .name_repair = "unique")
<syntaxhighlight lang='bash'>
# Example
sudo apt-get install libxml2-dev
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>
</syntaxhighlight>
and in R
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R
<syntaxhighlight lang='rsplus'>
:<syntaxhighlight lang='rsplus'>
install.packages("devtools")
library(writexl)
install.packages("XML")
mylst <- list(sheet1name = df1, sheet2name = df2)
install.packages("pbapply")
write_xlsx(mylst, "output.xlsx")
install.packages("dplyr")
devtools::install_github("56north/Rmazon")
product_info <- Rmazon::get_product_info("1593273843")
reviews <- Rmazon::get_reviews("1593273843")
reviews[1,6] # only show partial characters from the 1st review
nchar(reviews[1,6])
as.character(reviews[1,6]) # show the complete text from the 1st review
 
reviews <- Rmazon::get_reviews("B07BNGJXGS")
# Fetching 30 reviews of 'BOOX Note Ereader,Android 6.0 32 GB 10.3" Dual Touch HD Display'
#  |++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed = 02s
reviews
# A tibble: 30 x 6
  reviewRating reviewDate reviewFormat Verified_Purcha… reviewHeadline
          <dbl> <chr>      <lgl>        <lgl>            <chr>       
1            4 May 23, 2… NA          TRUE            Good for PDF …
2            3 May 8, 20… NA          FALSE            The reading s…
3            5 May 17, 2… NA          TRUE            E-reader and …
4            3 May 24, 2… NA          TRUE            Good hardware…
5            3 June 21, … NA          TRUE            Poor QC     
6            5 August 5,… NA          TRUE            Excellent for…
7            5 May 31, 2… NA          TRUE            Especially li…
8            5 July 4, 2… NA          TRUE            Android 6 rea…
9            4 July 15, … NA          TRUE            Remember the …
10            4 June 9, 2… NA          TRUE            Overall fanta…
# ... with 20 more rows, and 1 more variable: reviewText <chr>
reviews[1, 6] # 6-th column is the review text
</syntaxhighlight>
</syntaxhighlight>


=== [https://cran.r-project.org/web/packages/gutenbergr/index.html gutenbergr] ===
For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).
[https://blog.jumpingrivers.com/posts/2018/tidytext_edinbr_2018/ Edinbr: Text Mining with R]
{{Pre}}
> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
  UniqueID (Double-click) CloneID UGCluster
1                  HK1A1  21652 Hs.445981
2                  HK1A2  22012 Hs.119177
3                  HK1A4  22293 Hs.501376
                                                    Name Symbol EntrezID
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1    1495
2                              ADP-ribosylation factor 3  ARF3      377
3                          Uroporphyrinogen III synthase  UROS    7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1  5.000000        5q31.2              <NA>      1
2  12.000000        12q13              <NA>      1
3      <NA> 10q25.2-q26.3              <NA>      1
</pre>


=== Twitter ===
The hidden worksheets become visible (Not sure what are those first rows mean in the output).
[http://www.masalmon.eu/2017/03/19/facesofr/ Faces of #rstats Twitter]
{{Pre}}
 
> excel_sheets("~/Downloads/BRCA.xls")
=== OCR ===
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00
* [http://ropensci.org/blog/blog/2016/11/16/tesseract Tesseract package: High Quality OCR in R], [https://www.r-bloggers.com/how-to-do-optical-character-recognition-ocr-of-non-english-documents-in-r-using-tesseract/ How to do Optical Character Recognition (OCR) of non-English documents in R using Tesseract?]
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00
* https://cran.r-project.org/web/packages/abbyyR/index.html
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00
 
[1] "Experiment descriptors" "Filtered log ratio"    "Gene identifiers"     
=== Wikipedia ===
[4] "Gene annotations"      "CollateInfo"            "GeneSubsets"         
[https://github.com/ironholds/wikipedir WikipediR]: R's MediaWiki API client library
[7] "GeneSubsetsTemp"     
 
</pre>
== Creating local repository for CRAN and Bioconductor (focus on Windows binary packages only) ==
=== How to set up a local repository ===
 
* CRAN specific: http://cran.r-project.org/mirror-howto.html
* Bioconductor specific: http://www.bioconductor.org/about/mirrors/mirror-how-to/
* [https://rstudio.github.io/packrat/custom-repos.html How to Set Up a Custom CRAN-like Repository]
 
General guide: http://cran.r-project.org/doc/manuals/R-admin.html#Setting-up-a-package-repository


Utilities such as install.packages can be pointed at any CRAN-style repository, and R users may want to set up their own. The ‘base’ of a repository is a URL such as http://www.omegahat.org/R/: this must be an URL scheme that download.packages supports (which also includes ‘ftp://’ and ‘file://’, but not on most systems ‘https://’). '''Under that base URL there should be directory trees for one or more of the following types of package distributions:'''
The Chinese character works too.
{{Pre}}
> read_excel("~/Downloads/testChinese.xlsx", 1)
  中文 B C
1    a b c
2    1 2 3
</pre>
 
To read all worksheets we need a convenient function
{{Pre}}
read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc[[1]]) is a tibble.
</pre>


* "source": located at src/contrib and containing .tar.gz files. Other forms of compression can be used, e.g. .tar.bz2 or .tar.xz files.
=== [https://cran.r-project.org/web/packages/readr/ readr] ===
* '''"win.binary": located at bin/windows/contrib/x.y for R versions x.y.z and containing .zip files for Windows.'''
* "mac.binary.leopard": located at bin/macosx/leopard/contrib/x.y for R versions x.y.z and containing .tgz files.


Each terminal directory must also contain a PACKAGES file. This can be a concatenation of the DESCRIPTION files of the packages separated by blank lines, but only a few of the fields are needed. The simplest way to set up such a file is to use function write_PACKAGES in the tools package, and its help explains which fields are needed. Optionally there can also be a PACKAGES.gz file, a gzip-compressed version of PACKAGES—as this will be downloaded in preference to PACKAGES it should be included for large repositories. (If you have a mis-configured server that does not report correctly non-existent files you will need PACKAGES.gz.)
Compared to base equivalents like '''read.csv()''', '''readr''' is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.


To add your repository to the list offered by setRepositories(), see the help file for that function.
[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released. [https://www.tidyverse.org/blog/2021/07/readr-2-0-0/ readr 2.0.0] adds built-in support for reading multiple files at once, fast multi-threaded lazy reading and automatic guessing of delimiters among other changes.


A repository can contain subdirectories, when the descriptions in the PACKAGES file of packages in subdirectories must include a line of the form
Consider a [http://www.cs.utoronto.ca/~juris/data/cmapbatch/instmatx.21.txt text file] where the table (6100 x 22) has duplicated row names and the (1,1) element is empty. The column names are all unique.
* read.delim() will treat the first column as rownames but it does not allow duplicated row names. Even we use row.names=NULL, it still does not read correctly. It does give warnings (EOF within quoted string & number of items read is not a multiple of the number of columns). The dim is 5177 x 22.
* readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
* '''data.table::fread(Filename, sep = "\t")''' will detect the number of column names is less than the number of columns. Added 1 extra default column name for the first column which is guessed to be row names or an index. The dim is 6100 x 22. (Winner!)


<nowiki>Path: path/to/subdirectory</nowiki>
The '''readr::read_csv()''' function is as fast as '''data.table::fread()''' function. ''For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv().'' See 5.3 of Efficient R Programming book.


—once again write_PACKAGES is the simplest way to set this up.
Note that '''data.table::fread()''' can read a selection of the columns.


==== Space requirement if we want to mirror WHOLE repository ====
=== Speed comparison ===
* Whole CRAN takes about 92GB (rsync -avn  cran.r-project.org::CRAN > ~/Downloads/cran).
[https://predictivehacks.com/the-fastest-way-to-read-and-write-file-in-r/ The Fastest Way To Read And Write Files In R]. data.table >> readr >> base.
* Bioconductor is big (> 64G for BioC 2.11). Please check the size of what will be transferred with e.g. (rsync -avn bioconductor.org::2.11 > ~/Downloads/bioc) and make sure you have enough room on your local disk before you start.


On the other hand, we if only care about Windows binary part, the space requirement is largely reduced.
== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
* CRAN: 2.7GB
See [[Ggplot2|ggplot2]]
* Bioconductor: 28GB.


==== Misc notes ====
== Data Manipulation & Tidyverse ==
* If the binary package was built on R 2.15.1, then it cannot be installed on R 2.15.2. But vice is OK.
See [[Tidyverse|Tidyverse]].
* Remember to issue "--delete" option in rsync, otherwise old version of package will be installed.
* The repository still need src directory. If it is missing, we will get an error
<pre>
Warning: unable to access index for repository http://arraytools.no-ip.org/CRAN/src/contrib
Warning message:
package ‘glmnet’ is not available (for R version 2.15.2)
</pre>
The error was given by available.packages() function.


To bypass the requirement of src directory, I can use
== Data Science ==
<pre>
See [[Data_science|Data science]] page
install.packages("glmnet", contriburl = contrib.url(getOption('repos'), "win.binary"))
</pre>
but there may be a problem when we use biocLite() command.


I find a workaround. Since the error comes from missing CRAN/src directory, we just need to make sure the directory CRAN/src/contrib exists AND either CRAN/src/contrib/PACKAGES or CRAN/src/contrib/PACKAGES.gz exists.
== microbenchmark & rbenchmark ==
* [https://cran.r-project.org/web/packages/microbenchmark/index.html microbenchmark]
** [https://www.r-bloggers.com/using-the-microbenchmark-package-to-compare-the-execution-time-of-r-expressions/ Using the microbenchmark package to compare the execution time of R expressions]
* [https://cran.r-project.org/web/packages/rbenchmark/index.html rbenchmark] (not updated since 2012)


==== To create CRAN repository ====
== Plot, image ==
Before creating a local repository please give a dry run first. You don't want to be surprised how long will it take to mirror a directory.
=== [http://cran.r-project.org/web/packages/jpeg/index.html jpeg] ===
If we want to create the image on this wiki left hand side panel, we can use the '''jpeg''' package to read an existing plot and then edit and save it.


Dry run (-n option). Pipe out the process to a text file for an examination.
We can also use the jpeg package to import and manipulate a jpg image. See [http://moderndata.plot.ly/fun-with-heatmaps-and-plotly/ Fun with Heatmaps and Plotly].
<pre>
rsync -avn cran.r-project.org::CRAN > crandryrun.txt
</pre>
To mirror only partial repository, it is necessary to create directories before running rsync command.
<pre>
cd
mkdir -p ~/Rmirror/CRAN/bin/windows/contrib/2.15
rsync -rtlzv --delete cran.r-project.org::CRAN/bin/windows/contrib/2.15/ ~/Rmirror/CRAN/bin/windows/contrib/2.15
(one line with space before ~/Rmirror)


# src directory is very large (~27GB) since it contains source code for each R version.
=== EPS/postscript format ===
# We just need the files PACKAGES and PACKAGES.gz in CRAN/src/contrib. So I comment out the following line.
<ul>
# rsync -rtlzv --delete cran.r-project.org::CRAN/src/ ~/Rmirror/CRAN/src/
<li>Don't use postscript().  
mkdir -p ~/Rmirror/CRAN/src/contrib
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/PACKAGES ~/Rmirror/CRAN/src/contrib/
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/PACKAGES.gz ~/Rmirror/CRAN/src/contrib/
</pre>
And optionally
<pre>
library(tools)
write_PACKAGES("~/Rmirror/CRAN/bin/windows/contrib/2.15", type="win.binary")  
</pre>
and if we want to get src directory
<pre>
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/*.tar.gz ~/Rmirror/CRAN/src/contrib/
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/2.15.3 ~/Rmirror/CRAN/src/contrib/
</pre>


We can use '''du -h''' to check the folder size.  
<li>Use cairo_ps(). See [http://www.sthda.com/english/wiki/saving-high-resolution-ggplots-how-to-preserve-semi-transparency aving High-Resolution ggplots: How to Preserve Semi-Transparency]. It works on base R plots too.
<syntaxhighlight lang='r'>
cairo_ps(filename = "survival-curves.eps",
        width = 7, height = 7, pointsize = 12,
        fallback_resolution = 300)
print(p) # or any base R plots statements
dev.off()
</syntaxhighlight>


For example (as of 1/7/2013),
<li>[https://stackoverflow.com/a/8147482 Export a graph to .eps file with R].
* The results looks the same as using cairo_ps().
* The file size by setEPS() + postscript() is quite smaller compared to using cairo_ps().
* However, '''grep''' can find the characters shown on the plot generated by cairo_ps() but not setEPS() + postscript().
<pre>
<pre>
$ du -k ~/Rmirror --max-depth=1 --exclude ".*" | sort -nr | cut -f2 | xargs -d '\n' du -sh
setEPS()
30G /home/brb/Rmirror
postscript("whatever.eps") # 483 KB
28G /home/brb/Rmirror/Bioc
plot(rnorm(20000))
2.7G /home/brb/Rmirror/CRAN
dev.off()
# grep rnorm whatever.eps # Not found!
 
cairo_ps("whatever_cairo.eps")  # 2.4 MB
plot(rnorm(20000))
dev.off()
# grep rnorm whatever_cairo.eps  # Found!
</pre>
</pre>


==== To create Bioconductor repository ====
<li> View EPS files
Dry run
* Linux: evince. It is installed by default.
<pre>
* Mac: evince. ''' brew install evince'''
rsync -avn bioconductor.org::2.11 > biocdryrun.txt
* Windows. Install '''ghostscript''' [https://www.npackd.org/p/com.ghostscript.Ghostscript64/9.20 9.20] (10.x does not work with ghostview/GSview) and '''ghostview/GSview''' (5.0). In Ghostview, open Options -> Advanced Configure. Change '''Ghostscript DLL''' path AND '''Ghostscript include Path''' according to the ghostscript location ("C:\.
</pre>
Then creates directories before running rsync.  


<syntaxhighlight lang='bash'>
<li>Edit EPS files: Inkscape
cd
* Step 1: open the EPS file
mkdir -p ~/Rmirror/Bioc
* Step 2: EPS Input: Determine page orientation from text direction 'Page by page' - OK
wget -N http://www.bioconductor.org/biocLite.R -P ~/Rmirror/Bioc
* Step 3: PDF Import Settings: default is "Internal import", but we shall choose '''"Cairo import"'''.  
</syntaxhighlight>
* Step 4: '''Zoom in''' first.
where '''-N''' is to overwrite original file if the size or timestamp change and '''-P''' in wget means an output directory, not a file name.
* Step 5: Click on '''Layers and Objects''' tab on the RHS. Now we can select any lines or letters and edit them as we like. The selected objects are highlighted in the "Layers and Objects" panel. That is, we can select multiple objects using object names. The selected objects can be rotated (Object -> Rotate 90 CW), for example.
* Step 6: We can save the plot as any formats like svg, eps, pdf, html, pdf, ...
</ul>


Optionally, we can add the following in order to see the Bioconductor front page.
=== png and resolution ===
<syntaxhighlight lang='bash'>
It seems people use '''res=300''' as a definition of high resolution.  
rsync -zrtlv  --delete bioconductor.org::2.11/BiocViews.html ~/Rmirror/Bioc/packages/2.11/
rsync -zrtlv  --delete bioconductor.org::2.11/index.html ~/Rmirror/Bioc/packages/2.11/
</syntaxhighlight>


The software part (aka bioc directory) installation:
<ul>
<syntaxhighlight lang='bash'>
<li>Bottom line: fix res=300 and adjust height/width as needed. The default is res=72, height=width=480. If we increase res=300, the text font size will be increased, lines become thicker and the plot looks like a zoom-in.
cd
<li>[https://stackoverflow.com/a/51194014 Saving high resolution plot in png].  
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/bin/windows
<pre>
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/src
png("heatmap.png", width = 8, height = 6, units='in', res = 300)
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/bin/windows/ ~/Rmirror/Bioc/packages/2.11/bioc/bin/windows
# we can adjust width/height as we like
# Either rsync whole src directory or just essential files
# the pixel values will be width=8*300 and height=6*300 which is equivalent to
# rsync -zrtlv  --delete bioconductor.org::2.11/bioc/src/ ~/Rmirror/Bioc/packages/2.11/bioc/src
# 8*300 * 6*300/10^6 = 4.32 Megapixels (1M pixels = 10^6 pixels) in camera's term
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/src/contrib/PACKAGES ~/Rmirror/Bioc/packages/2.11/bioc/src/contrib/
# However, if we use png(, width=8*300, height=6*300, units='px'), it will produce
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/src/contrib/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/bioc/src/contrib/
# a plot with very large figure body and tiny text font size.
# Optionally the html part
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/html
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/html/ ~/Rmirror/Bioc/packages/2.11/bioc/html
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/vignettes
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/vignettes/ ~/Rmirror/Bioc/packages/2.11/bioc/vignettes
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/news
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/news/ ~/Rmirror/Bioc/packages/2.11/bioc/news
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/licenses
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/licenses/ ~/Rmirror/Bioc/packages/2.11/bioc/licenses
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/manuals
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/manuals/ ~/Rmirror/Bioc/packages/2.11/bioc/manuals
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/readmes
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/readmes/ ~/Rmirror/Bioc/packages/2.11/bioc/readmes
</syntaxhighlight>
and annotation (aka data directory) part:
<syntaxhighlight lang='bash'>
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/annotation/bin/windows
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/annotation/src/contrib
# one line for each of the following
rsync -zrtlv --delete bioconductor.org::2.11/data/annotation/bin/windows/ ~/Rmirror/Bioc/packages/2.11/data/annotation/bin/windows
rsync -zrtlv --delete bioconductor.org::2.11/data/annotation/src/contrib/PACKAGES ~/Rmirror/Bioc/packages/2.11/data/annotation/src/contrib/
rsync -zrtlv --delete bioconductor.org::2.11/data/annotation/src/contrib/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/data/annotation/src/contrib/
</syntaxhighlight>
and experiment directory:
<syntaxhighlight lang='bash'>
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/experiment/src/contrib
# one line for each of the following
# Note that we are cheating by only downloading PACKAGES and PACKAGES.gz files
rsync -zrtlv --delete bioconductor.org::2.11/data/experiment/bin/windows/contrib/2.15/PACKAGES ~/Rmirror/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15/
rsync -zrtlv --delete bioconductor.org::2.11/data/experiment/bin/windows/contrib/2.15/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15/
rsync -zrtlv --delete bioconductor.org::2.11/data/experiment/src/contrib/PACKAGES ~/Rmirror/Bioc/packages/2.11/data/experiment/src/contrib/
rsync -zrtlv --delete bioconductor.org::2.11/data/experiment/src/contrib/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/data/experiment/src/contrib/
</syntaxhighlight>
and extra directory:
<syntaxhighlight lang='bash'>
mkdir -p ~/Rmirror/Bioc/packages/2.11/extra/bin/windows/contrib/2.15
mkdir -p ~/Rmirror/Bioc/packages/2.11/extra/src/contrib
# one line for each of the following
# Note that we are cheating by only downloading PACKAGES and PACKAGES.gz files
rsync -zrtlv --delete bioconductor.org::2.11/extra/bin/windows/contrib/2.15/PACKAGES ~/Rmirror/Bioc/packages/2.11/extra/bin/windows/contrib/2.15/
rsync -zrtlv --delete bioconductor.org::2.11/extra/bin/windows/contrib/2.15/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/extra/bin/windows/contrib/2.15/
rsync -zrtlv --delete bioconductor.org::2.11/extra/src/contrib/PACKAGES ~/Rmirror/Bioc/packages/2.11/extra/src/contrib/
rsync -zrtlv --delete bioconductor.org::2.11/extra/src/contrib/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/extra/src/contrib/
</syntaxhighlight>


==== sync Bioconductor software packages ====
# It seems the following command gives the same result as above
To keep a copy of the bioc/source (software packages) code only,
png("heatmap.png", width = 8*300, height = 6*300, res = 300) # default units="px"
<syntaxhighlight lang='bash'>
</pre>
$ mkdir -p ~/bioc_release/bioc/
<li>Chapter 14.5 [https://r-graphics.org/recipe-output-bitmap Outputting to Bitmap (PNG/TIFF) Files] by R Graphics Cookbook
$ rsync -zrtlv --delete master.bioconductor.org::release/bioc/src ~/bioc_release/bioc/
* Changing the resolution affects the size (in pixels) of graphical objects like text, lines, and points.
<li>[https://blog.revolutionanalytics.com/2009/01/10-tips-for-making-your-r-graphics-look-their-best.html 10 tips for making your R graphics look their best] David Smith
* In Word you can resize the graphic to an appropriate size, but the high resolution gives you the flexibility to choose a size while not compromising on the quality.  I'd recommend '''at least 1200 pixels''' on the longest side for standard printers.
<li>[https://stat.ethz.ch/R-manual/R-devel/library/grDevices/html/png.html ?png]. The png function has default settings ppi=72, height=480, width=480, units="px".
* By default no resolution is recorded in the file, except for BMP.
* [https://www.adobe.com/creativecloud/file-types/image/comparison/bmp-vs-png.html BMP vs PNG format]. If you need a smaller file size and don’t mind a lossless compression, PNG might be a better choice. If you need to retain as much detail as possible and don’t mind a larger file size, BMP could be the way to go.
** '''Compression''': BMP files are raw and uncompressed, meaning they’re large files that retain as much detail as possible. On the other hand, PNG files are compressed but still lossless. This means you can reduce or expand PNGs without losing any information.
** '''File size''': BMPs are larger than PNGs. This is because PNG files automatically compress, and can be compressed again to make the file even smaller.
** '''Common uses''': BMP contains a maximum amount of details while PNGs are good for small illustrations, sketches, drawings, logos and icons.
** '''Quality''': No difference
** '''Transparency''': PNG supports transparency while BMP doesn't
<li>Some comparison about the ratio
* 11/8.5=1.29  (A4 paper)
* 8/6=1.33    (plot output)
* 1440/900=1.6 (my display)
<li>[https://babichmorrowc.github.io/post/2019-05-23-highres-figures/ Setting resolution and aspect ratios in R]
<li>The difference of '''res''' parameter for a simple plot. [https://www.tutorialspoint.com/how-to-change-the-resolution-of-a-plot-in-base-r How to change the resolution of a plot in base R?]
<li>[https://danieljhocking.wordpress.com/2013/03/12/high-resolution-figures-in-r/ High Resolution Figures in R].
<li>[https://magesblog.com/post/2013-10-29-high-resolution-graphics-with-r/ High resolution graphics with R]
<li>[https://stackoverflow.com/questions/8399100/r-plot-size-and-resolution R plot: size and resolution]
<li>[https://stackoverflow.com/a/22815896 How can I increase the resolution of my plot in R?], [https://cran.r-project.org/web/packages/devEMF/index.html devEMF] package
<li>See [[Images#Anti-alias_%E4%BF%AE%E9%82%8A|Images -> Anti-alias]].
<li>How to check DPI on PNG
* '''The width of a PNG file in terms of inches cannot be determined directly from the file itself''', as the file contains pixel dimensions, not physical dimensions. However, '''you can calculate the width in inches if you know the resolution (DPI, dots per inch) of the image'''. Remember that converting pixel measurements to physical measurements like inches involves a specific resolution (DPI), and different devices may display the same image at different sizes due to having different resolutions.
<li>[https://community.rstudio.com/t/save-high-resolution-figures-from-r-300dpi/62016/3 Cairo] case.
</ul>


$ du -h ~/bioc_release/bioc/
=== PowerPoint ===
# 20GB, 1565 items, Bioc 3.7
<ul>
</syntaxhighlight>
<li>For PP presentation, I found it is useful to use svg() to generate a small size figure. Then when we enlarge the plot, the text font size can be enlarged too. According to [https://www.rdocumentation.org/packages/grDevices/versions/3.6.2/topics/cairo svg], by default, width = 7, height = 7, pointsize = 12, family = '''sans'''.
Note ''-z'' - compress file data during the transfer, ''-t'' - preserve modification times, ''-l'' copy symbolic links as symbolic links. The option ''-zrtlv'' can be replaced by the common options ''-avz''.
<li>Try the following code. The font size is the same for both plots/files. However, the first plot can be enlarged without losing its quality.
<pre>
svg("svg4.svg", width=4, height=4)
plot(1:10, main="width=4, height=4")
dev.off()


To get the old versions of a packages (after the release of a version of Bioconductor), check out the ''Archive'' folder.
svg("svg7.svg", width=7, height=7) # default
plot(1:10, main="width=7, height=7")
dev.off()
</pre>
</ul>


Now we can create a cron job to do sync. ''' ''Note'' ''' my observation is Bioconductor has a daily update around 10:45AM. So I set time at 11:00AM.
=== magick ===
<syntaxhighlight lang='bash'>
https://cran.r-project.org/web/packages/magick/
echo "00 11 * * * rsync -avz --delete master.bioconductor.org::release/bioc/src ~/bioc_release/bioc/" >> \
  ~/Documents/cronjob  # everyday at 6am & 1pm
crontab ~/Documents/cronjob
crontab -l
</syntaxhighlight>


=== To test local repository ===
See an example [[:File:Progpreg.png|here]] I created.


==== Create soft links in Apache server ====
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
<pre>
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
su
ln -s /home/brb/Rmirror/CRAN /var/www/html/CRAN
ln -s /home/brb/Rmirror/Bioc /var/www/html/Bioc
ls -l /var/www/html
</pre>
The soft link mode should be 777.


==== To test CRAN ====
=== geDevices ===
Replace the host name arraytools.no-ip.org by IP address 10.133.2.111 if necessary.
* [https://www.jumpingrivers.com/blog/r-graphics-cairo-png-pdf-saving/ Saving R Graphics across OSs]. Use png(type="cairo-png") or the [https://cran.r-project.org/web/packages/ragg/index.html ragg] package which can be incorporated into RStudio.
* [https://www.jumpingrivers.com/blog/r-knitr-markdown-png-pdf-graphics/ Setting the Graphics Device in a RMarkdown Document]


<pre>
=== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ===
r <- getOption("repos"); r["CRAN"] <- "http://arraytools.no-ip.org/CRAN"
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
options(repos=r)
install.packages("glmnet")
</pre>
We can test if the backup server is working or not by installing a package which was removed from the CRAN. For example, 'ForImp' was removed from CRAN in 11/8/2012, but I still a local copy built on R 2.15.2 (run rsync on 11/6/2012).


For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
<pre>
<pre>
r <- getOption("repos"); r["CRAN"] <- "http://cran.r-project.org"
sudo apt-get install libgtk2.0-dev libcairo2-dev
r <- c(r, BRB='http://arraytools.no-ip.org/CRAN')
#                        CRAN                            CRANextra                                  BRB
# "http://cran.r-project.org" "http://www.stats.ox.ac.uk/pub/RWin"  "http://arraytools.no-ip.org/CRAN"
options(repos=r)
install.packages('ForImp')
</pre>
</pre>


Note by default, CRAN mirror is selected interactively.
On Windows OS, we may got the error: '''unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' '''. We need to follow the instruction in [http://tolstoy.newcastle.edu.au/R/e6/help/09/05/15613.html here].
<pre>
> getOption("repos")
                                CRAN                            CRANextra
                            "@CRAN@" "http://www.stats.ox.ac.uk/pub/RWin"
</pre>


==== To test Bioconductor ====
=== dpi requirement for publication ===
<pre>
[http://www.cookbook-r.com/Graphs/Output_to_a_file/ For import into PDF-incapable programs (MS Office)]
# CRAN part:
r <- getOption("repos"); r["CRAN"] <- "http://arraytools.no-ip.org/CRAN"
options(repos=r)
# Bioconductor part:
options("BioC_mirror" = "http://arraytools.no-ip.org/Bioc")
source("http://bioconductor.org/biocLite.R")
# This source biocLite.R line can be placed either before or after the previous 2 lines
biocLite("aCGH")
</pre>


If there is a connection problem, check folder attributes.
=== sketcher: photo to sketch effects ===
<pre>
https://htsuda.net/sketcher/
chmod -R 755 ~/CRAN/bin
</pre>


* Note that if a binary package was created for R 2.15.1, then it can be installed under R 2.15.1 but not R 2.15.2. The R console will show package xxx is not available (for R version 2.15.2).
=== httpgd ===
* https://nx10.github.io/httpgd/ A graphics device for R that is accessible via network protocols. Display graphics on browsers.
* [https://youtu.be/uxyhmhRVOfw Three tricks to make IDEs other than Rstudio better for R development]


* For binary installs, the function also checks for the availability of a source package on the same repository, and reports if the source package has a later version, or is available but no binary version is.
== [http://igraph.org/r/ igraph] ==
So for example, if the mirror does not have contents under src directory, we need to run the following line in order to successfully run ''install.packages()'' function.
[[R_web#igraph|R web -> igraph]]
<pre>
options(install.packages.check.source = "no")
</pre>


* If we only mirror the essential directories, we can run biocLite() successfully. However, the R console will give some warning
== Identifying dependencies of R functions and scripts ==
<pre>
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
> biocLite("aCGH")
{{Pre}}
BioC_mirror: http://arraytools.no-ip.org/Bioc
library(mvbutils)
Using Bioconductor version 2.11 (BiocInstaller 1.8.3), R version 2.15.
foodweb(where = "package:batr")
Installing package(s) 'aCGH'
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/data/experiment/src/contrib
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/extra/src/contrib
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/extra/bin/windows/contrib/2.15
trying URL 'http://arraytools.no-ip.org/Bioc/packages/2.11/bioc/bin/windows/contrib/2.15/aCGH_1.36.0.zip'
Content type 'application/zip' length 2431158 bytes (2.3 Mb)
opened URL
downloaded 2.3 Mb


package ‘aCGH’ successfully unpacked and MD5 sums checked
foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)


The downloaded binary packages are in
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
        C:\Users\limingc\AppData\Local\Temp\Rtmp8IGGyG\downloaded_packages
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/extra/bin/windows/contrib/2.15
> library()
</pre>
</pre>


=== CRAN repository directory structure ===
== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
The information below is specific to R 2.15.2. There are linux and macosx subdirecotries whenever there are windows subdirectory.
Iterator is useful over for-loop if the data is already a '''collection'''. It can be used to iterate over a vector, data frame, matrix, file
 
Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
 
== Colors ==
* [https://scales.r-lib.org/ scales] package. This is used in ggplot2 package.
<ul>
<li>[https://cran.r-project.org/web/packages/colorspace/index.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes. Popular! Many reverse imports/suggests; e.g. ComplexHeatmap. See my [[Ggplot2#colorspace_package|ggplot2]] page.  
<pre>
<pre>
bin/winows/contrib/2.15
hcl_palettes(plot = TRUE) # a quick overview
src/contrib
hcl_palettes(palette = "Dark 2", n=5, plot = T)
  /contrib/2.15.2
q4 <- qualitative_hcl(4, palette = "Dark 3")
  /contrib/Archive
</pre>
web/checks
</ul>
  /dcmeta
* [https://statisticsglobe.com/create-color-range-between-two-colors-in-r Create color range between two colors in R] using colorRampPalette()
  /packages
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
  /views
* palette_explorer() function from the [https://cran.r-project.org/web/packages/tmaptools/index.html tmaptools] package. See [https://www.computerworld.com/article/3184778/data-analytics/6-useful-r-functions-you-might-not-know.html selecting color palettes with shiny].
* [http://www.cookbook-r.com/ Cookbook for R]
* [http://ggplot2.tidyverse.org/reference/scale_brewer.html Sequential, diverging and qualitative colour scales/palettes from colorbrewer.org]: scale_colour_brewer(), scale_fill_brewer(), ...
* http://colorbrewer2.org/
* It seems there is no choice of getting only 2 colors no matter which set name we can use
* To see the set names used in brewer.pal, see
** [https://www.rdocumentation.org/packages/RColorBrewer/versions/1.1-2/topics/RColorBrewer RColorBrewer::display.brewer.all()]
** [https://rpubs.com/flowertear/224344 Output]
** Especially, '''[http://colorbrewer2.org/#type=qualitative&scheme=Set1&n=4 Set1]''' from http://colorbrewer2.org/
* To list all R color names, colors().
** [http://research.stowers.org/mcm/efg/R/Color/Chart/ColorChart.pdf Color Chart] (include Hex and RGB) & [http://research.stowers.org/mcm/efg/Report/UsingColorInR.pdf Using Color in R] from http://research.stowers.org
** Code to generate rectangles with colored background https://www.r-graph-gallery.com/42-colors-names/
* http://www.bauer.uh.edu/parks/truecolor.htm Interactive RGB, Alpha and Color Picker
* http://deanattali.com/blog/colourpicker-package/ Not sure what it is doing
* [http://www.lifehack.org/484519/how-to-choose-the-best-colors-for-your-data-charts How to Choose the Best Colors For Your Data Charts]
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* [http://sape.inf.usi.ch/quick-reference/ggplot2/colour Color names in R]
<ul>
<li>[https://stackoverflow.com/questions/28461326/convert-hex-color-code-to-color-name convert hex value to color names]
{{Pre}}
library(plotrix)
sapply(rainbow(4), color.id) # color.id is a function
          # it is used to identify closest match to a color
sapply(palette(), color.id)
sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
</pre>
</pre>
</li></ul>
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.3/topics/hsv hsv()] function. [https://eranraviv.com/matrix-style-screensaver-in-r/ Matrix-style screensaver in R]


A clickable map [http://taichi.selfip.net:81/RmirrorMap/Rmirror.html]
Below is an example using the option ''scale_fill_brewer''(palette = "[http://colorbrewer2.org/#type=qualitative&scheme=Paired&n=9 Paired]"). See the source code at [https://gist.github.com/JohannesFriedrich/c7d80b4e47b3331681cab8e9e7a46e17 gist]. Note that only '''set1''' and '''set3''' palettes in '''qualitative scheme''' can support up to 12 classes.


=== CRAN package download statistics from RStudio ===
According to the information from the colorbrew website, '''qualitative''' schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.
* Daily download statistics http://cran-logs.rstudio.com/. Note the page is split into 'package' download and 'R' download. It tracks
** Package: date, time, size, r_version, r_arch, r_os, package, version, country, ip_id.
** R: date, time, size, R version, os (win/src/osx), county, ip_id (reset daily).
* https://www.r-bloggers.com/finally-tracking-cran-packages-downloads/. The code still works.
** [https://gist.github.com/arraytools/f037a4fd90ae8fe8e2896db6fe5b7fad My modified code] for showing the top download R packages. The original code suffers from memory issue (tested on my 64GB Linux box) when it is calling rbindlist() from data.table package with large data. Still 64GB is required since the matrix is 369k by 4 (12GB).
** [http://www.bnosac.be/index.php/blog/12-popularity-bigdata-large-data-packages-in-r-and-ffbase-user-presentation Popularity bigdata / large data packages in R and ffbase useR presentation]
* https://strengejacke.wordpress.com/2015/03/07/cran-download-statistics-of-any-packages-rstats/


=== Bioconductor package download statistics ===
[[:File:GgplotPalette.svg]]
http://bioconductor.org/packages/stats/


=== Bioconductor repository directory structure ===
=== [http://rpubs.com/gaston/colortools colortools] ===
The information below is specific to Bioc 2.11 (R 2.15). There are linux and macosx subdirecotries whenever there are windows subdirectory.
Tools that allow users generate color schemes and palettes
<pre>
bioc/bin/windows/contrib/2.15
    /html
    /install
    /license
    /manuals
    /news
    /src
    /vignettes
data/annotation/bin/windows/contrib/2.15
              /html
              /licenses
              /manuals
              /src
              /vignettes
    /experiment/bin/windows/contrib/2.15
                /html
                /manuals
                /src/contrib
                /vignettes
extra/bin/windows/contrib
    /html
    /src
    /vignettes
</pre>


=== List all R packages from CRAN/Bioconductor ===
=== [https://github.com/daattali/colourpicker colourpicker] ===
<s>
A Colour Picker Tool for Shiny and for Selecting Colours in Plots
Check my daily result based on R 2.15 and Bioc 2.11 in [http://taichi.selfip.net:81/Rsummary/R_reposit.html]


# [http://taichi.selfip.net:81/Rsummary/cran.html CRAN]
=== eyedroppeR ===
# [http://taichi.selfip.net:81/Rsummary/bioc.html Bioc software]
[http://gradientdescending.com/select-colours-from-an-image-in-r-with-eyedropper/ Select colours from an image in R with {eyedroppeR}]
# [http://taichi.selfip.net:81/Rsummary/annotation.html Bioc annotation]
# [http://taichi.selfip.net:81/Rsummary/experiment.html Bioc experiment]
</s>


See [http://www.r-pkg.org/pkglist METACRAN] for packages hosted on CRAN. The 'https://github.com/metacran/PACKAGES' file contains the latest update.
== [https://github.com/kevinushey/rex rex] ==
Friendly Regular Expressions


== r-hub: the everything-builder the R community needs ==
== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
https://github.com/r-hub/proposal
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
=== Introducing R-hub, the R package builder service ===
* https://www.rstudio.com/resources/videos/r-hub-overview/
* http://blog.revolutionanalytics.com/2016/10/r-hub-public-beta.html


== Parallel Computing ==
See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.


# [http://shop.oreilly.com/product/0636920021421.do Example code] for the book Parallel R by McCallum and Weston.
# [http://www.win-vector.com/blog/2016/01/parallel-computing-in-r/ A gentle introduction to parallel computing in R]
# [http://www.stat.berkeley.edu/scf/paciorek-distribComp.pdf An introduction to distributed memory parallelism in R and C]
# [http://danielmarcelino.com/parallel-processing/Parallel Processing: When does it worth?]
=== Security warning from Windows/Mac ===
It seems it is safe to choose 'Cancel' when Windows Firewall tried to block R program when we use '''makeCluster()''' to create a socket cluster.
<pre>
<pre>
library(parallel)
library(formatR)
cl <- makeCluster(2)
tidy_source("Input.R", file = "output.R", width.cutoff=70)
clusterApply(cl, 1:2, get("+"), 3)
tidy_source("clipboard")  
stopCluster(cl)
# default width is getOption("width") which is 127 in my case.
</pre>
</pre>
[[File:WindowsSecurityAlert.png|100px]]  [[File:RegisterDoParallel mac.png|150px]]


If we like to see current firewall settings, just click Windows Start button, search 'Firewall' and choose 'Windows Firewall with Advanced Security'. In the 'Inbound Rules', we can see what programs (like, R for Windows GUI front-end, or Rserve) are among the rules. These rules are called 'private' in the 'Profile' column. Note that each of them may appear twice because one is 'TCP' protocol and the other one has a 'UDP' protocol.
Some issues
 
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
=== Detect number of cores ===
<pre>
<syntaxhighlight lang='rsplus'>
cat("abcd",
parallel::detectCores()
    # This is my comment
</syntaxhighlight>
    "defg")
Don't use the default option getOption("mc.cores", 2L) (PS it only returns 2.) in mclapply() unless you are a developer for a package.
</pre>
 
will result in
However, it is a different story when we run the R code in HPC cluster. Read the discussion [https://stackoverflow.com/questions/28954991/whether-to-use-the-detectcores-function-in-r-to-specify-the-number-of-cores-for Whether to use the detectCores function in R to specify the number of cores for parallel processing?]
<pre>
 
> tidy_source("clipboard")
On NIH's biowulf, even I specify an interactive session with 4 cores, the parallel::detectCores() function returns 56. This number is the same as the output from the bash command '''grep processor /proc/cpuinfo''' or (better) '''lscpu'''. The '''free -hm''' also returns a full 125GB size instead of my requested size (4GB by default).
Error in base::parse(text = code, srcfile = NULL) :
 
  3:1: unexpected string constant
=== parallel package (including parLapply, parSapply) ===
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
Parallel package was included in R 2.14.0. It is derived from the snow and multicore packages and provides many of the same functions as those packages.
3: "defg"
 
  ^
The parallel package provides several *apply functions for R users to quickly modify their code using parallel computing.
</pre>
 
* Comments appearing at the end of a line within a long complete statement ''won't break'' tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.  
* makeCluster(makePSOCKcluster, makeForkCluster), stopCluster. Other cluster types are passed to package '''snow'''.
<pre>
* '''clusterCall''', clusterEvalQ: source R files and/or load libraries
cat("abcd"
* clusterSplit
    ,"defg"  # This is my comment
* '''clusterApply''', '''clusterApplyLB''' (vs the foreach package)
  ,"ghij")
* '''clusterExport''': export variables
</pre>
* clusterMap
will become
* [https://stat.ethz.ch/R-manual/R-devel/library/parallel/html/mclapply.html parallel::mclapply()] and [https://stat.ethz.ch/R-manual/R-devel/library/parallel/html/clusterApply.html parallel::parLapply()]
<pre>
* parLapply, parSapply, parApply, parRapply, parCapply. Note that
cat("abcd", "defg"  # This is my comment
** '''parSapply()''' can be used to as a parallel version of the replicate() function. See [https://stackoverflow.com/questions/19281010/simplest-way-to-do-parallel-replicate?answertab=active#tab-top this example].
, "ghij")  
** An iteration parameter needs to be added to the first parameter of the main function.
</pre>
* parLapplyLB, parSapplyLB (load balance version)
Still bad!!
* clusterSetRNGStream, nextRNGStream, nextRNGSubStream
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
 
<pre>
Examples (See ?[http://www.inside-r.org/r-doc/parallel/clusterApply clusterApply])
cat("</p>",
<syntaxhighlight lang='rsplus'>
"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
library(parallel)
ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
cl <- makeCluster(2, type = "SOCK")
                    "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"),  
clusterApply(cl, 1:2, function(x) x*3)    # OR clusterApply(cl, 1:2, get("*"), 3)
file=ExternalFileName, sep="\n", append=T)
# [[1]]
</pre>
# [1] 3
will result in
#
<pre>
# [[2]]
> tidy_source("clipboard", width.cutoff=70)
# [1] 6
Error in base::parse(text = code, srcfile = NULL) :
parSapply(cl, 1:20, get("+"), 3)
  3:129: unexpected SPECIAL
#  [1]  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
stopCluster(cl)
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
</syntaxhighlight>
</pre>
An example of using clusterCall() or clusterEvalQ()
* ''width.cutoff'' parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
<syntaxhighlight lang='rsplus'>
<pre>
library(parallel)
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on",  
 
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName,
cl <- makeCluster(4)
    sep = "\n", append = T)
clusterCall(cl, function() {
</pre>
  source("test.R")
* It merges lines though I don't always want to do that. For example
})
<pre>
# clusterEvalQ(cl, {
cat("abcd"
#    source("test.R")
    ,"defg" 
# })
  ,"ghij")
 
</pre>
## do some parallel work
will become
stopCluster(cl)
<pre>
</syntaxhighlight>
cat("abcd", "defg", "ghij")  
 
</pre>
==== mclapply() from the 'parallel' package is a mult-core version of lapply() ====
* Be providing the number of cores in mclapply() using '''mc.cores''' argument (2 is used by default)
* Be careful on the need and the side-effect of using "L'Ecuyer-CMRG" seed.  
* '''[https://stackoverflow.com/questions/15070377/r-doesnt-reset-the-seed-when-lecuyer-cmrg-rng-is-used R doesn't reset the seed when “L'Ecuyer-CMRG” RNG is used?]''' <syntaxhighlight lang='rsplus'>
library(parallel)
system.time(mclapply(1:1e4L, function(x) rnorm(x)))
system.time(mclapply(1:1e4L, function(x) rnorm(x), mc.cores = 4))
 
set.seed(1234)
mclapply(1:3, function(x) rnorm(x))
set.seed(1234)
mclapply(1:3, function(x) rnorm(x)) # cannot reproduce the result
 
set.seed(123, "L'Ecuyer")
mclapply(1:3, function(x) rnorm(x))
mclapply(1:3, function(x) rnorm(x)) # results are not changed once we have run set.seed( , "L'Ecuyer")


set.seed(1234)                      # use set.seed() in order to get a new reproducible result
== styler ==
mclapply(1:3, function(x) rnorm(x))
https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.
mclapply(1:3, function(x) rnorm(x)) # results are not changed
</syntaxhighlight>
* An example of using [https://statcompute.wordpress.com/2019/02/23/gradient-free-optimization-for-glmnet-parameters/ parallel::mcMap()].


Note
== Download papers ==
# [https://stackoverflow.com/questions/15070377/r-doesnt-reset-the-seed-when-lecuyer-cmrg-rng-is-used R doesn't reset the seed when “L'Ecuyer-CMRG” RNG is used?]
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
# Windows OS can not use mclapply(). The mclapply() implementation relies on forking and Windows does not support forking. mclapply from the parallel package is implemented as a serial function on Windows systems. The ''parallelsugar'' package was created based on the above idea.
Search and Download Papers from the bioRxiv Preprint Server (biology)
# Another choice for Windows OS is to use parLapply() function in parallel package.
# [https://stackoverflow.com/questions/17196261/understanding-the-differences-between-mclapply-and-parlapply-in-r Understanding the differences between mclapply and parLapply in R] You don't have to worry about '''reproducing''' your environment on each of the cluster workers if mclapply() is used. <syntaxhighlight lang='rsplus'>
ncores <- as.integer( Sys.getenv('NUMBER_OF_PROCESSORS') )
cl <- makeCluster(getOption("cl.cores", ncores))
LLID2GOIDs2 <- parLapply(cl, rLLID, function(x) {
                                    library(org.Hs.eg.db); get("org.Hs.egGO")[[x]]}
                        )
stopCluster(cl)
</syntaxhighlight>It does work. Cut the computing time from 100 sec to 29 sec on 4 cores.


==== mclapply() vs foreach() ====
=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
https://stackoverflow.com/questions/44806048/r-mclapply-vs-foreach
Interface to the arXiv API


==== parallel vs doParallel package ====
=== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ===
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
* https://ropensci.org/technotes/2018/12/14/pdftools-20/


==== parallelsugar package ====
== [https://github.com/ColinFay/aside aside]: set it aside ==
* http://edustatistics.org/nathanvan/2015/10/14/parallelsugar-an-implementation-of-mclapply-for-windows/
An RStudio addin to run long R commands aside your current session.


If we load parallelsugar, the default implementation of parallel::mclapply, which used fork based clusters, will be overwritten by parallelsugar::mclapply, which is implemented with socket clusters.
== Teaching ==
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics


<syntaxhighlight lang='rsplus'>
== Organize R research project ==
library(parallel)  
* [https://cran.r-project.org/web/views/ReproducibleResearch.html CRAN Task View: Reproducible Research]
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
* [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow]. It suggests the [https://github.com/r-lib/here here] package. Don't use '''setwd()''' and '''rm(list = ls())'''.
** [https://rstats.wtf/safe-paths.html Practice safe paths]. Use projects and the [https://cran.r-project.org/web/packages/here/index.html here] package.
** In RStudio, if we try to send a few lines of code and one of the line contains '''setwd()''', it will give a message: ''The working directory was changed to XXX inside a notebook chunk. The working directory will be reset when the chunk is finished running. Use the knitr root.dir option in the setup chunk to change the working directory for notebook chunks.''
** [http://jenrichmond.rbind.io/post/how-to-use-the-here-package/ how to use the `here` package]
** No update for the ''here'' package after 2020-12. Consider [https://github.com/r-lib/usethis usethis] package (Automate project and package setup).
* drake project
** [https://ropensci.org/blog/2018/02/06/drake/ The prequel to the drake R package]
** [https://ropenscilabs.github.io/drake-manual/index.html The drake R Package User Manual]
* [https://docs.ropensci.org/targets/ targets] package
* [http://projecttemplate.net/ ProjectTemplate]


system.time( mclapply(1:4, function(xx){ Sys.sleep(10) }) )
=== How to save (and load) datasets in R (.RData vs .Rds file) ===
##    user  system elapsed
[https://rcrastinate.rbind.io/post/how-to-save-and-load-data-in-r-an-overview/ How to save (and load) datasets in R: An overview]
##    0.00    0.00  40.06


library(parallelsugar)
=== Naming convention ===
##
<ul>
## Attaching package: ‘parallelsugar’
<li>[https://stackoverflow.com/a/1946879 What is your preferred style for naming variables in R?]
##
* Use of period separator: they can get mixed up in simple method dispatch. However, it is used by base R ([https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/make.names make.names()], read.table(), et al)
## The following object is masked from ‘package:parallel’:
* Use of underscores: really annoying for ESS users
##  
* '''camelCase''': Winner
##    mclapply
<li>However, the [https://stackoverflow.com/a/13413278 survey] said (no surprises perhaps) that
* '''lowerCamelCase''' was most often used for function names,
* '''period.separated''' names most often used for parameters.
<li>[https://datamanagement.hms.harvard.edu/collect/file-naming-conventions What are file naming conventions?]
<li>[https://www.r-bloggers.com/2014/07/consistent-naming-conventions-in-r/ Consistent naming conventions in R]
<li>http://adv-r.had.co.nz/Style.html
<li>[https://www.r-bloggers.com/2011/07/testing-for-valid-variable-names/ Testing for valid variable names]
<li>R reserved words ?Reserved
* [https://www.datamentor.io/r-programming/reserved-words/ R Reserved Words]
* Among these words, if, else, repeat, while, function, for, '''in''', next and break are used for conditions, loops and user defined functions.
<li>Microarray/RNA-seq data
<pre>
clinicalDesignData  # clnDesignData
geneExpressionData  # gExpData
geneAnnotationData  # gAnnoData


system.time( mclapply(1:4, function(xx){ Sys.sleep(10) }) )
dataClinicalDesign
##    user  system elapsed
dataGeneExpression
##    0.04    0.08  12.98
dataAnnotation
</syntaxhighlight>
</pre>
<pre>
# Search all variables ending with .Data
ls()[grep("\\.Data$", ls())]
# Search all variables starting with data_
ls()[grep("^data_", ls())]
</pre>
</ul>


=== [http://cran.r-project.org/web/packages/snow/index.html snow] package ===
=== Efficient Data Management in R ===
[https://www.mzes.uni-mannheim.de/socialsciencedatalab/article/efficient-data-r/ Efficient Data Management in R]. .Rprofile, renv package and dplyr package.


Supported cluster types are "SOCK", "PVM", "MPI", and "NWS".
== Text to speech ==
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]


=== [http://cran.r-project.org/web/packages/multicore/index.html multicore] package ===
== Speech to text ==
This package is removed from CRAN.
https://github.com/ggerganov/whisper.cpp and an R package [https://github.com/bnosac/audio.whisper audio.whisper]


Consider using package ‘parallel’ instead.
== Weather data ==
* [https://github.com/ropensci/prism prism] package
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]


=== [http://cran.r-project.org/web/packages/foreach/index.html foreach] package ===
== logR ==
This package depends on one of the following
https://github.com/jangorecki/logR
* doParallel - Foreach parallel adaptor for the parallel package
* doSNOW - Foreach parallel adaptor for the snow package
* doMC - Foreach parallel adaptor for the multicore package. Used in [https://web.stanford.edu/~hastie/glmnet/glmnet_alpha.html glmnet] vignette.
* doMPI - Foreach parallel adaptor for the Rmpi package
* doRedis - Foreach parallel adapter for the rredis package
as a backend.


<syntaxhighlight lang='rsplus'>
== Progress bar ==
library(foreach)
https://github.com/r-lib/progress#readme
library(doParallel)


m <- matrix(rnorm(9), 3, 3)
Configurable Progress bars, they may include percentage, elapsed time, and/or the estimated completion time. They work in terminals, in 'Emacs' 'ESS', 'RStudio', 'Windows' 'Rgui' and the 'macOS'.


cl <- makeCluster(2, type = "SOCK")
== cron ==
registerDoParallel(cl) # register the parallel backend with the foreach package
* [https://github.com/bnosac/cronr cronR]
foreach(i=1:nrow(m), .combine=rbind) %dopar%
* [https://mathewanalytics.com/building-a-simple-pipeline-in-r/ Building a Simple Pipeline in R]
  (m[i,] / mean(m[i,]))


stopCluster(cl)
== beepr: Play A Short Sound ==
</syntaxhighlight>
https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".


See also this post [http://blog.revolutionanalytics.com/2015/10/updates-to-the-foreach-package-and-its-friends.html Updates to the foreach package and its friends] on Oct 2015.
== utils package ==
https://www.rdocumentation.org/packages/utils/versions/3.6.2


* [https://statcompute.wordpress.com/2015/12/13/calculate-leave-one-out-prediction-for-glm/ Cross validation in prediction for glm]
== tools package ==
* [http://gforge.se/2015/02/how-to-go-parallel-in-r-basics-tips/#The_foreach_package How-to go parallel in R – basics + tips]
* https://www.rdocumentation.org/packages/tools/versions/3.6.2
* [https://www.r-bloggers.com/2023/08/three-four-r-functions-i-enjoyed-this-week/ Where in the file are there non ASCII characters?], [https://rdocumentation.org/packages/tools/versions/3.6.2/topics/showNonASCII tools::showNonASCIIfile(<filename>)]


==== combine list of lists ====
= Different ways of using R =
* .combine argument https://stackoverflow.com/questions/27279164/output-list-of-two-rbinded-data-frames-with-foreach-in-r
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)
* [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Merge lists] by mapply() or base::Map()


<syntaxhighlight lang='rsplus'>
== 10 things R can do that might surprise you ==
comb <- function(...) {
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/
  mapply('cbind', ..., SIMPLIFY=FALSE)
}


library(foreach)
== R call C/C++ ==
library(doParallel)
Mainly talks about .C() and .Call().


cl <- makeCluster(2)
Note that scalars and arrays must be passed using pointers. So if we want to access a function not exported from a package, we may need to modify the function to make the arguments as pointers.
registerDoParallel(cl) # register the parallel backend with the foreach package


m <- rbind(rep(1,3), rep(2,3))
* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
* http://www.stat.harvard.edu/ccr2005/
* http://mazamascience.com/WorkingWithData/?p=1099
* [https://youtube.com/playlist?list=PLwc48KSH3D1OkObQ22NHbFwEzof2CguJJ Make an R package with C++ code] (a playlist from youtube)
* [https://working-with-data.mazamascience.com/2021/07/16/using-r-calling-c-code-hello-world/ Using R – Calling C code ‘Hello World!’]
* [http://www.haowulab.org//pages/computing.html Computing tip] by Hao Wu


# nrow(m) can represents number of permutations (2 in this toy example)
=== .Call ===
tmp <- foreach(i=1:nrow(m)) %dopar% {
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/CallExternal ?.Call]
  a <- m[i, ]
* [http://mazamascience.com/WorkingWithData/?p=1099 Using R — .Call(“hello”)]
  b <- a * 10
* http://adv-r.had.co.nz/C-interface.html
  list(a, b)
* [https://working-with-data.mazamascience.com/2021/07/16/using-r-callhello/ Using R – .Call(“hello”)]
}; tmp
# [[1]]
# [[1]][[1]]
# [1] 1 1 1
#
# [[1]][[2]]
# [1] 10 10 10
#
#
# [[2]]
# [[2]][[1]]
# [1] 2 2 2
#
# [[2]][[2]]
# [1] 20 20 20


foreach(i=1:nrow(m), .combine = "comb") %dopar% {
Be sure to add the ''PACKAGE'' parameter to avoid an error like
  a <- m[i,]
<pre>
  b <- a * 10
cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group,  
  list(a, b)
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
}
Error in .Call("standardize", X) :
# [[1]]
   "standardize" not resolved from current namespace (grpreg)
#      [,1] [,2]
</pre>
# [1,]    1   2
# [2,]    1    2
# [3,]    1    2
#
# [[2]]
#      [,1] [,2]
# [1,]   10  20
# [2,]  10  20
# [3,]  10  20
stopCluster(cl)
</syntaxhighlight>


==== Replacing double loops ====
=== NAMESPACE file & useDynLib ===
* https://stackoverflow.com/questions/30927693/how-can-i-parallelize-a-double-for-loop-in-r
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
* http://www.exegetic.biz/blog/2013/08/the-wonders-of-foreach/
* We don't need to include double quotes around the C/Fortran subroutines in .C() or .Fortran()
<syntaxhighlight lang='rsplus'>
* digest package example: [https://github.com/cran/digest/blob/master/NAMESPACE NAMESPACE] and [https://github.com/cran/digest/blob/master/R/digest.R R functions] using .Call().
library(foreach)
* stats example: [https://github.com/wch/r-source/blob/trunk/src/library/stats/NAMESPACE NAMESPACE]
library(doParallel)


nc <- 4
(From [https://cran.r-project.org/doc/manuals/r-release/R-exts.html#dyn_002eload-and-dyn_002eunload Writing R Extensions manual]) Loading is most often done automatically based on the '''useDynLib()''' declaration in the '''NAMESPACE''' file, but may be done explicitly via a call to '''library.dynam()'''. This has the form
nr <- 2
{{Pre}}
library.dynam("libname", package, lib.loc)
</pre>


cores=detectCores()
=== library.dynam.unload() ===
cl <- makeCluster(cores[1]-1)
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
registerDoParallel(cl)
* http://r-pkgs.had.co.nz/src.html. The '''library.dynam.unload()''' function should be placed in '''.onUnload()''' function. This function can be saved in any R files.
# set.seed(1234) # not work
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]
# set.seed(1234, "L'Ecuyer-CMRG") # not work either
# library("doRNG")
# registerDoRNG(seed = 1985)    # not work with nested foreach
# Error in list(e1 = list(args = (1:nr)(), argnames = "i", evalenv = <environment>,  :
#  nested/conditional foreach loops are not supported yet.
m <- foreach (i = 1:nr, .combine='rbind') %:% # nesting operator
  foreach (j = 1:nc) %dopar% {
    rnorm(1, i*5, j) # code to parallelise
}
m
stopCluster(cl)
</syntaxhighlight>
Note that since the random seed (see the next session) does not work on nested loop, it is better to convert nested loop (two indices) to a single loop (one index).


==== set seed and [https://cran.r-project.org/web/packages/doRNG/ doRNG] package ====
=== gcc ===
* [https://cran.r-project.org/web/packages/doRNG/vignettes/doRNG.pdf Vignette], [https://www.rdocumentation.org/packages/doRNG/versions/1.7.1/topics/doRNG-package Documentation]
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
* [http://michaeljkoontz.weebly.com/uploads/1/9/9/4/19940979/parallel.pdf#page=4 doRNG] package example
* [https://stackoverflow.com/questions/8358098/how-to-set-seed-for-random-simulations-with-foreach-and-domc-packages How to set seed for random simulations with foreach and doMC packages?]
* Use '''clusterSetRNGStream()''' from the parallel package; see [http://gforge.se/2015/02/how-to-go-parallel-in-r-basics-tips/ How-to go parallel in R – basics + tips]
* http://www.stat.colostate.edu/~scharfh/CSP_parallel/handouts/foreach_handout.html#random-numbers <syntaxhighlight lang='rsplus'>
library("doRNG") # doRNG does not need to be loaded after doParallel
library("doParallel")


cl <- makeCluster(2)
=== Primitive functions ===
registerDoParallel(cl)
[https://nathaneastwood.github.io/2020/02/01/primitive-functions-list/ Primitive Functions List]


registerDoRNG(seed = 1234) # works for a single loop
== SEXP ==
m1 <- foreach(i = 1:5, .combine = 'c') %dopar% rnorm(1)
Some examples from packages
registerDoRNG(seed = 1234)
m2 <- foreach(i = 1:5, .combine = 'c') %dopar% rnorm(1)
identical(m1, m2)
stopCluster(cl)


attr(m1, "rng") <- NULL # remove rng attribute
* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
</syntaxhighlight>
* Another way to use the seed is to supply '''[https://www.rdocumentation.org/packages/doRNG/versions/1.7.1/topics/%25dorng%25 .options.RNG]''' in foreach() function. <syntaxhighlight lang='rsplus'>
r1 <- foreach(i=1:4, .options.RNG=1234) %dorng% { runif(1) }
</syntaxhighlight>


==== Export libraries and variables ====
== R call Fortran ==
* http://stat.ethz.ch/R-manual/R-devel/library/parallel/html/clusterApply.html
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
<syntaxhighlight lang='rsplus'>
* [https://www.r-bloggers.com/the-need-for-speed-part-1-building-an-r-package-with-fortran-or-c/ The Need for Speed Part 1: Building an R Package with Fortran (or C)] (Very detailed)
clusterEvalQ(cl, {
  library(biospear)
  library(glmnet)
  library(survival)
})
clusterExport(cl, list("var1", "foo2"))
</syntaxhighlight>


==== Summary the result ====
== Embedding R ==
foreach returns the result in a list. For example, if each component is a matrix we can use


* Reduce("+", res)/length(res) # Reduce("+", res, na.rm = TRUE) not working
* See [http://cran.r-project.org/doc/manuals/R-exts.html#Linking-GUIs-and-other-front_002dends-to-R Writing for R Extensions] Manual Chapter 8.
* apply(simplify2array(res), 1:2, mean, na.rm = TRUE)
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
* [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf Technical report]  by Friedrich Leisch in 2007.
* https://stat.ethz.ch/pipermail/r-help/attachments/20110729/b7d86ed7/attachment.pl


to get the average of matrices over the list.
=== An very simple example (do not return from shell) from Writing R Extensions manual ===
The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.


==== Read a list files ====
This example can be run by
[https://danielmarcelino.github.io/2018/reading-list-faster.html Reading List Faster With parallel, doParallel, and pbapply]
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>


=== snowfall package ===
Note:
http://www.imbi.uni-freiburg.de/parallel/docs/Reisensburg2009_TutParallelComputing_Knaus_Porzelius.pdf
# '''R_HOME/bin/exec/R''' is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
# '''R_HOME/bin/R''' is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to ''/usr/local/bin/R''. When we run ''R_HOME/bin/R'', it actually runs ''R_HOME/bin/R CMD R_HOME/bin/exec/R'' (see line 259 of ''R_HOME/bin/R'' as in R 3.0.2) so we know the important role of ''R_HOME/bin/exec/R''.


=== [http://cran.r-project.org/web/packages/Rmpi/index.html Rmpi] package ===
More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
Some examples/tutorials


* http://trac.nchc.org.tw/grid/wiki/R-MPI_Install
=== An example from Bioconductor workshop ===
* http://www.arc.vt.edu/resources/software/r/index.php
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* https://www.sharcnet.ca/help/index.php/Using_R_and_MPI
* Use eval() function. See R-Ext [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Unix_002dalikes 8.1] and [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Windows 8.2] and [http://cran.r-project.org/doc/manuals/R-exts.html#Evaluating-R-expressions-from-C 5.11].
* http://math.acadiau.ca/ACMMaC/Rmpi/examples.html
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* http://www.umbc.edu/hpcf/resources-tara/how-to-run-R.html
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c
* [http://www.slideshare.net/bytemining/taking-r-to-the-limit-high-performance-computing-in-r-part-1-parallelization-la-r-users-group-727 Ryan Rosario]
* http://pj.freefaculty.org/guides/Rcourse/parallel-1/parallel-1.pdf
* * http://biowulf.nih.gov/apps/R.html


=== OpenMP ===
Example:
* [http://www.parallelr.com/r-and-openmp-boosting-compiled-code-on-multi-core-cpu-s/ R and openMP: boosting compiled code on multi-core cpu-s] from parallelr.com.
Create [https://gist.github.com/arraytools/7d32d92fee88ffc029365d178bc09e75#file-embed-c embed.c] file.
Then build the executable. Note that I don't need to create R_HOME variable.
<pre>
cd
tar xzvf
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest


=== [http://www.bioconductor.org/packages/release/bioc/html/BiocParallel.html BiocParallel] ===
nano embed.c
* [http://rpubs.com/seandavi/KallistoFromR Orchestrating a small, parallel, RNA-seq pre-processing workflow using R]
# Using a single line will give an error and cannot not show the real problem.
 
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
=== [https://cran.r-project.org/web/packages/RcppParallel/index.html RcppParallel] ===
# A better way is to run compile and link separately
 
gcc -I../../include -c embed.c
=== future & [https://cran.r-project.org/web/packages/future.apply/index.html future.apply] & doFuture packages ===
gcc -o embed embed.o -L../../lib -lR -lRblas
* [https://alexioannides.com/2016/11/02/asynchronous-and-distributed-programming-in-r-with-the-future-package/ Asynchronous and Distributed Programming in R with the Future Package]
../../bin/R CMD ./embed
* [https://www.jottr.org/2018/06/23/future.apply_1.0.0/ Parallelize Any Base R Apply Function]
</pre>
* [https://www.jottr.org/2019/01/11/parallelize-a-for-loop-by-rewriting-it-as-an-lapply-call/ Parallelize a For-Loop by Rewriting it as an Lapply Call]
* [https://blog.revolutionanalytics.com/2019/01/future-package.html Use foreach with HPC schedulers thanks to the future package]
 
=== Apache Spark ===
* [http://files.meetup.com/3576292/Dubravko%20Dulic%20SparkR%20June%202016.pdf Introduction to Apache Spark]


=== Microsoft R Server ===
Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying '''R_HOME''' variable and including the directories used in linking R in '''LD_LIBRARY_PATH'''. This is based on the inform provided by [http://cran.r-project.org/doc/manuals/r-devel/R-exts.html Writing R Extensions].
* [http://files.meetup.com/3576292/Stefan%20Cronjaeger%20R%20Server.pdf Microsoft R '''Server'''] (not Microsoft R Open)
<pre>
export R_HOME=/home/brb/Downloads/R-3.0.2
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
./embed # No need to include R CMD in front.
</pre>


=== GPU ===
Question: Create a data frame in C? Answer: [https://stat.ethz.ch/pipermail/r-devel/2013-August/067107.html Use data.frame() via an eval() call from C]. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as [https://stat.ethz.ch/pipermail/r-devel/2013-August/067109.html here].
* [http://www.parallelr.com/r-gpu-programming-for-all-with-gpur/ GPU Programming for All with ‘gpuR] from parallelr.com. The gpuR is available on [https://cran.r-project.org/web/packages/gpuR/index.html CRAN].
* [https://cran.r-project.org/web/packages/gputools/index.html gputools]


=== Threads ===
Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
* [https://cran.r-project.org/web/packages/Rdsm/index.html Rdsm] package
* [https://random-remarks.net/2016/12/11/a-very-experimental-threading-in-r/ (A Very) Experimental Threading in R] and a post from [https://matloff.wordpress.com/2016/12/11/threading-in-r/ Mad Scientist]


=== Benchmark ===
=== Create a Simple Socket Server in R ===
[http://rpsychologist.com/benchmark-parallel-sim Are parallel simulations in the cloud worth it? Benchmarking my MBP vs my Workstation vs Amazon EC2]
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].


R functions to run timing
Create an R function
<syntaxhighlight lang='rsplus'>
<pre>
# Method 1
simpleServer <- function(port=6543)
system.time( invisible(rnorm(10000)))
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}
</pre>
Then run simpleServer(). Open another terminal and try to communicate with the server
<pre>
$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.


# Method 2
Welcome to R!
btime <- Sys.time()
R> summary(iris[, 3:5])
invisible(rnorm(10000))
  Petal.Length    Petal.Width          Species 
Sys.time() - btime
Min.   :1.000  Min.  :0.100  setosa    :50 
</syntaxhighlight>
1st Qu.:1.600  1st Qu.:0.300  versicolor:50 
 
Median :4.350  Median :1.300  virginica :50 
== Cloud Computing ==
Mean  :3.758  Mean  :1.199                 
 
3rd Qu.:5.100  3rd Qu.:1.800                 
=== Install R on Amazon EC2 ===
Max.  :6.900  Max.  :2.500                 
http://randyzwitch.com/r-amazon-ec2/


=== Bioconductor on Amazon EC2 ===
R> quit
http://www.bioconductor.org/help/bioconductor-cloud-ami/
Connection closed by foreign host.
</pre>


== Big Data Analysis ==
=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
* [https://cran.r-project.org/web/views/HighPerformanceComputing.html CRAN Task View: High-Performance and Parallel Computing with R]
Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See [[R#An_example_from_Bioconductor_workshop|Example from Bioconductor workshop]].
* [http://www.xmind.net/m/LKF2/ R for big data] in one picture
* [https://rstudio-pubs-static.s3.amazonaws.com/72295_692737b667614d369bd87cb0f51c9a4b.html Handling large data sets in R]


=== bigmemory, biganalytics, bigtabulate ===
See my [[Rserve]] page.


=== ff, ffbase ===
=== outsider ===
* tapply does not work. [https://stackoverflow.com/questions/16470677/using-tapply-ave-functions-for-ff-vectors-in-r Using tapply, ave functions for ff vectors in R]
* [https://joss.theoj.org/papers/10.21105/joss.02038 outsider]: Install and run programs, outside of R, inside of R
* [http://www.bnosac.be/index.php/blog/12-popularity-bigdata-large-data-packages-in-r-and-ffbase-user-presentation Popularity bigdata / large data packages in R and ffbase useR presentation]
* [https://github.com/stephenturner/om..bcftools Run bcftools with outsider in R]


=== biglm ===
=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
 
== Useful R packages ==
* [https://support.rstudio.com/hc/en-us/articles/201057987-Quick-list-of-useful-R-packages Quick list of useful R packages]
* [https://github.com/qinwf/awesome-R awesome-R]
* [https://stevenmortimer.com/one-r-package-a-day/ One R package a day]


=== RInside ===
=== [http://rdotnet.codeplex.com/ R.NET] ===
* http://dirk.eddelbuettel.com/code/rinside.html
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf


==== Ubuntu ====
=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on [http://www.youtube.com/watch?v=UQ8yKQcPTg0 Youtube]. I have tested this '''qtdensity''' example successfully using Qt 4.8.5.
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
# Follow the instruction [[#cairoDevice|cairoDevice]] to install required libraries for cairoDevice package and then cairoDevice itself.
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.
# Install [[Qt|Qt]]. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
# Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.
# Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the ''interactive'' program qtdensity appears on your desktop.
[[File:qtdensity.png|100px]].


With RInside + [http://www.webtoolkit.eu/wt Wt web toolkit] installed, we can also create a web application. To demonstrate the example in ''examples/wt'' directory, we can do
Terminal
<pre>
{{Pre}}
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
# jdk 7
make
sudo apt-get install openjdk-7-*
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk
</pre>
and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.
* Create the file '''/etc/ld.so.conf.d/java.conf''' with the following entries:
<pre>
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
</pre>
</pre>
Then we can go to the browser's address bar and type ''http://localhost:8080'' to see how it works (a screenshot is in [http://dirk.eddelbuettel.com/blog/2011/11/30/ here]).
* And then run '''sudo ldconfig'''


==== Windows 7 ====
Now go back to R
To make RInside works on Windows OS, try the following
{{Pre}}
# Make sure R is installed under '''C:\''' instead of '''C:\Program Files''' if we don't want to get an error like ''g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory''.
install.packages("rJava")
# Install RTools
# Instal RInside package from source (the binary version will give an [http://stackoverflow.com/questions/13137770/fatal-error-unable-to-open-the-base-package error ])
# Create a DOS batch file containing necessary paths in PATH environment variable
<pre>
@echo off
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set R_HOME=C:\R\R-3.0.1
echo Setting environment for using R
cmd
</pre>
</pre>
In the Windows command prompt, run
Done!
 
If above does not work, a simple way is by (under Ubuntu) running
<pre>
<pre>
cd C:\R\R-3.0.1\library\RInside\examples\standard
sudo apt-get install r-cran-rjava
make -f Makefile.win
</pre>
</pre>
Now we can test by running any of executable files that '''make''' generates. For example, ''rinside_sample0''.
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.
 
=== RCaller ===
 
=== RApache ===
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf
 
=== Rscript, arguments and commandArgs() ===
[https://www.r-bloggers.com/passing-arguments-to-an-r-script-from-command-lines/ Passing arguments to an R script from command lines]
Syntax:
<pre>
<pre>
rinside_sample0
$ Rscript --help
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]
</pre>
</pre>


As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See  some discussions in
Example:
* http://stackoverflow.com/questions/12280707/using-rinside-with-qt-in-windows
<pre>
* http://www.mail-archive.com/[email protected]/msg04377.html
args = commandArgs(trailingOnly=TRUE)
So the Qt and Wt web tool applications on Windows may or may not be possible.
# test if there is at least one argument: if not, return an error
 
if (length(args)==0) {
=== GUI ===
  stop("At least one argument must be supplied (input file).n", call.=FALSE)
==== Qt and R ====
} else if (length(args)==1) {
* http://cran.r-project.org/web/packages/qtbase/index.html [https://stat.ethz.ch/pipermail/r-devel/2015-July/071495.html QtDesigner is such a tool, and its output is compatible with the qtbase R package]
  # default output file
* http://qtinterfaces.r-forge.r-project.org
  args[2] = "out.txt"
 
}
=== tkrplot ===
cat("args[1] = ", args[1], "\n")
On Ubuntu, we need to install tk packages, such as by
cat("args[2] = ", args[2], "\n")
</pre>
<pre>
<pre>
sudo apt-get install tk-dev
Rscript --vanilla sillyScript.R iris.txt out.txt
# args[1] =  iris.txt
# args[2] =  out.txt
</pre>
</pre>


=== reticulate - Interface to 'Python' ===
=== Rscript, #! Shebang and optparse package ===
* https://cran.r-project.org/web/packages/reticulate/index.html
<ul>
* [https://www.brodrigues.co/blog/2018-12-30-reticulate/ R or Python? Why not both? Using Anaconda Python within R with {reticulate}]
<li>Writing [https://www.r-bloggers.com/2014/05/r-scripts/ R scripts] like linux bash files.
 
<li>[https://www.makeuseof.com/shebang-in-linux/ What Is the Shebang (#!) Character Sequence in Linux?]
=== Hadoop (eg ~100 terabytes) ===
<li>[https://blog.rmhogervorst.nl/blog/2020/04/14/where-does-the-output-of-rscript-go/ Where does the output of Rscript go?]
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
<li>Create a file <shebang.R>.  
 
* RHadoop
* Hive
* [http://cran.r-project.org/web/packages/mapReduce/ MapReduce]. Introduction by [http://www.linuxjournal.com/content/introduction-mapreduce-hadoop-linux Linux Journal].
* http://www.techspritz.com/category/tutorials/hadoopmapredcue/ Single node or multinode cluster setup using Ubuntu with VirtualBox (Excellent)
* [http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-single-node-cluster/ Running Hadoop on Ubuntu Linux (Single-Node Cluster)]
* Ubuntu 12.04 http://www.youtube.com/watch?v=WN2tJk_oL6E and [https://www.dropbox.com/s/05aurcp42asuktp/Chiu%20Hadoop%20Pig%20Install%20Instructions.docx instruction]
* Linux Mint http://blog.hackedexistence.com/installing-hadoop-single-node-on-linux-mint
* http://www.r-bloggers.com/search/hadoop
 
==== [https://github.com/RevolutionAnalytics/RHadoop/wiki RHadoop] ====
* [http://www.rdatamining.com/tutorials/r-hadoop-setup-guide RDataMining.com] based on Mac.
* Ubuntu 12.04 - [http://crishantha.com/wp/?p=1414 Crishantha.com], [http://nikhilshah123sh.blogspot.com/2014/03/setting-up-rhadoop-in-ubuntu-1204.html nikhilshah123sh.blogspot.com].[http://bighadoop.wordpress.com/2013/02/25/r-and-hadoop-data-analysis-rhadoop/ Bighadoop.wordpress] contains an example.
* RapReduce in R by [https://github.com/RevolutionAnalytics/rmr2/blob/master/docs/tutorial.md RevolutionAnalytics] with a few examples.
* https://twitter.com/hashtag/rhadoop
* [http://bigd8ta.com/step-by-step-guide-to-setting-up-an-r-hadoop-system/ Bigd8ta.com] based on Ubuntu 14.04.
 
==== Snowdoop: an alternative to MapReduce algorithm ====
* http://matloff.wordpress.com/2014/11/26/how-about-a-snowdoop-package/
* http://matloff.wordpress.com/2014/12/26/snowdooppartools-update/comment-page-1/#comment-665
 
=== [http://cran.r-project.org/web/packages/XML/index.html XML] ===
On Ubuntu, we need to install libxml2-dev before we can install XML package.
<pre>
<pre>
sudo apt-get update
#!/usr/bin/env Rscript
sudo apt-get install libxml2-dev
print ("shebang works")
</pre>
</pre>
 
Then in the command line
On CentOS,
<pre>
<pre>
yum -y install libxml2 libxml2-devel
chmod u+x shebang.R
./shebang.R
</pre>
</pre>
<li>[http://www.cureffi.org/2014/01/15/running-r-batch-mode-linux/ Running R in batch mode on Linux]
<li>[https://cran.r-project.org/web/packages/optparse/index.html optparse] package. Check out its vignette.
<li>[https://cran.r-project.org/web/packages/getopt/index.html getopt]: C-Like 'getopt' Behavior.
</ul>


==== XML ====
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
* http://giventhedata.blogspot.com/2012/06/r-and-web-for-beginners-part-ii-xml-in.html. It gave an example of extracting the XML-values from each XML-tag for all nodes and save them in a data frame using '''xmlSApply()'''.
Provides hash-bang (#!) capability for R
* http://www.quantumforest.com/2011/10/reading-html-pages-in-r-for-text-processing/
 
* https://tonybreyal.wordpress.com/2011/11/18/htmltotext-extracting-text-from-html-via-xpath/
FAQs:
* https://www.tutorialspoint.com/r/r_xml_files.htm
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
* https://www.datacamp.com/community/tutorials/r-data-import-tutorial#xml
* [https://stackoverflow.com/questions/3412911/r-exe-rcmd-exe-rscript-exe-and-rterm-exe-whats-the-difference Whats the difference between Rscript and R CMD BATCH]
* [http://www.stat.berkeley.edu/~statcur/Workshop2/Presentations/XML.pdf Extracting data from XML] PubMed and Zillow are used to illustrate. xmlTreeParse(),  xmlRoot(),  xmlName() and xmlSApply().
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
* https://yihui.name/en/2010/10/grabbing-tables-in-webpages-using-the-xml-package/
{{Pre}}
<syntaxhighlight lang='rsplus'>
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
library(XML)
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"


# Read and parse HTML file
-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)
                                              # Example: R -q -e "date()"


# Extract all the paragraphs (HTML tag is p, starting at
-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
# the root of the document). Unlist flattens the list to
                                              # It won't show the startup message when it is used in the command line.
# create a character vector.
                                              # Example: Rscript -e "date()"
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))
</pre>


# Replace all by spaces
We can install littler using two ways.
doc.text = gsub('\n', ' ', doc.text)
* install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ''~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r''). You need to create a soft link in order to access it globally.
* sudo apt install littler. This will install 'r' globally; however, the installed version may be old.


# Join all the elements of the character vector into a single
After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The [http://dirk.eddelbuettel.com/code/littler.examples.html web version of examples] does not have the TOC.
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')
</syntaxhighlight>


This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.
'''r''' was not meant to run interactively like '''R'''. See ''man r''.
<syntaxhighlight lang='rsplus'>
> library(RCurl) # getURL()
> library(XML)  # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # <span class="css-truncate-target">2.5.3a</span>
> plain.text
[1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"   
[6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"


> # try picard
=== RInside: Embed R in C++ ===
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
See [[R#RInside|RInside]]
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
[10] "2.6.0"
</syntaxhighlight>
This method can be used to monitor new tags/releases from some projects like [https://github.com/Ultimaker/Cura/releases Cura], BWA, Picard, [https://github.com/alexdobin/STAR/releases STAR]. But for some projects like [https://github.com/ncbi/sra-tools sratools] the '''class''' attribute in the '''span''' element ("css-truncate-target") can be different (such as "tag-name").


==== xmlview ====
(''From RInside documentation'') The RInside package makes it easier to embed R in your C++ applications. There is no code you would execute directly from the R environment. Rather, you write C++ programs that embed R which is illustrated by some the included examples.
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/


=== RCurl ===
The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
<syntaxhighlight lang='bash'>
# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>


==== Scrape google scholar results ====
To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R


No google ID is required
To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>


Seems not work
The real build process looks like (check <Makefile> for completeness)
<pre>
<pre>
  Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :
g++ -I/home/brb/Downloads/R-3.0.2/include \
  arguments imply differing number of rows: 2, 0  
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include  \
    rinside_sample0.cpp \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0
</pre>
</pre>


==== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ====
Hello World example of embedding R in C++.
'''devtools''' package depends on Curl.  
<pre>
<syntaxhighlight lang='bash'>
#include <RInside.h>                    // for the embedded R via RInside
# Test on Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev
</syntaxhighlight>


[https://github.com/wch/movies/issues/3 Lazy-load database XXX is corrupt. internal error -3]. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.
int main(int argc, char *argv[]) {


==== [https://github.com/hadley/httr httr] ====
    RInside R(argc, argv);              // create an embedded R instance
httr imports curl, jsonlite, mime, openssl and R6 packages.


When I tried to install httr package, I got an error and some message:
    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
<pre>
Configuration failed because openssl was not found. Try installing:
* deb: libssl-dev (Debian, Ubuntu, etc)
* rpm: openssl-devel (Fedora, CentOS, RHEL)
* csw: libssl_dev (Solaris)
* brew: openssl (Mac OSX)
If openssl is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
--------------------------------------------------------------------
ERROR: configuration failed for package ‘openssl’
</pre>
It turns out after I run '''sudo apt-get install libssl-dev''' in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!


Real example: see [http://stackoverflow.com/questions/27371372/httr-retrieving-data-with-post this post]. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).
    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns


Since httr package was used in many other packages, take a look at how others use it. For example, [https://github.com/ropensci/aRxiv aRxiv] package.
    exit(0);
}
</pre>


==== [http://cran.r-project.org/web/packages/curl/ curl] ====
The above can be compared to the Hello world example in Qt.
curl is independent of RCurl package.
<pre>
#include <QApplication.h>
#include <QPushButton.h>


* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
int main( int argc, char **argv )
* https://www.opencpu.org/posts/curl-release-0-8/
{
    QApplication app( argc, argv );


<syntaxhighlight lang='rsplus'>
    QPushButton hello( "Hello world!", 0 );
library(curl)
    hello.resize( 100, 30 );
h <- new_handle()
handle_setform(h,
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)
</syntaxhighlight>


==== [http://ropensci.org/packages/index.html rOpenSci] packages ====
    app.setMainWidget( &hello );
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
    hello.show();


=== [https://cran.r-project.org/web/packages/remotes/index.html remotes] ===
    return app.exec();
Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'.
}
 
Example:
<syntaxhighlight lang='rsplus'>
# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')
</syntaxhighlight>
 
=== DirichletMultinomial ===
On Ubuntu, we do
<pre>
sudo apt-get install libgsl0-dev
</pre>
</pre>


=== Create GUI ===
=== [http://www.rfortran.org/ RFortran] ===
==== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ====
RFortran is an open source project with the following aim:


=== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ===
''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
[http://statistical-research.com/simulating-random-multivariate-correlated-data-categorical-variables/?utm_source=rss&utm_medium=rss&utm_campaign=simulating-random-multivariate-correlated-data-categorical-variables here]


=== [http://cran.r-project.org/web/packages/rjson/index.html rjson] ===
It works only on Windows platform with Microsoft Visual Studio installed:(
http://heuristically.wordpress.com/2013/05/20/geolocate-ip-addresses-in-r/


=== [http://cran.r-project.org/web/packages/RJSONIO/index.html RJSONIO] ===
== Call R from other languages ==
==== Accessing Bitcoin Data with R ====
=== C ===
http://blog.revolutionanalytics.com/2015/11/accessing-bitcoin-data-with-r.html
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]


==== Plot IP on google map ====
Error: [https://stackoverflow.com/questions/43662542/not-resolved-from-current-namespace-error-when-calling-c-routines-from-r “not resolved from current namespace” error, when calling C routines from R]
* http://thebiobucket.blogspot.com/2011/12/some-fun-with-googlevis-plotting-blog.html#more  (RCurl, RJONIO, plyr, googleVis)
* http://devblog.icans-gmbh.com/using-the-maxmind-geoip-api-with-r/ (RCurl, RJONIO, maps)
* http://cran.r-project.org/web/packages/geoPlot/index.html (geoPlot package (deprecated as 8/12/2013))
* http://archive09.linux.com/feature/135384  (Not R) ApacheMap
* http://batchgeo.com/features/geolocation-ip-lookup/    (Not R)  (Enter a spreadsheet of adress, city, zip or a column of IPs and it will show the location on google map)
* http://code.google.com/p/apachegeomap/


The following example is modified from the first of above list.
Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
<pre>
require(RJSONIO) # fromJSON
require(RCurl)  # getURL


temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
=== JRI ===
                          ssl.verifypeer = FALSE)
http://www.rforge.net/JRI/
ip <- read.table(textConnection(temp), as.is=TRUE)
names(ip) <- "IP"
nr = nrow(ip)
Lon <- as.numeric(rep(NA, nr))
Lat <- Lon
Coords <- data.frame(Lon, Lat)
ip2coordinates <- function(ip) {
  api <- "http://freegeoip.net/json/"
  get.ips <- getURL(paste(api, URLencode(ip), sep=""))
  # result <- ldply(fromJSON(get.ips), data.frame)
  result <- data.frame(fromJSON(get.ips))
  names(result)[1] <- "ip.address"
  return(result)
}


for (i in 1:nr){
=== ryp2 ===
  cat(i, "\n")
http://rpy.sourceforge.net/rpy2.html
  try(
  Coords[i, 1:2] <- ip2coordinates(ip$IP[i])[c("longitude", "latitude")]
  )
}
# append to log-file:
logfile <- data.frame(ip, Lat = Coords$Lat, Long = Coords$Lon,
                                      LatLong = paste(round(Coords$Lat, 1), round(Coords$Lon, 1), sep = ":"))
log_gmap <- logfile[!is.na(logfile$Lat), ]


require(googleVis) # gvisMap
== Create a standalone Rmath library ==
gmap <- gvisMap(log_gmap, "LatLong",
R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of [http://cran.r-project.org/doc/manuals/R-admin.html#The-standalone-Rmath-library R-admin manual].
                options = list(showTip = TRUE, enableScrollWheel = TRUE,
                              mapType = 'hybrid', useMapTypeControl = TRUE,
                              width = 1024, height = 800))
plot(gmap)
</pre>
[[File:GoogleVis.png|200px]]


The plot.gvis() method in googleVis packages also teaches the startDynamicHelp() function in the tools package, which was used to launch a http server. See
Here is my experience based on R 3.0.2 on Windows OS.
[http://jeffreyhorner.tumblr.com/page/3 Jeffrey Horner's note about deploying Rook App].


=== Map ===
=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
==== [https://rstudio.github.io/leaflet/ leaflet] ====
Suppose we have downloaded R source code and build R from its source. See [[R#Build_R_from_its_source|Build_R_from_its_source]]. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.
* rstudio.github.io/leaflet/#installation-and-use
<pre>
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>


==== choroplethr ====
=== Use Rmath library in our code ===
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
<pre>
* http://www.arilamstein.com/blog/2015/06/25/learn-to-map-census-data-in-r/
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
* http://www.arilamstein.com/blog/2015/09/10/user-question-how-to-add-a-state-border-to-a-zip-code-map/
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.


==== ggplot2 ====
# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
# It is OK to save the cpp file under any directory.


=== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ===
# Force to link against the static library <libRmath.a>
See an example from [[R#RJSONIO|RJSONIO]] above.
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe


=== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ===
# Force to link against dynamic library <Rmath.dll>
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
</pre>
Test the executable program. Note that the executable program ''RmathEx1.exe'' can be transferred to and run in another computer without R installed. Isn't it cool!
<pre>
c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>


=== gtrendsR - Google Trends ===
Below is the cpp program <RmathEx1.cpp>.
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
<pre>
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
//RmathEx1.cpp
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=microarray%20analysis microarray analysis] from 2004-04-01
#define MATHLIB_STANDALONE
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=ngs%20next%20generation%20sequencing ngs next generation sequencing] from 2004-04-01
#include <iostream>
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=dna%20sequencing dna sequencing] from 2004-01-01.
#include "Rmath.h"
* [https://trends.google.com/trends/explore?date=2004-01-01%202017-09-04&q=rna%20sequencing rna sequencing] from 2004-01-01. It can be seen RNA sequencing >> DNA sequencing.
* [http://www.kdnuggets.com/2017/09/python-vs-r-data-science-machine-learning.html?utm_content=buffere1df7&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer Python vs R – Who Is Really Ahead in Data Science, Machine Learning?] and [https://stackoverflow.blog/2017/09/06/incredible-growth-python/ The Incredible Growth of Python] by [https://twitter.com/drob?lang=en David Robinson]


=== quantmod ===
using std::cout; using std::cin; using std::endl;
[http://www.thertrader.com/2015/12/13/maintaining-a-database-of-price-files-in-r/ Maintaining a database of price files in R]. It consists of 3 steps.


# Initial data downloading
int main()
# Update existing data
{
# Create a batch file
  double x1, x2;
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;


=== [http://cran.r-project.org/web/packages/Rcpp/index.html Rcpp] ===
  cout << "Prob(Z <= " << x1 << ") = " <<
 
    pnorm(x1, 0, 1, 1, 0)  << endl;
* [http://lists.r-forge.r-project.org/pipermail/rcpp-devel/ Discussion archive]
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
* (Video) [https://www.rstudio.com/resources/videos/extending-r-with-c-a-brief-introduction-to-rcpp/ Extending R with C++: A Brief Introduction to Rcpp]
    pchisq(x2, 1, 1, 0) << endl;
* [http://dirk.eddelbuettel.com/blog/2017/06/13/#007_c++14_r_travis C++14, R and Travis -- A useful hack]
  return 0;
 
}
It may be necessary to install dependency packages for RcppEigen.
</pre>
<syntaxhighlight lang='rsplus'>
sudo apt-get install libblas-dev liblapack-dev
sudo apt-get install gfortran
</syntaxhighlight>


==== Speed Comparison ====
== Calling R.dll directly ==
* [http://blog.revolutionanalytics.com/2015/06/a-comparison-of-high-performance-computing-techniques-in-r.html A comparison of high-performance computing techniques in R]. It compares Rcpp to an R looping operator (like mapply), a parallelized version of a looping operator (like mcmapply), explicit parallelization, via the parallel package or the ParallelR suite.
See Chapter 8.2.2 of [http://cran.r-project.org/doc/manuals/R-exts.html#Calling-R_002edll-directly|Writing R Extensions]. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.
* In the following example, C++ avoids the overhead of creating an intermediate object (eg vector of the same length as the original vector). The c++ uses an intermediate scalar. So C++ wins R over memory management in this case. <syntaxhighlight lang='rsplus'>
# http://blog.mckuhn.de/2016/03/avoiding-unnecessary-memory-allocations.html
library(Rcpp)


`%count<%` <- cppFunction('
== Create HTML report ==
size_t count_less(NumericVector x, NumericVector y) {
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor. See [[Genome#ReportingTools|Genome->ReportingTools]].
  const size_t nx = x.size();
  const size_t ny = y.size();
  if (nx > 1 & ny > 1) stop("Only one parameter can be a vector!");
  size_t count = 0;
  if (nx == 1) {
    double c = x[0];
    for (int i = 0; i < ny; i++) count += c < y[i];
  } else {
    double c = y[0];
    for (int i = 0; i < nx; i++) count += x[i] < c;
  }
  return count;
}
')


set.seed(42)
=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.


N <- 10^7
* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
v <- runif(N, 0, 10000)
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
* [http://gforge.se/2020/07/news-in-htmltable-2-0/ News in htmlTable 2.0]


# Testing on my ODroid xu4 running ubuntu 15.10
=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
system.time(sum(v < 5000))
* https://github.com/renkun-ken/formattable
#  user  system elapsed
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
#  1.135  0.305  1.453
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]
system.time(v %count<% 5000)
#  user  system elapsed
#  0.535  0.000  0.540
</syntaxhighlight>
* [http://blog.ephorie.de/why-r-for-data-science-and-not-python Why R for data science – and not Python?]<syntaxhighlight lang='rsplus'>
library(Rcpp)
bmi_R <- function(weight, height) {
  weight / (height * height)
}
bmi_R(80, 1.85) # body mass index of person with 80 kg and 185 cm
## [1] 23.37473
cppFunction("
  float bmi_cpp(float weight, float height) {
    return weight / (height * height);
  }
")
bmi_cpp(80, 1.85) # same with cpp function
## [1] 23.37473
</syntaxhighlight>
* [https://www.enchufa2.es/archives/boost-the-speed-of-r-calls-from-rcpp.html Boost the speed of R calls from Rcpp]


==== Use Rcpp in RStudio ====
=== [https://github.com/crubba/htmltab htmltab] package ===
RStudio makes it easy to use Rcpp package.
This package is NOT used to CREATE html report but EXTRACT html table.


Open RStudio, click New File -> C++ File. It will create a C++ template on the RStudio editor
=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
<pre>
Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.
#include <Rcpp.h>
using namespace Rcpp;


// Below is a simple example of exporting a C++ function to R. You can
== Create academic report ==
// source this function into an R session using the Rcpp::sourceCpp
[http://cran.r-project.org/web/packages/reports/index.html reports] package in CRAN and in [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.
// function (or via the Source button on the editor toolbar)


// For more on using Rcpp click the Help button on the editor toolbar
== Create pdf and epub files ==
 
{{Pre}}
// [[Rcpp::export]]
# Idea:
int timesTwo(int x) {
#        knitr        pdflatex
  return x * 2;
#  rnw -------> tex ----------> pdf
}
library(knitr)
knit("example.rnw") # create example.tex file
</pre>
</pre>
Now in R console, type
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
<pre>
{{Pre}}
library(Rcpp)
git clone https://github.com/yihui/knitr-examples.git
sourceCpp("~/Downloads/timesTwo.cpp")
timesTwo(9)
# [1] 18
</pre>
</pre>
See more examples on http://adv-r.had.co.nz/Rcpp.html and [http://blog.revolutionanalytics.com/2017/08/kmeans-r-rcpp.html Calculating a fuzzy kmeans membership matrix]
* <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run '''sudo apt-get install texlive-fonts-recommended''' to install missing fonts. It works!


If we wan to test Boost library, we can try it in RStudio. Consider the following example in [http://stackoverflow.com/questions/19034564/can-the-bh-r-package-link-to-boost-math-and-numeric stackoverflow.com].
To see a real example, check out [http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html DESeq2] package (inst/doc subdirectory). In addition to DESeq2, I also need to install '''DESeq, BiocStyle, airway, vsn, gplots''', and '''pasilla''' packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.
<pre>
// [[Rcpp::depends(BH)]]
#include <Rcpp.h>
#include <boost/foreach.hpp>
#include <boost/math/special_functions/gamma.hpp>


#define foreach BOOST_FOREACH
Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
 
{{Pre}}
using namespace boost::math;
# Idea:
#        knitr        pandoc
#  rmd -------> md ----------> pdf


//[[Rcpp::export]]
git clone https://github.com/yihui/knitr-examples.git
Rcpp::NumericVector boost_gamma( Rcpp::NumericVector x ) {
cd knitr-examples
  foreach( double& elem, x ) {
R -e "library(knitr); knit('001-minimal.Rmd')"
    elem = boost::math::tgamma(elem);
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
  };
 
  return x;
}
</pre>
</pre>
Then the R console
<pre>
boost_gamma(0:10 + 1)
#  [1]      1      1      2      6      24    120    720    5040  40320
# [10]  362880 3628800


identical( boost_gamma(0:10 + 1), factorial(0:10) )
To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
# [1] TRUE
{{Pre}}
</pre>
# Idea:
#       knitr        pandoc
#  rnw -------> tex ----------> markdown or epub


==== Example 1. convolution example ====
library(knitr)
First, Rcpp package should be installed (I am working on Linux system). Next we try one example shipped in Rcpp package.
knit("DESeq2.Rnw") # create DESeq2.tex
 
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
PS. If R was not available in global environment (such as built by ourselves), we need to modify 'Makefile' file by replacing 'R' command with its complete path (4 places).
<pre>
cd ~/R/x86_64-pc-linux-gnu-library/3.0/Rcpp/examples/ConvolveBenchmarks/
make
R
</pre>
Then type the following in an R session to see how it works. Note that we don't need to issue '''library(Rcpp)''' in R.
<pre>
dyn.load("convolve3_cpp.so")
x <- .Call("convolve3cpp", 1:3, 4:6)
x # 4 13 28 27 18
</pre>
</pre>


If we have our own cpp file, we need to use the following way to create dynamic loaded library file. Note that the  character ([http://bash.cyberciti.biz/guide/Command_substitution grave accent]) ` is not (single quote)'. If you mistakenly use ', it won't work.
Convert tex to epub
<pre>
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
export PKG_CXXFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
export PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
R CMD SHLIB xxxx.cpp
</pre>


==== Example 2. Use together with inline package ====
=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
* http://adv-r.had.co.nz/C-interface.html#calling-c-functions-from-r
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
<pre>
library(inline)
src <-'
Rcpp::NumericVector xa(a);
Rcpp::NumericVector xb(b);
int n_xa = xa.size(), n_xb = xb.size();


Rcpp::NumericVector xab(n_xa + n_xb - 1);
* https://rmarkdown.rstudio.com/lesson-7.html
for (int i = 0; i < n_xa; i++)
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
for (int j = 0; j < n_xb; j++)
* http://kbroman.org/knitr_knutshell/pages/figs_tables.html
xab[i + j] += xa[i] * xb[j];
* https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/
return xab;
* [https://cran.r-project.org/web/packages/kableExtra/vignettes/awesome_table_in_html.html kableExtra] package
'
fun <- cxxfunction(signature(a = "numeric", b = "numeric"),
src, plugin = "Rcpp")
fun(1:3, 1:4)
# [1] 1  4 10 16 17 12
</pre>


==== Example 3. Calling an R function ====
== Create Word report ==


==== [http://cran.r-project.org/web/packages/RcppParallel/index.html RcppParallel] ====
=== Using the power of Word ===
[https://www.rforecology.com/post/exporting-tables-from-r-to-microsoft-word/ How to go from R to nice tables in Microsoft Word]


=== [http://cran.r-project.org/web/packages/caret/index.html caret] ===
=== knitr + pandoc ===
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* https://www.r-project.org/conferences/useR-2013/Tutorials/kuhn/user_caret_2up.pdf
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* https://github.com/cran/caret source code mirrored on github
* http://rmarkdown.rstudio.com/articles_docx.html
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/


=== Tool for connecting Excel with R ===
It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.
* https://bert-toolkit.com/
<pre>
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
# Idea:
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html
#        knitr      pandoc
#  rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files
</pre>
and then
<pre>
FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
</pre>
Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.


=== Read/Write Excel files package ===
Another way is
* http://www.milanor.net/blog/?p=779
<pre>
* [https://www.displayr.com/how-to-read-an-excel-file-into-r/?utm_medium=Feed&utm_source=Syndication flipAPI]. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
library(pander)
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
name = "demo"
* [http://cran.r-project.org/web/packages/openxlsx/index.html openxlsx]: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
knit(paste0(name, ".Rmd"), encoding = "utf-8")
** It can't be used to open .xls or .xlm files.
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
</pre>
** [https://fabiomarroni.wordpress.com/2018/08/07/use-r-to-write-multiple-tables-to-a-single-excel-file/ Use R to write multiple tables to a single Excel file]
* [https://github.com/hadley/readxl readxl]: it does not depend on anything although it can only read but not write Excel files.  [https://github.com/rstudio/webinars/tree/master/36-readxl readxl webinar]. One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a '''tbl_df''', a '''tbl''' and a '''data.frame.'''
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R


Tested it on Ubuntu machine with R 3.1.3 using <BRCA.xls> file. Usage:
Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:
<syntaxhighlight lang='rsplus'>
* A pdf file: pandoc -s report.md -t latex -o report.pdf
library(readxl)
* A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
read_excel(path, sheet = 1, col_names = TRUE, col_types = NULL, na = "", skip = 0)
* Openoffice: pandoc report.md -o report.odt
</syntaxhighlight>
* Word docx: pandoc report.md -o report.docx
For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).
<syntaxhighlight lang='rsplus'>
> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
  UniqueID (Double-click) CloneID UGCluster
1                  HK1A1  21652 Hs.445981
2                  HK1A2  22012 Hs.119177
3                  HK1A4  22293 Hs.501376
                                                    Name Symbol EntrezID
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1    1495
2                              ADP-ribosylation factor 3  ARF3      377
3                          Uroporphyrinogen III synthase  UROS    7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1  5.000000        5q31.2              <NA>      1
2  12.000000        12q13              <NA>      1
3      <NA> 10q25.2-q26.3              <NA>      1
</syntaxhighlight>


The hidden worksheets become visible (Not sure what are those first rows mean in the output).
We can also create the epub file for reading on Kobo ereader. For example, download [https://gist.github.com/jeromyanglim/2716336 this file] and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!
<syntaxhighlight lang='rsplus'>
<pre>
> excel_sheets("~/Downloads/BRCA.xls")
knit("example.Rmd")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00
pandoc("example.md", format="epub")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00
</pre>
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00
[1] "Experiment descriptors" "Filtered log ratio"     "Gene identifiers"     
[4] "Gene annotations"      "CollateInfo"            "GeneSubsets"         
[7] "GeneSubsetsTemp"     
</syntaxhighlight>


The Chinese character works too.
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
<syntaxhighlight lang='rsplus'>
<pre>
> read_excel("~/Downloads/testChinese.xlsx", 1)
> pandoc("Rmd_to_Epub.md", format="epub")
  中文 B C
executing pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
1     a b c
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
2    1 2 3
Error in (function (input, format, ext, cfg) : conversion failed
</syntaxhighlight>
In addition: Warning message:
running command 'pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1
</pre>


To read all worksheets we need a convenient function
=== pander ===
<syntaxhighlight lang='rsplus'>
Try pandoc[1] with a minimal reproducible example, you might give a try to my "[http://cran.r-project.org/web/packages/pander/ pander]" package [2] too:
read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc[[1]]) is a tibble.
</syntaxhighlight>


=== [https://cran.r-project.org/web/packages/readr/ readr] ===
<pre>
Note: '' '''readr''' package is not designed to read Excel files.''
library(pander)
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
            output = tempfile(), convert = 'docx')
</pre>
Where the content of the "minimal.brew" file is something you might have
got used to with Sweave - although it's using "brew" syntax instead. See
the examples of pander [3] for more details. Please note that pandoc should
be installed first, which is pretty easy on Windows.


Compared to base equivalents like '''read.csv()''', '''readr''' is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.
# http://johnmacfarlane.net/pandoc/
# http://rapporter.github.com/pander/
# http://rapporter.github.com/pander/#examples


[https://blog.rstudio.org/2016/08/05/readr-1-0-0/ 1.0.0] released.
=== R2wd ===
Use [http://cran.r-project.org/web/packages/R2wd/ R2wd] package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.  
<pre>
> library(R2wd)
> wdGet()
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
    installstatconnDCOM()


The '''read_csv()''' function from the '''readr''' package is as fast as '''fread()''' function from '''data.table''' package. ''For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv().'' See 5.3 of Efficient R Programming book.
This will download and install the current version of statconnDCOM


Note that '''fread()''' can read-n a selection of the columns.
You will need a working Internet connection
because installation needs to download a file.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
  argument is of length zero
</pre>


=== Colors ===
The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
* [http://colorspace.r-forge.r-project.org/articles/colorspace.html colorspace]: A Toolbox for Manipulating and Assessing Colors and Palettes.
 
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
=== Convert from pdf to word ===
* palette_explorer() function from the [https://cran.r-project.org/web/packages/tmaptools/index.html tmaptools] package. See [https://www.computerworld.com/article/3184778/data-analytics/6-useful-r-functions-you-might-not-know.html selecting color palettes with shiny].
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert
* [http://www.ucl.ac.uk/~zctpep9/Archived%20webpages/Cookbook%20for%20R%20%C2%BB%20Colors%20(ggplot2).htm Cookbook for R]
* [http://ggplot2.tidyverse.org/reference/scale_brewer.html Sequential, diverging and qualitative colour scales/palettes from colorbrewer.org]: scale_colour_brewer(), scale_fill_brewer(), ...
* http://colorbrewer2.org/
* It seems there is no choice of getting only 2 colors no matter which set name we can use
* To see the set names used in brewer.pal, see
** RColorBrewer::display.brewer.all()
** For example, [http://colorbrewer2.org/#type=qualitative&scheme=Set1&n=4 Set1] from http://colorbrewer2.org/
* To list all R color names, colors()
* [https://stackoverflow.com/questions/28461326/convert-hex-color-code-to-color-name convert hex value to color names] <syntaxhighlight lang='rsplus'>
library(plotrix)
sapply(rainbow(4), color.id)
sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
</syntaxhighlight>


Below is an example using the option ''scale_fill_brewer''(palette = "[http://colorbrewer2.org/#type=qualitative&scheme=Paired&n=9 Paired]"). See the source code at [https://gist.github.com/JohannesFriedrich/c7d80b4e47b3331681cab8e9e7a46e17 gist]. Note that only 'set1' and 'set3' palettes in '''qualitative scheme''' can support up to 12 classes.  
=== rtf ===
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.


According to the information from the colorbrew website, '''qualitative''' schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.
=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
Package xtable will produce html output.
{{Pre}}
print(xtable(X), type="html")
</pre>


[[File:GgplotPalette.svg|300px]]
If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.


=== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ===
=== officer ===
See [[Ggplot2|ggplot2]]
<ul>
<li>[https://cran.r-project.org/web/packages/officer/index.html CRAN]. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.
<li>The [https://gist.github.com/arraytools/4f182b036ae7f95a31924ba5d5d3f069 gist] includes a comprehensive example that encompasses various elements such as sections, subsections, and tables. It also incorporates a detailed paragraph, along with visual representations created using base R plots and ggplots.
<li>Add a line space
<pre>
doc <- body_add_par(doc, "")


=== Data Manipulation & Tidyverse ===
# Function to add n line spaces
* [https://www.rstudio.com/resources/webinars/pipelines-for-data-analysis-in-r/ Pipelines for data analysis in R], [https://www.rstudio.com/resources/videos/data-science-in-the-tidyverse/ Data Science in the Tidyverse]
body_add_par_n <- function (doc, n) {
  for(i in 1:n){
    doc <- body_add_par(doc, "")
  }
  return(doc)
}
body_add_par_n(3)
</pre>
<li>[https://ardata-fr.github.io/officeverse/officer-for-word.html Figures] from the documentation of '''officeverse'''.
<li>See [https://stackoverflow.com/a/25427314 Data frame to word table?].
<li>See [[Office#Tables|Office]] page for some code.
<li>[https://www.r-bloggers.com/2020/07/how-to-read-and-create-word-documents-in-r/ How to read and create Word Documents in R] where we can extracting tables from Word Documents.
<pre>
<pre>
  Import
x = read_docx("myfile.docx")
    |
content <- docx_summary(x) # a vector
    | readr, readxl
grep("nlme", content$text, ignore.case = T, value = T)
    | haven, DBI, httr  +----- Visualize ------+
    |                    |    ggplot2, ggvis    |
    |                    |                      |
  Tidy ------------- Transform
  tibble              dplyr                  Model
  tidyr                  |                    broom
                          +------ Model ---------+
</pre>
</pre>
* [http://r4ds.had.co.nz/ R for Data Science] and [http://tidyverse.org/ tidyverse] package (it is a collection of '''ggplot2, tibble, tidyr, readr, purrr''' & '''dplyr''' packages).
</ul>
** tidyverse, among others, was used at [http://juliasilge.com/blog/Mining-CRAN-DESCRIPTION/ Mining CRAN DESCRIPTION Files] (tbl_df(), %>%, summarise(), count(), mutate(), arrange(), unite(), ggplot(), filter(), select(), ...). Note that there is a problem to reproduce the result. I need to run ''cran <- cran[, -14]'' to remove the MD5sum column.
** [http://brettklamer.com/diversions/statistical/compile-r-for-data-science-to-a-pdf/ Compile R for Data Science to a PDF]
* [https://www.rstudio.com/wp-content/uploads/2015/02/data-wrangling-cheatsheet.pdf Data Wrangling with dplyr and tidyr Cheat Sheet]
* [https://hbctraining.github.io/Intro-to-R/lessons/07_intro_tidyverse.html Data Wrangling with Tidyverse] from the Harvard Chan School of Public Health.
* [http://datascienceplus.com/best-packages-for-data-manipulation-in-r/ Best packages for data manipulation in R]. It demonstrates to perform the same tasks using '''data.table''' and '''dplyr''' packages. '''data.table''' is faster and it may be a go-to package when performance and memory are the constraints.


==== [http://rpubs.com/danmirman/Rgroup-part1 5 most useful data manipulation functions] ====
== Powerpoint ==
* subset() for making subsets of data (natch)
<ul>
* merge() for combining data sets in a smart and easy way
<li>[https://cran.r-project.org/web/packages/officer/index.html officer] package  (formerly ReporteRs). [http://theautomatic.net/2020/07/28/how-to-create-powerpoint-reports-with-r/ How to create powerpoint reports with R]
* '''melt()'''-reshape2 package for converting from wide to long data formats. See an example [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others here] where we want to combine multiple columns of values into 1 column.
</li>
* dcast()-reshape2 package for converting from long to wide data formats (or just use [https://datascienceplus.com/building-barplots-with-error-bars/ tapply()]), and for making summary tables
<li>[https://davidgohel.github.io/flextable/ flextable] (imports '''officer''')
* ddply()-plyr package for doing split-apply-combine operations, which covers a huge swath of the most tricky data operations
</li>
<li>[https://stackoverflow.com/a/21558466 R data.frame to table image for presentation].  
<pre>
library(gridExtra)
grid.newpage()
grid.table(mydf)
</pre>
</li>
<li>[https://bookdown.org/yihui/rmarkdown/powerpoint-presentation.html Rmarkdown]
</li>
</ul>


==== [https://cran.r-project.org/web/packages/data.table/index.html data.table] ====
== PDF manipulation ==
Fast aggregation of large data (e.g. 100GB in RAM or just several GB size file), fast ordered joins, fast add/modify/delete of columns by group using no copies at all, list columns and a fast file reader (fread).
[https://github.com/pridiltal/staplr staplr]


[https://github.com/Rdatatable/data.table/wiki/Installation#openmp-enabled-compiler-for-mac OpenMP enabled compiler for Mac]. This instruction works on my Mac El Capitan (10.11.6) when I need to upgrade the data.table version from 1.11.4 to 1.11.6.
== R Graphs Gallery ==
* [https://www.facebook.com/pages/R-Graph-Gallery/169231589826661 Romain François]
* [http://shinyapps.stat.ubc.ca/r-graph-catalog/ R Graph Catalog] written using R + Shiny. The source code is available on [https://github.com/jennybc/r-graph-catalog Github].
* Forest plot. See the packages [https://cran.r-project.org/web/packages/rmeta/index.html rmeta] and [https://cran.r-project.org/web/packages/forestplot/ forestplot]. The forest plot can be used to plot the quantities like relative risk (with 95% CI) in survival data.
** [http://www.danieldsjoberg.com/bstfun/dev/reference/add_inline_forest_plot.html Inline forest plot]


Question: how to make use multicore with data.table package?
== COM client or server ==


* [https://github.com/rstudio/cheatsheets/raw/master/datatable.pdf Cheat sheet] from [https://www.rstudio.com/resources/cheatsheets/ RStudio]
=== Client ===
* [https://www.r-bloggers.com/importing-data-into-r-part-two/ Reading large data tables in R]
* [http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
<syntaxhighlight lang='rsplus'>
* [https://www.r-bloggers.com/2024/06/how-to-execute-vba-code-in-excel-via-r-using-rdcomclient/ How to Execute VBA Code in Excel via R using RDCOMClient]
library(data.table)
x <- fread("mylargefile.txt")
</syntaxhighlight>
* Note that '''x[, 2]'' always return 2. If you want to do the thing you want, use ''x[, 2, with=FALSE]'' or ''x[, V2]'' where V2 is the header name. See the FAQ #1 in [http://datatable.r-forge.r-project.org/datatable-faq.pdf data.table].
* [http://r-norberg.blogspot.com/2016/06/understanding-datatable-rolling-joins.html Understanding data.table Rolling Joins]
* [https://rollingyours.wordpress.com/2016/06/14/fast-aggregation-of-large-data-with-the-data-table-package/ Intro to The data.table Package]
* In the [https://cran.r-project.org/web/packages/data.table/vignettes/datatable-intro-vignette.html Introduction to data.table] vignette, the data.table::order() function is SLOWER than base::order() from my Odroid xu4 (running Ubuntu 14.04.4 trusty on uSD)
<syntaxhighlight lang='rsplus'>
odt = data.table(col=sample(1e7))
(t1 <- system.time(ans1 <- odt[base::order(col)]))  ## uses order from base R
#  user  system elapsed
#  2.730  0.210  2.947
(t2 <- system.time(ans2 <- odt[order(col)]))        ## uses data.table's order
#  user  system elapsed
#  2.830  0.215  3.052
(identical(ans1, ans2))
# [1] TRUE
</syntaxhighlight>
* [https://jangorecki.github.io/blog/2016-06-30/Boost-Your-Data-Munging-with-R.html Boost Your Data Munging with R]
* [https://www.rdocumentation.org/packages/data.table/versions/1.12.0/topics/rbindlist rbindlist()]. One problem, it uses too much memory. In fact, when I try to analyze R package downloads, the command "dat <- rbindlist(logs)" uses up my 64GB memory (OS becomes unresponsive).


==== reshape & reshape2 ====
=== Server ===
* [http://r-exercises.com/2016/07/06/data-shape-transformation-with-reshape/ Data Shape Transformation With Reshape()]
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]
* Use '''acast()''' function in reshape2 package. It will convert data.frame used for analysis to a table-like data.frame good for display.
* http://lamages.blogspot.com/2013/10/creating-matrix-from-long-dataframe.html


==== [http://cran.r-project.org/web/packages/tidyr/index.html tidyr] and benchmark ====
== Use R under proxy ==
An evolution of reshape2. It's designed specifically for data tidying (not general reshaping or aggregating) and works well with dplyr data pipelines.
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy


* [https://cran.r-project.org/web/packages/tidyr/vignettes/tidy-data.html vignette("tidy-data")] & [https://github.com/rstudio/cheatsheets/blob/master/data-import.pdf Cheat sheet]
== RStudio ==
* Main functions
* [https://github.com/rstudio/rstudio Github]
** Reshape data: '''gather()''' & '''spread()'''
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
** Split cells: '''separate()''' & '''unite()'''
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]
** Handle missing: drop_na() & fill() & replace_na()
* http://blog.rstudio.org/2014/07/22/introducing-tidyr/
* http://rpubs.com/seandavi/GEOMetadbSurvey2014
* http://timelyportfolio.github.io/rCharts_factor_analytics/factors_with_new_R.html
* [http://www.milanor.net/blog/reshape-data-r-tidyr-vs-reshape2/ tidyr vs reshape2]
* [http://r-posts.com/benchmarking-cast-in-r-from-long-data-frame-to-wide-matrix/ Benchmarking cast in R from long data frame to wide matrix]
 
Make wide tables long with '''gather()''' (see 6.3.1 of Efficient R Programming)
<syntaxhighlight lang='rsplus'>
library(tidyr)
library(efficient)
data(pew) # wide table
dim(pew) # 18 x 10,  (religion, '<$10k', '$10--20k', '$20--30k', ..., '>150k')
pewt <- gather(data = pew, key = Income, value = Count, -religion)
dim(pew) # 162 x 3,  (religion, Income, Count)


args(gather)
=== rstudio.cloud ===
# function(data, key, value, ..., na.rm = FALSE, convert = FALSE, factor_key = FALSE)
https://rstudio.cloud/
</syntaxhighlight>
where the three arguments of gather() requires:
* data: a data frame in which column names will become row vaues
* key: the name of the categorical variable into which the column names in the original datasets are converted.
* value: the name of cell value columns


In this example, the 'religion' column will not be included (-religion).
=== Launch RStudio ===
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]


==== dplyr, plyr packages ====
=== Create .Rproj file ===
* Essential functions: 3 rows functions, 3 column functions and 1 mixed function.
If you have an existing package that doesn't have an .Rproj file, you can use '''devtools::use_rstudio("path/to/package")''' to add it.
<pre>
          select, mutate, rename
            +------------------+
filter      +                  +
arrange    +                  +
group_by    +                  +
            + summarise        +
            +------------------+
</pre>
* These functions works on data frames and tibble objects.
<syntaxhighlight lang='rsplus'>
iris %>% filter(Species == "setosa") %>% count()
head(iris %>% filter(Species == "setosa") %>% arrange(Sepal.Length))
</syntaxhighlight>
* [http://r4ds.had.co.nz/transform.html Data Transformation] in the book '''R for Data Science'''. Five key functions in the '''dplyr''' package:
** Filter rows: filter()
** Arrange rows: arrange()
** Select columns: select()
** Add new variables: mutate()
** Grouped summaries: group_by() & summarise()
<syntaxhighlight lang='rsplus'>
# filter
jan1 <- filter(flights, month == 1, day == 1)
filter(flights, month == 11 | month == 12)
filter(flights, arr_delay <= 120, dep_delay <= 120)
df <- tibble(x = c(1, NA, 3))
filter(df, x > 1)
filter(df, is.na(x) | x > 1)


# arrange
With an RStudio project file, you can
arrange(flights, year, month, day)
* Restore .RData into workspace at startup
arrange(flights, desc(arr_delay))
* Save workspace to .RData on exit (or '''save.image'''("Robj.RData") & load("Robj.RData"))
* Always save history (even if no saving .RData, '''savehistory'''(".Rhistory") & loadhistory(".Rhistory"))
* etc


# select
=== package search ===
select(flights, year, month, day)
https://github.com/RhoInc/CRANsearcher
select(flights, year:day)
select(flights, -(year:day))


# mutate
=== Git ===
flights_sml <- select(flights,
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
  year:day,
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]
  ends_with("delay"),
  distance,
  air_time
)
mutate(flights_sml,
  gain = arr_delay - dep_delay,
  speed = distance / air_time * 60
)
# if you only want to keep the new variables
transmute(flights,
  gain = arr_delay - dep_delay,
  hours = air_time / 60,
  gain_per_hour = gain / hours
)


# summarise()
== Visual Studio ==
by_day <- group_by(flights, year, month, day)
[http://blog.revolutionanalytics.com/2017/05/r-and-python-support-now-built-in-to-visual-studio-2017.html R and Python support now built in to Visual Studio 2017]
summarise(by_day, delay = mean(dep_delay, na.rm = TRUE))


# pipe. Note summarise() can return more than 1 variable.
== List files using regular expression ==
delays <- flights %>%
* Extension
  group_by(dest) %>%
<pre>
  summarise(
list.files(pattern = "\\.txt$")
    count = n(),
</pre>
    dist = mean(distance, na.rm = TRUE),
where the dot (.) is a metacharacter. It is used to refer to any character.
    delay = mean(arr_delay, na.rm = TRUE)
* Start with
  ) %>%
<pre>
  filter(count > 20, dest != "HNL")
list.files(pattern = "^Something")
flights %>%
</pre>
  group_by(year, month, day) %>%
  summarise(mean = mean(dep_delay, na.rm = TRUE))
</syntaxhighlight>
* Videos
** [https://youtu.be/jWjqLW-u3hc Hands-on dplyr tutorial for faster data manipulation in R] by Data School. At time 17:00, it compares the '''%>%''' operator, '''with()''' and '''aggregate()''' for finding group mean.
** https://youtu.be/aywFompr1F4 (shorter video) by Roger Peng
** https://youtu.be/8SGif63VW6E by Hadley Wickham
** [https://www.rstudio.com/resources/videos/tidy-eval-programming-with-dplyr-tidyr-and-ggplot2/ Tidy eval: Programming with dplyr, tidyr, and ggplot2]. Bang bang "!!" operator was introduced for use in a function call.
* [https://csgillespie.github.io/efficientR/data-carpentry.html#dplyr Efficient R Programming]
* [http://www.r-exercises.com/2017/07/19/data-wrangling-transforming-23/ Data wrangling: Transformation] from R-exercises.
* [https://rollingyours.wordpress.com/2016/06/29/express-intro-to-dplyr/ Express Intro to dplyr] by rollingyours.
* [https://martinsbioblogg.wordpress.com/2017/05/21/using-r-when-using-do-in-dplyr-dont-forget-the-dot/ the dot].
* [http://martinsbioblogg.wordpress.com/2013/03/24/using-r-reading-tables-that-need-a-little-cleaning/ stringr and plyr] A '''data.frame''' is pretty much a list of vectors, so we use plyr to apply over the list and stringr to search and replace in the vectors.
* https://randomjohn.github.io/r-maps-with-census-data/ dplyr and stringr are used
* [https://datascienceplus.com/5-interesting-subtle-insights-from-ted-videos-data-analysis-in-r/ 5 interesting subtle insights from TED videos data analysis in R]
* [https://www.mango-solutions.com/blog/what-is-tidy-eval-and-why-should-i-care What is tidy eval and why should I care?]


==== stringr ====
Using '''Sys.glob()"' as
* https://www.rstudio.com/wp-content/uploads/2016/09/RegExCheatsheet.pdf
<pre>
* [https://github.com/rstudio/cheatsheets/raw/master/strings.pdf stringr Cheat sheet] (2 pages, this will immediately download the pdf file)
> Sys.glob("~/Downloads/*.txt")
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
</pre>


==== [https://github.com/smbache/magrittr magrittr] ====
== Hidden tool: rsync in Rtools ==
* [https://cran.r-project.org/web/packages/magrittr/vignettes/magrittr.html Vignettes]
<pre>
* [http://www.win-vector.com/blog/2018/04/magrittr-and-wrapr-pipes-in-r-an-examination/ magrittr and wrapr Pipes in R, an Examination]
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe


Instead of nested statements, it is using pipe operator '''%>%'''. So the code is easier to read. Impressive!
sent 323142 bytes received 31 bytes 646346.00 bytes/sec
<syntaxhighlight lang='rsplus'>
total size is 1198416 speedup is 3.71
x %>% f    # f(x)
x %>% f(y) # f(x, y)
x %>% f(arg=y) # f(x, arg=y)
x %>% f(z, .) # f(z, x)
x %>% f(y) %>% g(z) #  g(f(x, y), z)


x %>% select(which(colSums(!is.na(.))>0))  # remove columns with all missing data
c:\Rtools\bin>
x %>% select(which(colSums(!is.na(.))>0)) %>% filter((rowSums(!is.na(.))>0)) # remove all-NA columns _and_ rows
</pre>
</syntaxhighlight>
* [http://www.win-vector.com/blog/2018/03/r-tip-make-arguments-explicit-in-magrittr-dplyr-pipelines/ Make Arguments Explicit in magrittr/dplyr Pipelines]
<syntaxhighlight lang='rsplus'>
suppressPackageStartupMessages(library("dplyr"))
starwars %>%
  filter(., height > 200) %>%
  select(., height, mass) %>%
  head(.)
# instead of
starwars %>%
  filter(height > 200) %>%
  select(height, mass) %>%
  head
</syntaxhighlight>
* [https://stackoverflow.com/questions/27100678/how-to-extract-subset-an-element-from-a-list-with-the-magrittr-pipe Subset an element from a list]
<syntaxhighlight lang='rsplus'>
iris$Species
iris[["Species"]]


iris %>%
Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also [https://superuser.com/a/69764 rsync file permissions on windows].
`[[`("Species")


iris %>%
== Install rgdal package (geospatial Data) on ubuntu ==
`[[`(5)
Terminal
{{Pre}}
sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411
</pre>


iris %>%
R
  subset(select = "Species")
{{Pre}}
</syntaxhighlight>
install.packages("rgdal")
* '''Split-apply-combine''': group + summarize + sort/arrange + top n. The following example is from [https://csgillespie.github.io/efficientR/data-carpentry.html#data-aggregation Efficient R programming].
</pre>
<syntaxhighlight lang='rsplus'>
data(wb_ineq, package = "efficient")
wb_ineq %>%
  filter(grepl("g", Country)) %>%
  group_by(Year) %>%
  summarise(gini = mean(gini, na.rm  = TRUE)) %>%
  arrange(desc(gini)) %>%
  top_n(n = 5)
</syntaxhighlight>
* [https://drdoane.com/writing-pipe-friendly-functions/ Writing Pipe-friendly Functions]
* http://rud.is/b/2015/02/04/a-step-to-the-right-in-r-assignments/
* http://rpubs.com/tjmahr/pipelines_2015
* http://danielmarcelino.com/i-loved-this-crosstable/
* http://moderndata.plot.ly/using-the-pipe-operator-in-r-with-plotly/
* Videos
** [https://www.rstudio.com/resources/videos/writing-readable-code-with-pipes/ Writing Readable Code with Pipes]
** [https://youtu.be/iIBTI_qiq9g Pipes in R - An Introduction to magrittr package]
<syntaxhighlight lang='rsplus'>
# Examples from R for Data Science-Import, Tidy, Transform, Visualize, and Model
diamonds <- ggplot2::diamonds
diamonds2 <- diamonds %>% dplyr::mutate(price_per_carat = price / carat)


pryr::object_size(diamonds)
== Install sf package ==
pryr::object_size(diamonds2)
I got the following error even I have installed some libraries.
pryr::object_size(diamonds, diamonds2)
<pre>
checking GDAL version >= 2.0.1... no
configure: error: sf is not compatible with GDAL versions below 2.0.1
</pre>
Then I follow the instruction here
{{Pre}}
sudo apt remove libgdal-dev
sudo apt remove libproj-dev
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable


rnorm(100) %>% matrix(ncol = 2) %>% plot() %>% str()
sudo apt update
rnorm(100) %>% matrix(ncol = 2) %T>% plot() %>% str() # 'tee' pipe
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
    # %T>% works like %>% except that it returns the lefthand side (rnorm(100) %>% matrix(ncol = 2)) 
    # instead of the righthand side.


# If a function does not have a data frame based api, you can use %$%.
sudo apt install libgdal-dev # works on ubuntu 20.04 too
# It explodes out the variables in a data frame.
                            # no need the previous lines
mtcars %$% cor(disp, mpg)
</pre>


# For assignment, magrittr provides the %<>% operator
== Database ==
mtcars <- mtcars %>% transform(cyl = cyl * 2) # can be simplified by
* https://cran.r-project.org/web/views/Databases.html
mtcars %<>% transform(cyl = cyl * 2)
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]
</syntaxhighlight>


Upsides of using magrittr: no need to create intermediate objects, code is easy to read.
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite


When not to use the pipe
'''Creating a new database''':
* your pipes are longer than (say) 10 steps
{{Pre}}
* you have multiple inputs or outputs
library(DBI)
* Functions that use the current environment: assign(), get(), load()
* Functions that use lazy evaluation: tryCatch(), try()


==== outer() ====
mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
dbDisconnect(mydb)
unlink("my-db.sqlite")


==== Genomic sequence ====
# temporary database
* chartr
mydb <- dbConnect(RSQLite::SQLite(), "")
<syntaxhighlight lang='bash'>
dbDisconnect(mydb)
> yourSeq <- "AAAACCCGGGTTTNNN"
</pre>
> chartr("ACGT", "TGCA", yourSeq)
[1] "TTTTGGGCCCAAANNN"
</syntaxhighlight>


==== lobstr package - dig into the internal representation and structure of R objects ====
'''Loading data''':
[https://www.tidyverse.org/articles/2018/12/lobstr/ lobstr 1.0.0]
{{Pre}}
mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)


=== Data Science ===
dbListTables(mydb)
See [[Data_science|Data science]] page


=== [http://cran.r-project.org/web/packages/jpeg/index.html jpeg] ===
dbListFields(con, "mtcars")
If we want to create the image on this wiki left hand side panel, we can use the '''jpeg''' package to read an existing plot and then edit and save it.


We can also use the jpeg package to import and manipulate a jpg image. See [http://moderndata.plot.ly/fun-with-heatmaps-and-plotly/ Fun with Heatmaps and Plotly].
dbReadTable(con, "mtcars")
 
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().
 
=== [https://cran.r-project.org/web/packages/cairoDevice/ cairoDevice] ===
PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).
 
For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
<pre>
sudo apt-get install libgtk2.0-dev libcairo2-dev
</pre>
</pre>


On Windows OS, we may got the error: '''unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' '''. We need to follow the instruction in [http://tolstoy.newcastle.edu.au/R/e6/help/09/05/15613.html here].
'''Queries''':
{{Pre}}
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


=== [http://igraph.org/r/ igraph] ===
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]


=== Identifying dependencies of R functions and scripts ===
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
<syntaxhighlight lang='rsplus'>
library(mvbutils)
foodweb(where = "package:batr")


foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
dbFetch(res)
</pre>


foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
'''Batched queries''':
</syntaxhighlight>
{{Pre}}
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


=== [http://cran.r-project.org/web/packages/iterators/ iterators] ===
dbClearResult(rs)
Iterator is useful over for-loop if the data is already a '''collection'''. It can be used to iterate over a vector, data frame, matrix, file
</pre>


Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.
'''Multiple parameterised queries''':
{{Pre}}
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)
</pre>


=== Colors ===
'''Statements''':
* http://www.bauer.uh.edu/parks/truecolor.htm Interactive RGB, Alpha and Color Picker
{{Pre}}
* http://deanattali.com/blog/colourpicker-package/ Not sure what it is doing
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
* [http://www.lifehack.org/484519/how-to-choose-the-best-colors-for-your-data-charts How to Choose the Best Colors For Your Data Charts]
#> [1] 0
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
* [http://sape.inf.usi.ch/quick-reference/ggplot2/colour Color names in R]
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)
</pre>


==== [http://rpubs.com/gaston/colortools colortools] ====
=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
Tools that allow users generate color schemes and palettes
Manipulate R data frames using SQL. Depends on RSQLite. [http://datascienceplus.com/a-use-of-gsub-reshape2-and-sqldf-with-healthcare-data/ A use of gsub, reshape2 and sqldf with healthcare data]


==== [https://github.com/daattali/colourpicker colourpicker] ====
=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===
A Colour Picker Tool for Shiny and for Selecting Colours in Plots


==== [https://cran.r-project.org/web/packages/inlmisc/index.html inlmisc] ====
=== [[MySQL#Use_through_R|RMySQL]] ===
[https://owi.usgs.gov/blog/tolcolors/ GetTolColors()]. Lots of examples.
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.


=== [https://github.com/kevinushey/rex rex] ===
=== MongoDB ===
Friendly Regular Expressions
* http://www.r-bloggers.com/r-and-mongodb/
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/


=== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ===
=== odbc ===
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''


See also [http://stackoverflow.com/questions/3017877/tool-to-auto-format-r-code this discussion] on stackoverflow talks about R code reformatting.
=== RODBC ===


<pre>
=== DBI ===
library(formatR)
tidy_source("Input.R", file = "output.R", width.cutoff=70)
tidy_source("clipboard")
# default width is getOption("width") which is 127 in my case.
</pre>


Some issues
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
* To use databases with dplyr, you need to first install dbplyr
<pre>
* https://db.rstudio.com/dplyr/
cat("abcd",
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
    # This is my comment
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html
    "defg")
 
'''Create a new SQLite database''':
{{Pre}}
surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")
 
my_db_file <- "portal-database.sqlite"
my_db <- src_sqlite(my_db_file, create = TRUE)
 
copy_to(my_db, surveys)
copy_to(my_db, plots)
my_db
</pre>
</pre>
will result in
 
<pre>
'''Connect to a database''':
> tidy_source("clipboard")
{{Pre}}
Error in base::parse(text = code, srcfile = NULL) :
download.file(url = "https://ndownloader.figshare.com/files/2292171",
  3:1: unexpected string constant
              destfile = "portal_mammals.sqlite", mode = "wb")
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
 
3: "defg"
library(dbplyr)
  ^
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")
</pre>
</pre>
* Comments appearing at the end of a line within a long complete statement ''won't break'' tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.
 
<pre>
'''Querying the database with the SQL syntax''':
cat("abcd"
{{Pre}}
    ,"defg"  # This is my comment
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
  ,"ghij")
</pre>
</pre>
will become
 
<pre>
'''Querying the database with the dplyr syntax''':
cat("abcd", "defg"  # This is my comment
{{Pre}}
, "ghij")  
surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)
 
show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
</pre>
</pre>
Still bad!!
 
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
'''Simple database queries''':
<pre>
{{Pre}}
cat("</p>",
surveys %>%
"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
  filter(weight < 5) %>%
ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
  select(species_id, sex, weight)
                    "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"),
file=ExternalFileName, sep="\n", append=T)
</pre>
</pre>
will result in
 
<pre>
'''Laziness''' (instruct R to stop being lazy):
> tidy_source("clipboard", width.cutoff=70)
{{Pre}}
Error in base::parse(text = code, srcfile = NULL) :
data_subset <- surveys %>%
  3:129: unexpected SPECIAL
  filter(weight < 5) %>%
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
  select(species_id, sex, weight) %>%
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
  collect()
</pre>
* ''width.cutoff'' parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
<pre>
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on",
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName,
    sep = "\n", append = T)
</pre>
* It merges lines though I don't always want to do that. For example
<pre>
cat("abcd"
    ,"defg" 
  ,"ghij")
</pre>
will become
<pre>
cat("abcd", "defg", "ghij")  
</pre>
</pre>


=== Download papers ===
'''Complex database queries''':
==== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ====
{{Pre}}
Search and Download Papers from the bioRxiv Preprint Server
plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table
 
surveys # The plot_id column also features in the surveys table


==== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ====
# Join databases method 1
Interface to the arXiv API
plots %>%
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()
</pre>


==== [https://cran.r-project.org/web/packages/pdftools/index.html pdftools] ====
=== NoSQL ===
* http://ropensci.org/blog/2016/03/01/pdftools-and-jeroen
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
* http://r-posts.com/how-to-extract-data-from-a-pdf-file-with-r/
* https://ropensci.org/technotes/2018/12/14/pdftools-20/


=== [https://github.com/ColinFay/aside aside]: set it aside ===
== Github ==
An RStudio addin to run long R commands aside your current session.


=== Teaching ===
=== R source  ===
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics
https://github.com/wch/r-source/ Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.


=== packrat for reproducible search ===
If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
* [https://cran.r-project.org/web/packages/packrat/ CRAN] & [https://rstudio.github.io/packrat/ Github]
* Videos:
** https://www.rstudio.com/resources/webinars/managing-package-dependencies-in-r-with-packrat/
** https://www.rstudio.com/resources/webinars/rstudio-essentials-webinar-series-managing-part-3/
* Packrat will not only store all packages, but also all project files.
* Packrat is integrated in RStudio’s user interface. It allows you to share projects along co-workers easily. See [https://rstudio.github.io/packrat/rstudio.html Using Packrat with RStudio].
* [https://rstudio.github.io/packrat/limitations.html limitations].
* [https://stackoverflow.com/questions/36187543/using-r-with-git-and-packrat Git and packrat]. The ''packrat/src'' directory can be very large. ''If you don't want them available in your git-repo, you simply add packrat/src/ to the .gitignore. But, this will mean that anyone accessing the git-repo will not have access to the package source code, and the files will be downloaded from CRAN, or from wherever the source line dictates within the packrat.lock file.''
** [https://www.joelnitta.com/post/packrat/ Using packrat with git for (better) version control] Jun 2018


'''Create a snapshot''':
=== R packages (only) source (metacran) ===
* Do we really need to call packrat::snapshot()? The [https://rstudio.github.io/packrat/walkthrough.html walk through] page says it is not needed but the lock file is not updated from my testing.
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.
* I got an error when it is trying to fetch the source code from bioconductor and local repositories: packrat is trying to fetch the source from CRAN in these two packages.
** On normal case, the packrat/packrat.lock file contains two entries in 'Repos' field (line 4).
** The cause of the error is I ran snapshot() after I quitted R and entered again. So the solution is to add bioc and local repositories to options(repos).
** So what is important of running snapshot()?
** Check out the [https://groups.google.com/forum/#!forum/packrat-discuss forum].
<syntaxhighlight lang='rsplus'>
> dir.create("~/projects/babynames", recu=T)
> packrat::init("~/projects/babynames")
Initializing packrat project in directory:
- "~/projects/babynames"


Adding these packages to packrat:
=== Bioconductor packages source ===
            _
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>
    packrat  0.4.9-3


Fetching sources for packrat (0.4.9-3) ... OK (CRAN current)
=== Send local repository to Github in R by using reports package ===
Snapshot written to '/home/brb/projects/babynames/packrat/packrat.lock'
http://www.youtube.com/watch?v=WdOI_-aZV0Y
Installing packrat (0.4.9-3) ...
OK (built source)
Initialization complete!
Unloading packages in user library:
- packrat
Packrat mode on. Using library in directory:
- "~/projects/babynames/packrat/lib"


> install.packages("reshape2")
=== My collection ===
> packrat::snapshot()
* https://github.com/arraytools
 
* https://gist.github.com/4383351 heatmap using leukemia data
> system("tree -L 2 ~/projects/babynames/packrat/")
* https://gist.github.com/4382774 heatmap using sequential data
/home/brb/projects/babynames/packrat/
* https://gist.github.com/4484270 biocLite
├── init.R
├── lib
│   └── x86_64-pc-linux-gnu
├── lib-ext
│   └── x86_64-pc-linux-gnu
├── lib-R            # base packages
│   └── x86_64-pc-linux-gnu
├── packrat.lock
├── packrat.opts
└── src
    ├── bitops
    ├── glue
    ├── magrittr
    ├── packrat
    ├── plyr
    ├── Rcpp
    ├── reshape2
    ├── stringi
    └── stringr
</syntaxhighlight>


'''Restoring snapshots''':
=== How to download ===


Suppose a packrat project was created on Ubuntu 16.04 and we now want to repeat the analysis on Ubuntu 18.04. We first copy the whole project directory ('babynames') to Ubuntu 18.04. Then we should delete the library subdirectory ('packrat/lib') which contains binary files (*.so) that do not work on the new OS. After we delete the library subdirectory, start R from the project directory. Now if we run '''packrat::restore()''' command, it will re-install all missing libraries. Bingo! NOTE: Maybe I should use '''packrat::bundle()''' instead of manually copy the whole project folder.
Clone ~ Download.  
 
* Command line
Note: some OS level libraries (e.g. libXXX-dev) need to be installed manually beforehand in order for the magic to work.
<pre>
<syntaxhighlight lang='rsplus'>
git clone https://gist.github.com/4484270.git
$ rm -rf ~/projects/babynames/packrat/lib
</pre>
$ cd ~/projects/babynames/
This will create a subdirectory called '4484270' with all cloned files there.
$ R
>  
> packrat::status()
> remove.packages("plyr")
> packrat::status()
> packrat::restore()
</syntaxhighlight>


'''Set Up a Custom CRAN-like Repository''':
* Within R
<pre>
library(devtools)
source_gist("4484270")
</pre>
or
First download the json file from
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
library(RJSONIO)
x <- fromJSON("~/Downloads/gists.json")
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})
</pre>


See https://rstudio.github.io/packrat/custom-repos.html. Note the personal repository name ('sushi' in this example) used in "Repository" field of the personal package will be used in <packrat/packrat.lock> file. So as long as we work on the same computer, it is easy to restore a packrat project containing packages coming from personal repository.
=== Jekyll ===
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]


'''[https://rstudio.github.io/packrat/commands.html Common functions]''':
== Connect R with Arduino ==
* packrat::init()
* https://zhuhao.org/post/connect-arduino-chips-with-r/
* packrat::snapshot(), packrat::restore()
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
* packrat::clean()
* http://jean-robert.github.io/2012/11/11/thermometer-R-using-Arduino-Java.html
* packrat::status()
* http://bio7.org/?p=2049
* packrat::install_local() # http://rstudio.github.io/packrat/limitations.html
* http://www.rforge.net/Arduino/svn.html
* packrat::bundle() # see @28:44 of the [https://www.rstudio.com/resources/webinars/managing-package-dependencies-in-r-with-packrat/ video], packrat::unbundle() # see @29:17 of the same video. This will rebuild all packages
* packrat::on(), packrat::off()
* packrat::get_opts()
* packrat::set_opts() # http://rstudio.github.io/packrat/limitations.html
* packrat::opts$local.repos("~/local-cran")
* packrat::opts$external.packages(c("devtools")) # break the isolation
* packrat::extlib()
* packrat::with_extlib()
* packrat::project_dir(), .libPaths()


'''Warning'''
== Android App ==
* If we download and modify some function definition from a package in CRAN without changing DESCRIPTION file or the package name, the snapshot created using packrat::snapshot() will contain the package source from CRAN instead of local repository. This is because (I guess) the DESCRIPTION file contains a field 'Repository' with the value 'CRAN'.
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]


=== Organize R research project ===
== Common plots tips ==
[https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
=== Create an empty plot ===
'''plot.new()'''   


=== Text to speech ===
=== Overlay plots ===
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]
[https://finnstats.com/index.php/2021/08/15/how-to-overlay-plots-in-r/ How to Overlay Plots in R-Quick Guide with Example].
<pre>
#Step1:-create scatterplot
plot(x1, y1)
#Step 2:-overlay line plot
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)
</pre>


=== Weather data ===
=== Save the par() and restore it ===
* [https://github.com/ropensci/prism prism] package
'''Example 1''': Don't use old.par <- par() directly. no.readonly = FALSE by default. * The '''`no.readonly = TRUE`''' argument in the [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par par()] function in R is used to get the full list of graphical parameters '''that can be restored'''.
* [http://www.weatherbase.com/weather/weather.php3?s=507781&cityname=Rockville-Maryland-United-States-of-America Weatherbase]
* When you call `par()` with no arguments or `par(no.readonly = TRUE)`, it returns an invisible named list of all the graphical parameters. This includes both parameters that can be set and those that are read-only.
* If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
<pre>
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
old.par <- par(mar = c(5, 4, 4, 2) - 2)
## do plotting stuff with new settings
par(old.par)
</pre>
'''Example 2''': Use it inside a function with the [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/on.exit on.exit(0] function.
<pre>
ex <- function() {
  old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
  on.exit(par(old.par))
  ## ... do lots of par() settings and plots
  ## ...
  invisible() #-- now,  par(old.par)  will be executed
}
</pre>
'''Example 3''': It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.
<pre>
ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
</pre>
<pre>
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar
</pre>


=== logR ===
=== Grouped boxplots ===
https://github.com/jangorecki/logR
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
** 'at' parameter in boxplot() to change the equal spaced boxplots
** embed par(mar=) in boxplot()
** mtext(line=) to solve the problem the xlab overlapped with labels.
* [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others ggplot2 approach] (Hint: '''facet_grid''' is used)


=== Progress bar ===
=== [https://www.samruston.co.uk/ Weather Time Line] ===
https://github.com/r-lib/progress#readme
The plot looks similar to a boxplot though it is not. See a [https://www.samruston.co.uk/images/screens/screen_2.png screenshot] on Android by [https://www.samruston.co.uk/ Sam Ruston].


Configurable Progress bars, they may include percentage, elapsed time, and/or the estimated completion time. They work in terminals, in 'Emacs' 'ESS', 'RStudio', 'Windows' 'Rgui' and the 'macOS'.
=== Horizontal bar plot ===
{{Pre}}
library(ggplot2)
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
  coord_flip() + xlab("") + ylab("Fold Change") 
</pre>


=== cron ===
[[:File:Ggplot2bar.svg]]
[https://github.com/bnosac/cronr cronR]


== Different ways of using R ==
=== Include bar values in a barplot ===
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
* [http://stackoverflow.com/questions/12481430/how-to-display-the-frequency-at-the-top-of-each-factor-in-a-barplot-in-r barplot(), text() and axis()] functions. The data can be from a table() object.
* [https://stackoverflow.com/questions/11938293/how-to-label-a-barplot-bar-with-positive-and-negative-bars-with-ggplot2 How to label a barplot bar with positive and negative bars with ggplot2]


=== dyn.load ===
Use text().  
Error: [https://stackoverflow.com/questions/43662542/not-resolved-from-current-namespace-error-when-calling-c-routines-from-r “not resolved from current namespace” error, when calling C routines from R]


Solution: add '''getNativeSymbolInfo()''' around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace
Or use geom_text() if we are using the ggplot2 package. See an example [http://dsgeek.com/2014/09/19/Customizingggplot2charts.html here] or [https://rpubs.com/escott8908/RGC_Ch3_Gar_Graphs this].


=== R call C/C++ ===
For stacked barplot, see [http://t-redactyl.io/blog/2016/01/creating-plots-in-r-using-ggplot2-part-4-stacked-bar-plots.html this] post.
Mainly talks about .C() and .Call().


* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
=== Grouped barplots ===
* http://faculty.washington.edu/kenrice/sisg-adv/sisg-07.pdf
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)
* http://www.stat.berkeley.edu/scf/paciorek-cppWorkshop.pdf (Very useful)
{{Pre}}
* http://www.stat.harvard.edu/ccr2005/
library(ggplot2)
* http://mazamascience.com/WorkingWithData/?p=1099
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
      geom_bar(position="dodge", stat="identity")
</pre>
* https://datascienceplus.com/building-barplots-with-error-bars/. The error bars define 2 se (95% interval) for the black-and-white version and 1 se (68% interval) for ggplots. Be careful.
{{Pre}}
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042
</pre>
* [http://stackoverflow.com/questions/27466035/adding-values-to-barplot-of-table-in-r two bars in one factor] (stack). The data can be a 2-dim matrix with numerical values.
* [http://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r two bars in one factor], [https://stats.stackexchange.com/questions/14118/drawing-multiple-barplots-on-a-graph-in-r Drawing multiple barplots on a graph in R] (next to each other)
** [https://datascienceplus.com/building-barplots-with-error-bars/ Include error bars]
* [http://bl.ocks.org/patilv/raw/7360425/ Three variables] barplots
* [https://peltiertech.com/stacked-bar-chart-alternatives/ More alternatives] (not done by R)


=== SEXP ===
=== Unicode symbols ===
Some examples from packages
[https://www.r-bloggers.com/2024/09/mind-reader-game-and-unicode-symbols/ Mind reader game, and Unicode symbols]


* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
=== Math expression ===
 
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
=== R call Fortran ===
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
* Some cases
* [https://www.r-bloggers.com/the-need-for-speed-part-1-building-an-r-package-with-fortran-or-c/ The Need for Speed Part 1: Building an R Package with Fortran (or C)] (Very detailed)
** Use [https://www.rdocumentation.org/packages/base/versions/3.6.0/topics/expression expression()] function
** Don't need the backslash; use ''eta'' instead of ''\eta''. ''eta'' will be recognized as a special keyword in expression()
** Use parentheses instead of curly braces; use ''hat(eta)'' instead of ''hat{eta}''
** Summary: use expression(hat(eta)) instead of expression(\hat{\eta})
** [] means subscript, while ^ means superscript. See [https://statisticsglobe.com/add-subscript-and-superscript-to-plot-in-r Add Subscript and Superscript to Plot in R]
** Spacing can be done with ~.
** Mix math symbols and text using paste()
** Using substitute() and paste() if we need to substitute text (this part is advanced)
{{Pre}}
# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
    ylab = expression(phi^{rho + a}),
    main = "Pure Expressions")


=== Embedding R ===
# Superscript
plot(1:10, main = expression("My Title"^2))
# Subscript
plot(1:10, main = expression("My Title"[2])) 


* See [http://cran.r-project.org/doc/manuals/R-exts.html#Linking-GUIs-and-other-front_002dends-to-R Writing for R Extensions] Manual Chapter 8.
# Expressions with Spacing
* [http://www.ci.tuwien.ac.at/Conferences/useR-2004/abstracts/supplements/Urbanek.pdf Talk by Simon Urbanek] in UseR 2004.
# '~' is to add space and '*' is to squish characters together
* [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf Technical report]  by Friedrich Leisch in 2007.
plot(1:10, xlab= expression(Delta * 'C'))
* https://stat.ethz.ch/pipermail/r-help/attachments/20110729/b7d86ed7/attachment.pl
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
    ylab = expression(phi^{rho + a} * z * w),
    main = "Pure Expressions with Spacing")


==== An very simple example (do not return from shell) from Writing R Extensions manual ====
# Expressions with Text
The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.
plot(x,y,
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
    ylab = expression(paste("Here is some text of ", phi^{rho})),  
    main = "Expressions with Text")


This example can be run by
# Substituting Expressions
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>
plot(x,y,
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
    main = "Substituted Expressions")
</pre>


Note:
=== Impose a line to a scatter plot ===
# '''R_HOME/bin/exec/R''' is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
* abline + lsfit # least squares
# '''R_HOME/bin/R''' is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to ''/usr/local/bin/R''. When we run ''R_HOME/bin/R'', it actually runs ''R_HOME/bin/R CMD R_HOME/bin/exec/R'' (see line 259 of ''R_HOME/bin/R'' as in R 3.0.2) so we know the important role of ''R_HOME/bin/exec/R''.
{{Pre}}
plot(cars)
abline(lsfit(cars[, 1], cars[, 2]))
# OR
abline(lm(cars[,2] ~ cars[,1]))
</pre>
* abline + line # robust line fitting
{{Pre}}
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
</pre>
* lines
{{Pre}}
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')
</pre>


More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.
=== How to actually make a quality scatterplot in R: axis(), mtext() ===
[https://www.r-bloggers.com/2021/08/how-to-actually-make-a-quality-scatterplot-in-r/ How to actually make a quality scatterplot in R]


==== An example from Bioconductor workshop ====
=== 3D scatterplot ===
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* [http://sthda.com/english/wiki/scatterplot3d-3d-graphics-r-software-and-data-visualization Scatterplot3d: 3D graphics - R software and data visualization]. [https://stackoverflow.com/a/24510286 how to add legend to scatterplot3d in R] and consider '''xpd=TRUE'''.
* Use eval() function. See R-Ext [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Unix_002dalikes 8.1] and [http://cran.r-project.org/doc/manuals/R-exts.html#Embedding-R-under-Windows 8.2] and [http://cran.r-project.org/doc/manuals/R-exts.html#Evaluating-R-expressions-from-C 5.11].
* [[R_web#plotly|R web > plotly]]
* http://stackoverflow.com/questions/2463437/r-from-c-simplest-possible-helloworld (obtained from searching R_tryEval on google)
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c


Example:
=== Rotating x axis labels for barplot ===
Create <embed.c> file
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
<pre>
{{Pre}}
#include <Rembedded.h>
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
#include <Rdefines.h>
</pre>


static void doSplinesExample();
=== Set R plots x axis to show at y=0 ===
int
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
main(int argc, char *argv[])
{{Pre}}
{
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
    Rf_initEmbeddedR(argc, argv);
</pre>
    doSplinesExample();
    Rf_endEmbeddedR(0);
    return 0;
}
static void
doSplinesExample()
{
    SEXP e, result;
    int errorOccurred;


    // create and evaluate 'library(splines)'
=== Different colors of axis labels in barplot ===
    PROTECT(e = lang2(install("library"), mkString("splines")));
See [https://stackoverflow.com/questions/18839731/vary-colors-of-axis-labels-in-r-based-on-another-variable Vary colors of axis labels in R based on another variable]
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
    if (errorOccurred) {
        // handle error
    }
    UNPROTECT(1);


    // 'options(FALSE)' ...
Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
    PROTECT(e = lang2(install("options"), ScalarLogical(0)));
{{Pre}}
    // ... modified to 'options(example.ask=FALSE)' (this is obscure)
tN <- table(Ni <- stats::rpois(100, lambda = 5))
    SET_TAG(CDR(e), install("example.ask"));
r <- barplot(tN, col = rainbow(20))
    R_tryEval(e, R_GlobalEnv, NULL);
axis(1, 1, LETTERS[1], col.axis="red", col="red")
    UNPROTECT(1);
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
</pre>


    // 'example("ns")'
Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
    PROTECT(e = lang2(install("example"), mkString("ns")));
{{Pre}}
    R_tryEval(e, R_GlobalEnv, &errorOccurred);
barplot(tN, col = rainbow(20), axisnames = F)
    UNPROTECT(1);
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
}
</pre>
</pre>
Then build the executable. Note that I don't need to create R_HOME variable.
<pre>
cd
tar xzvf
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest


nano embed.c
=== Use text() to draw labels on X/Y-axis including rotation ===
# Using a single line will give an error and cannot not show the real problem.
* adj = 1 means top/right alignment. For left-bottom alignment, set adj = 0. The default is to center the text. [[https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text ?text]
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
* [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/par par("usr")] gives the extremes of the user coordinates of the plotting region of the form c(x1, x2, y1, y2).
# A better way is to run compile and link separately
** par("usr") is determined *after* a plot has been created
gcc -I../../include -c embed.c
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
gcc -o embed embed.o -L../../lib -lR -lRblas
* https://datascienceplus.com/building-barplots-with-error-bars/
../../bin/R CMD ./embed
{{Pre}}
par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
    adj = 1, labels = myData$names, xpd = TRUE)
</pre>
</pre>
* https://www.r-bloggers.com/rotated-axis-labels-in-r-plots/
=== Vertically stacked plots with the same x axis ===
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in


Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying '''R_HOME''' variable and including the directories used in linking R in '''LD_LIBRARY_PATH'''. This is based on the inform provided by [http://cran.r-project.org/doc/manuals/r-devel/R-exts.html Writing R Extensions].
=== Include labels on the top axis/margin: axis() and mtext() ===
<pre>
<pre>
export R_HOME=/home/brb/Downloads/R-3.0.2
plot(1:4, rnorm(4), axes = FALSE)
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
axis(3, at=1:4, labels = LETTERS[1:4], tick = FALSE, line = -0.5) # las, cex.axis
./embed # No need to include R CMD in front.
box()
mtext("Groups selected", cex = 0.8, line = 1.5) # default side = 3
</pre>
</pre>
See also [[#15_Questions_All_R_Users_Have_About_Plots| 15_Questions_All_R_Users_Have_About_Plots]]


Question: Create a data frame in C? Answer: [https://stat.ethz.ch/pipermail/r-devel/2013-August/067107.html Use data.frame() via an eval() call from C]. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as [https://stat.ethz.ch/pipermail/r-devel/2013-August/067109.html here].
This can be used to annotate each plot with the script name, date, ...
<pre>
mtext(text=paste("Prepared on", format(Sys.time(), "%d %B %Y at %H:%M")),
      adj=.99,  # text align to right
      cex=.75, side=3, las=1, line=2)
</pre>
 
ggplot2 uses '''breaks''' instead of '''at''' parameter. See [[Ggplot2#Add_axis_on_top_or_right_hand_side|ggplot2 &rarr; Add axis on top or right hand side]], [[Ggplot2#ggplot2::scale_-_axes.2Faxis.2C_legend|ggplot2 &rarr; scale_x_continus(name, breaks, labels)]] and the [https://ggplot2.tidyverse.org/reference/scale_continuous.html scale_continuous documentation].


Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
=== Legend tips ===
[https://r-coder.com/add-legend-r/ Add legend to a plot in R]


==== Create a Simple Socket Server in R ====
[https://stackoverflow.com/a/36842578 Increase/decrease legend font size] '''cex''' & [[Ggplot2#Legend_size|ggplot2]] package case.
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].  
{{Pre}}
plot(rnorm(100))
# op <- par(cex=2)
legend("topleft", legend = 1:4, col=1:4, pch=1, lwd=2, lty = 1, cex =2)
# par(op)
</pre>


Create an R function
'''legend inset'''. Default is 0. % (from 0 to 1) to draw the legend away from x and y axis. The inset argument with [https://stackoverflow.com/a/10528078 negative values moves the legend outside the plot].
<pre>
<pre>
simpleServer <- function(port=6543)
legend("bottomright", inset=.05, )
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}
</pre>
</pre>
Then run simpleServer(). Open another terminal and try to communicate with the server
 
'''legend without a box'''
<pre>
<pre>
$ telnet localhost 6543
legend(, bty = "n")
Trying 127.0.0.1...
</pre>
Connected to localhost.
Escape character is '^]'.


Welcome to R!
'''Add a legend title'''
R> summary(iris[, 3:5])
<pre>
  Petal.Length    Petal.Width          Species 
legend(, title = "")
Min.  :1.000  Min.  :0.100  setosa    :50 
1st Qu.:1.600  1st Qu.:0.300  versicolor:50 
Median :4.350  Median :1.300  virginica :50 
Mean  :3.758  Mean  :1.199                 
3rd Qu.:5.100  3rd Qu.:1.800                 
Max.  :6.900  Max.  :2.500                 
 
R> quit
Connection closed by foreign host.
</pre>
</pre>


==== [http://www.rforge.net/Rserve/doc.html Rserve] ====
[https://stackoverflow.com/a/60971923 Add a common legend to multiple plots]. Use the layout function.
Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See [[R#An_example_from_Bioconductor_workshop|Example from Bioconductor workshop]].


See my [[Rserve]] page.
=== Superimpose a density plot or any curves ===
Use '''lines()'''.  


==== (Commercial) [http://www.statconn.com/ StatconnDcom] ====
Example 1
{{Pre}}
plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))


==== [http://rdotnet.codeplex.com/ R.NET] ====
plot(density(x), col = "#6F69AC", lwd = 3)
 
lines(density(y), col = "#95DAC1", lwd = 3)
==== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ====
lines(density(z), col = "#FFEBA1", lwd = 3)
* [https://jozefhajnala.gitlab.io/r/r901-primer-java-from-r-1/ A primer in using Java from R - part 1]
* Note rJava is needed by [https://cran.r-project.org/web/packages/xlsx/index.html xlsx] package.
 
Terminal
<syntaxhighlight lang='bash'>
# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk
</syntaxhighlight>
and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.
* Create the file '''/etc/ld.so.conf.d/java.conf''' with the following entries:
<pre>
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
</pre>
</pre>
* And then run '''sudo ldconfig'''


Now go back to R
Example 2
<syntaxhighlight lang='rsplus'>
{{Pre}}
install.packages("rJava")
require(survival)
</syntaxhighlight>
n = 10000
Done!
beta1 = 2; beta2 = -1
 
lambdaT = 1 # baseline hazard
If above does not work, a simple way is by (under Ubuntu) running
lambdaC = 2  # hazard of censoring
<pre>
set.seed(1234)
sudo apt-get install r-cran-rjava
x1 = rnorm(n,0)
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2))
C <- rweibull(n, shape=1, scale=lambdaC) 
time = pmin(T,C) 
status <- 1*(T <= C)  
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1),
    ylab="Survival probability",
    main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull
</pre>
</pre>
which will create new package 'default-jre' (under '''/usr/lib/jvm''') and 'default-jre-headless'.


==== RCaller ====
Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".


==== RApache ====
=== log scale ===
* http://www.stat.ucla.edu/~jeroen/files/seminar.pdf
If we set y-axis to use log-scale, then what we display is the value log(Y) or log10(Y) though we still label the values using the input. For example, when we plot c(1, 10, 100) using the log scale, it is like we draw log10(c(1, 10, 100)) = c(0,1,2) on the plot but label the axis using the true values c(1, 10, 100).


==== [http://dirk.eddelbuettel.com/code/littler.html littler] ====
[[:File:Logscale.png]]
Provides hash-bang (#!) capability for R


FAQs:
=== Custom scales ===
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
* [https://stackoverflow.com/questions/3412911/r-exe-rcmd-exe-rscript-exe-and-rterm-exe-whats-the-difference Whats the difference between Rscript and R CMD BATCH]
* [https://stackoverflow.com/questions/21969145/why-or-when-is-rscript-or-littler-better-than-r-cmd-batch Why (or when) is Rscript (or littler) better than R CMD BATCH?]
<syntaxhighlight lang='bash'>
root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"


-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R       # text, R --help
== Time series ==
                                              # Example: R -q -e "date()"
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]


-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
=== Time series stock price plot ===
                                              # It won't show the startup message when it is used in the command line.
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
                                              # Example: Rscript -e "date()"
* [https://datascienceplus.com/visualize-your-portfolios-performance-and-generate-a-nice-report-with-r/ Visualize your Portfolio’s Performance and Generate a Nice Report with R]
</syntaxhighlight>
* https://timelyportfolio.github.io/rCharts_time_series/history.html


We can install littler using two ways.
{{Pre}}
* install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ''~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r''). You need to create a soft link in order to access it globally.
library(quantmod)
* sudo apt install littler. This will install 'r' globally; however, the installed version may be old.
getSymbols("AAPL")
getSymbols("IBM") # similar to AAPL
getSymbols("CSCO") # much smaller than AAPL, IBM
getSymbols("DJI") # Dow Jones, huge
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)",
    col='orange', subset = '2017::2017-08')


After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The [http://dirk.eddelbuettel.com/code/littler.examples.html web version of examples] does not have the TOC.
tail(Cl(DJI))
</pre>


'''r''' was not meant to run interactively like '''R'''. See ''man r''.
=== tidyquant: Getting stock data ===
[http://varianceexplained.org/r/stock-changes/ The 'largest stock profit or loss' puzzle: efficient computation in R]


==== RInside: Embed R in C++ ====
=== Timeline plot ===
See [[R#RInside|RInside]]
* https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
* [https://github.com/shosaco/vistime vistime] - Pretty Timelines in R


(''From RInside documentation'') The RInside package makes it easier to embed R in your C++ applications. There is no code you would execute directly from the R environment. Rather, you write C++ programs that embed R which is illustrated by some the included examples.
=== Clockify ===
[https://datawookie.dev/blog/2021/09/clockify-time-tracking-from-r/ Clockify]


The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.
== Circular plot ==
* http://freakonometrics.hypotheses.org/20667 which uses [https://cran.r-project.org/web/packages/circlize/ circlize] package; see also the '''ComplexHeatmap''' package.
* https://www.biostars.org/p/17728/
* [https://cran.r-project.org/web/packages/RCircos/ RCircos] package from CRAN.
* [http://www.bioconductor.org/packages/release/bioc/html/OmicCircos.html OmicCircos] from Bioconductor.


To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.
== Word cloud ==
* [http://www.sthda.com/english/wiki/text-mining-and-word-cloud-fundamentals-in-r-5-simple-steps-you-should-know Text mining and word cloud fundamentals in R : 5 simple steps you should know]
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
* [https://github.com/lepennec/ggwordcloud?s=09 ggwordcloud]


To run any executable program, we need to specify '''LD_LIBRARY_PATH''' variable, something like
== Text mining ==
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>
* [https://cran.r-project.org/web/packages/tm/index.html tm] package. It was used by [https://github.com/jtleek/swfdr/blob/master/getPvalues.R R code] of [https://doi.org/10.1093/biostatistics/kxt007 An estimate of the science-wise false discovery rate and application to the top medical literature].


The real build process looks like (check <Makefile> for completeness)
== World map ==
<pre>
[https://www.enchufa2.es/archives/visualising-ssh-attacks-with-r.html Visualising SSH attacks with R] ([https://cran.r-project.org/package=rworldmap rworldmap] and [https://cran.r-project.org/package=rgeolocate rgeolocate] packages)
g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include  \
    rinside_sample0.cpp  \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0
</pre>


Hello World example of embedding R in C++.
== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
<pre>
* [https://finnstats.com/index.php/2021/06/29/transition-plot-in-r-change-in-time-visualization/ Transition plot in R-change in time visualization]
#include <RInside.h>                    // for the embedded R via RInside


int main(int argc, char *argv[]) {
=== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ===
* [https://blog.rstudio.com/2015/05/01/rstudio-v0-99-preview-graphviz-and-diagrammer/ Graphviz and DiagrammeR]
* http://rich-iannone.github.io/DiagrammeR/,  
** [http://rich-iannone.github.io/DiagrammeR/io.html#r-markdown rmarkdown]
** [http://rich-iannone.github.io/DiagrammeR/graphviz_and_mermaid.html graphviz and mermaid] doc and examples
* https://donlelek.github.io/2015-03-31-dags-with-r/
* [https://mikeyharper.uk/flowcharts-in-r-using-diagrammer/ Data-driven flowcharts in R using DiagrammeR]


    RInside R(argc, argv);              // create an embedded R instance
=== [https://cran.r-project.org/web/packages/diagram/ diagram] ===
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
 
=== DAGitty (browser-based and R package) ===
* http://dagitty.net/
* https://cran.r-project.org/web/packages/dagitty/index.html
 
=== dagR ===
* https://cran.r-project.org/web/packages/dagR


    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
=== Gmisc ===
[http://gforge.se/2020/08/easy-flowchart/ Easiest flowcharts eveR?]


    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
=== Concept Maps ===
[https://github.com/rstudio/concept-maps/ concept-maps] where the diagrams are generated from https://app.diagrams.net/.


    exit(0);
=== flow ===
}
[https://cran.r-project.org/web/packages/flow/ flow], [https://predictivehacks.com/?all-tips=how-to-draw-flow-diagrams-in-r How To Draw Flow Diagrams In R]
</pre>


The above can be compared to the Hello world example in Qt.
== Venn Diagram ==
<pre>
[[Venn_diagram|Venn diagram]]
#include <QApplication.h>
#include <QPushButton.h>


int main( int argc, char **argv )
== hexbin plot ==
{
* [https://datasciencetut.com/how-to-create-a-hexbin-chart-in-r/ How to create a hexbin chart in R]
    QApplication app( argc, argv );
* [https://cran.r-project.org/web/packages/hextri/index.html hextri]: Hexbin Plots with Triangles. See an example on this https://www.pnas.org/content/117/48/30266#F4 paper] about the postpi method.


    QPushButton hello( "Hello world!", 0 );
== Bump chart/Metro map ==
    hello.resize( 100, 30 );
https://dominikkoch.github.io/Bump-Chart/


    app.setMainWidget( &hello );
== Amazing/special plots ==
    hello.show();
See [[Amazing_plot|Amazing plot]].


    return app.exec();
== Google Analytics ==
}
=== GAR package ===
</pre>
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html


==== [http://www.rfortran.org/ RFortran] ====
== Linear Programming ==
RFortran is an open source project with the following aim:
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/


''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
== Linear Algebra ==
* [https://jimskinner.github.io/post/elegant-linear-algebra-in-r-with-the-matrix-package/ Elegant linear algebra in R with the Matrix package]. Matrix package is used.
* [https://datascienceplus.com/linear-algebra-for-machine-learning-and-deep-learning-in-r/ Linear Algebra for Machine Learning and Deep Learning in R]. MASS library is used.


It works only on Windows platform with Microsoft Visual Studio installed:(
== Amazon Alexa ==
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html


=== Call R from other languages ===
== R and Singularity ==
==== C ====
https://rviews.rstudio.com/2017/03/29/r-and-singularity/
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]


==== JRI ====
== Teach kids about R with Minecraft ==
http://www.rforge.net/JRI/
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html


==== ryp2 ====
== Secure API keys ==
http://rpy.sourceforge.net/rpy2.html
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]


=== Create a standalone Rmath library ===
== Credentials and secrets ==
R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of [http://cran.r-project.org/doc/manuals/R-admin.html#The-standalone-Rmath-library R-admin manual].
[https://datascienceplus.com/how-to-manage-credentials-and-secrets-safely-in-r/ How to manage credentials and secrets safely in R]


Here is my experience based on R 3.0.2 on Windows OS.
== Hide a password ==
=== keyring package ===
* https://cran.r-project.org/web/packages/keyring/index.html
* [http://theautomatic.net/2019/06/25/how-to-hide-a-password-in-r-with-the-keyring-package/ How to hide a password in R with the Keyring package]


==== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ====
=== getPass ===
Suppose we have downloaded R source code and build R from its source. See [[R#Build_R_from_its_source|Build_R_from_its_source]]. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.
[https://cran.r-project.org/web/packages/getPass/README.html getPass]
<pre>
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>


==== Use Rmath library in our code ====
== Vision and image recognition ==
<pre>
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
* [http://www.bnosac.be/index.php/blog/66-computer-vision-algorithms-for-r-users Computer Vision Algorithms for R users] and https://github.com/bnosac/image
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.


# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
== Creating a Dataset from an Image ==
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
[https://ivelasq.rbind.io/blog/reticulate-data-recreation/ Creating a Dataset from an Image in R Markdown using reticulate]
# It is OK to save the cpp file under any directory.


# Force to link against the static library <libRmath.a>
== Turn pictures into coloring pages ==
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe


# Force to link against dynamic library <Rmath.dll>
== Numerical optimization ==
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
[https://cran.r-project.org/web/views/NumericalMathematics.html CRAN Task View: Numerical Mathematics], [https://cran.r-project.org/web/views/Optimization.html CRAN Task View: Optimization and Mathematical Programming]
</pre>
Test the executable program. Note that the executable program ''RmathEx1.exe'' can be transferred to and run in another computer without R installed. Isn't it cool!
<pre>
c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689
</pre>


Below is the cpp program <RmathEx1.cpp>.
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/uniroot.html uniroot]: One Dimensional Root (Zero) Finding. This is used in [http://onlinelibrary.wiley.com/doi/10.1002/sim.7178/full simulating survival data for predefined censoring rate]
<pre>
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
//RmathEx1.cpp
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optim.html optim]: General-purpose optimization based on Nelder–Mead, quasi-Newton and conjugate-gradient algorithms.  
#define MATHLIB_STANDALONE
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
#include <iostream>
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
#include "Rmath.h"
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares
* [https://blogs.rstudio.com/ai/posts/2021-04-22-torch-for-optimization/ torch for optimization]. L-BFGS optimizer.


using std::cout; using std::cin; using std::endl;
== Ryacas: R Interface to the 'Yacas' Computer Algebra System ==
[https://blog.ephorie.de/doing-maths-symbolically-r-as-a-computer-algebra-system-cas Doing Maths Symbolically: R as a Computer Algebra System (CAS)]
 
== Game ==
* [https://kbroman.org/miner_book/?s=09 R Programming with Minecraft]
* [https://cran.r-project.org/web/packages/pixelpuzzle/index.html pixelpuzzle]
* [https://www.rostrum.blog/2022/09/24/pixeltrix/ Interactive pixel art in R with {pixeltrix}]
* [https://rtaoist.blogspot.com/2021/03/r-shiny-maths-games-for-6-years-old.html Shiny math game]
* [https://cran.microsoft.com/web/packages/mazing/index.html mazing]: Utilities for Making and Plotting Mazes
* [https://github.com/jeroenjanssens/raylibr/blob/main/demo/snake.R snake] which is based on [https://github.com/jeroenjanssens/raylibr raylibr]
 
== Music ==
* [https://flujoo.github.io/gm/ gm]. Require to install [https://musescore.org/en MuseScore], an open source and free notation software.
 
== SAS ==
[https://github.com/MangoTheCat/sasMap sasMap] Static code analysis for SAS scripts


int main()
= R packages =
{
[[R_packages|R packages]]
  double x1, x2;
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;


  cout << "Prob(Z <= " << x1 << ") = " <<
= Tricks =
    pnorm(x1, 0, 1, 1, 0)  << endl;
  cout << "Prob(Chi^2 <= " << x2 << ")= " <<
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}
</pre>


=== Calling R.dll directly ===
== Getting help ==
See Chapter 8.2.2 of [http://cran.r-project.org/doc/manuals/R-exts.html#Calling-R_002edll-directly|Writing R Extensions]. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.  
* https://stat.ethz.ch/pipermail/r-help/
* https://stat.ethz.ch/pipermail/r-devel/


=== Create HTML report ===
== Better Coder/coding, best practices ==
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor.
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
* [https://stackoverflow.com/a/2258292 What best practices do you use for programming in R?]
* [https://onlinelibrary.wiley.com/doi/abs/10.1002/sim.9169?campaign=woletoc Best practices in statistical computing] Sanchez 2021


==== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ====
== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23


* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
== Getting user's home directory ==
* http://gforge.se/2014/01/fast-track-publishing-using-knitr-part-iv/
See [https://cran.r-project.org/bin/windows/base/rw-FAQ.html#What-are-HOME-and-working-directories_003f What are HOME and working directories?]
{{Pre}}
# Windows
normalizePath("~")  # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")  # "C:/Users/brb/Documents"


==== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ====
# Mac
* https://github.com/renkun-ken/formattable
normalizePath("~")  # [1] "/Users/brb"
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
Sys.getenv("R_USER") # [1] ""
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]
Sys.getenv("HOME")  # "/Users/brb"


==== [https://github.com/crubba/htmltab htmltab] package ====
# Linux
This package is NOT used to CREATE html report but EXTRACT html table.
normalizePath("~")  # [1] "/home/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")  # [1] "/home/brb"
</pre>


==== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ====
== tempdir() ==
Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.
* The path is a per-session temporary directory. On parallel use, R processes forked by functions such as '''mclapply''' and '''makeForkCluster''' in package '''parallel''' share a per-session temporary directory.
* [https://www.r-bloggers.com/2024/07/r-set-temporary-folder-for-r-in-rstudio-server/ Set temporary folder for R in Rstudio server]


=== Create academic report ===
== Distinguish Windows and Linux/Mac, R.Version() ==
[http://cran.r-project.org/web/packages/reports/index.html reports] package in CRAN and in [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.
identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.


=== Create pdf and epub files ===
* [https://www.r-bloggers.com/identifying-the-os-from-r/ Identifying the OS from R]
<syntaxhighlight lang='rsplus'>
* [https://stackoverflow.com/questions/4747715/how-to-check-the-os-within-r How to check the OS within R]
# Idea:
<pre>
#        knitr        pdflatex
get_os <- function(){
#  rnw -------> tex ----------> pdf
  sysinf <- Sys.info()
library(knitr)
  if (!is.null(sysinf)){
knit("example.rnw") # create example.tex file
    os <- sysinf['sysname']
</syntaxhighlight>
    if (os == 'Darwin')
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
      os <- "osx"
<syntaxhighlight lang='bash'>
  } else { ## mystery machine
git clone https://github.com/yihui/knitr-examples.git
    os <- .Platform$OS.type
</syntaxhighlight>
    if (grepl("^darwin", R.version$os))
* <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run '''sudo apt-get install texlive-fonts-recommended''' to install missing fonts. It works!
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
  tolower(os)
}
</pre>
<pre>
names(R.Version())
#  [1] "platform"      "arch"          "os"            "system"       
#  [5] "status"        "major"          "minor"          "year"         
#  [9] "month"          "day"            "svn rev"        "language"     
# [13] "version.string" "nickname"
getRversion()
# [1] ‘4.3.0’
</pre>


To see a real example, check out [http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html DESeq2] package (inst/doc subdirectory). In addition to DESeq2, I also need to install '''DESeq, BiocStyle, airway, vsn, gplots''', and '''pasilla''' packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.
== Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only) ==
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site'''). Put R statements.
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site'''). Define environment variables.
* https://cran.r-project.org/doc/manuals/r-release/R-intro.html ('''Rprofile.site, Renviron.site, Rconsole''' (Windows only))
* [http://blog.revolutionanalytics.com/2015/11/how-to-store-and-use-authentication-details-with-r.html How to store and use webservice keys and authentication details]
* [http://itsalocke.com/use-rprofile-give-important-notifications/ Use your .Rprofile to give you important notifications]
* [https://rviews.rstudio.com/2017/04/19/r-for-enterprise-understanding-r-s-startup/ *R for Enterprise: Understanding R’s Startup]
* [https://support.rstudio.com/hc/en-us/articles/360047157094-Managing-R-with-Rprofile-Renviron-Rprofile-site-Renviron-site-rsession-conf-and-repos-conf *Managing R with .Rprofile, .Renviron, Rprofile.site, Renviron.site, rsession.conf, and repos.conf]


Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
If we like to install R packages to a personal directory, follow [https://stat.ethz.ch/pipermail/r-devel/2015-July/071562.html this]. Just add the line
<syntaxhighlight lang='bash'>
# Idea:
#        knitr        pandoc
#  rmd -------> md ----------> pdf
 
git clone https://github.com/yihui/knitr-examples.git
cd knitr-examples
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!
</syntaxhighlight>
 
To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
<syntaxhighlight lang='rsplus'>
# Idea:
#        knitr        pandoc
#  rnw -------> tex ----------> markdown or epub
 
library(knitr)
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
</syntaxhighlight>
<pre>
<pre>
## Windows OS, epub cannot be built
R_LIBS_SITE=F:/R/library
pandoc:
</pre>
Error:
to the file '''R_HOME/etc/x64/Renviron.site'''. In R, run '''Sys.getenv("R_LIBS_SITE")''' or '''Sys.getenv("R_LIBS_USER")''' to query the environment variable. See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/EnvVar.html Environment Variables].
"source" (line 41, column 7):
unexpected "k"
expecting "{document}"


## Linux OS, epub missing figures and R codes.
=== What is the best place to save Rconsole on Windows platform ===
## First install texlive base and extra packages
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
## sudo apt-get install texlive-latex-base texlive-latex-extra
pandoc: Could not find media `figure/SchwederSpjotvoll-1', skipping...
pandoc: Could not find media `figure/sortedP-1', skipping...
pandoc: Could not find media `figure/figHeatmap2c-1', skipping...
pandoc: Could not find media `figure/figHeatmap2b-1', skipping...
pandoc: Could not find media `figure/figHeatmap2a-1', skipping...
pandoc: Could not find media `figure/plotCountsAdv-1', skipping...
pandoc: Could not find media `figure/plotCounts-1', skipping...
pandoc: Could not find media `figure/MA-1', skipping...
pandoc: Could not find media `figure/MANoPrior-1', skipping...
</pre>
The problems are at least
* figures need to be generated under the same directory as the source code
* figures cannot be in the format of pdf (DESeq2 generates both pdf and png files format)
* missing R codes


Convert tex to epub
My preferred settings:
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
* Size: 12
* background: black
* normaltext: white
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow


==== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ====
and others (default options)
Create Tables In LaTeX, HTML, Markdown And ReStructuredText
* pagebg: white
* pagetext: navy
* highlight: DarkRed
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


* https://rmarkdown.rstudio.com/lesson-7.html
A copy of the Rconsole is saved in [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb github].
* https://stackoverflow.com/questions/20942466/creating-good-kable-output-in-rstudio
* http://kbroman.org/knitr_knutshell/pages/figs_tables.html
* https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/
* [https://cran.r-project.org/web/packages/kableExtra/vignettes/awesome_table_in_html.html kableExtra] package


=== Create Word report ===
=== How R starts up ===
https://rstats.wtf/r-startup.html


==== knitr + pandoc ====
=== startup - Friendly R Startup Configuration ===
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
https://github.com/henrikbengtsson/startup
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* http://rmarkdown.rstudio.com/articles_docx.html


It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.
== Saving and loading history automatically: .Rprofile & local() ==
<ul>
<li>[http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html savehistory("filename")]. It will save everything from the beginning to the command savehistory() to a text file.
<li>'''.Rprofile''' will automatically be loaded when R has started from that directory
<li>Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See [https://www.tidyverse.org/articles/2017/12/workflow-vs-script/ Project-oriented workflow].
<li>'''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file and [[R_packages|R packages &rarr; packrat]].
<li>[http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]
* You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.
* At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
<pre>
# Idea:
options(continue=" ") # default is "+ "
#        knitr      pandoc
options(prompt="R> ", continue=" ")
#  rmd -------> md --------> docx
options(editor="nano") # default is "vi" on Linux
library(knitr)
# options(htmlhelp=TRUE)  
knit2html("example.rmd") #Create md and html files
</pre>
and then
<pre>
FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))
</pre>
Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.


Another way is
local({r <- getOption("repos")
<pre>
      r["CRAN"] <- "https://cran.rstudio.com"
library(pander)
      options(repos=r)})
name = "demo"
 
knit(paste0(name, ".Rmd"), encoding = "utf-8")
.First <- function(){
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")
# library(tidyverse)
</pre>
cat("\nWelcome at", date(), "\n")
}


Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:
.Last <- function(){
* A pdf file: pandoc -s report.md -t latex -o report.pdf
cat("\nGoodbye at ", date(), "\n")
* A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
* Openoffice: pandoc report.md -o report.odt
</pre>
* Word docx: pandoc report.md -o report.docx
<li>https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
<li>The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
<li>[https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)
</ul>
'''Linux''' or '''Mac'''


We can also create the epub file for reading on Kobo ereader. For example, download [https://gist.github.com/jeromyanglim/2716336 this file] and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!
In '''~/.profile''' or '''~/.bashrc''' I put:
<pre>
<pre>
knit("example.Rmd")
export R_HISTFILE=~/.Rhistory
pandoc("example.md", format="epub")
</pre>
</pre>
 
In '''~/.Rprofile''' I put:
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
<pre>
<pre>
> pandoc("Rmd_to_Epub.md", format="epub")
if (interactive()) {
executing pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
   if (.Platform$OS.type == "unix".First <- function() try(utils::loadhistory("~/.Rhistory"))
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
Error in (function (input, format, ext, cfg)  : conversion failed
}
In addition: Warning message:
running command 'pandoc  -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1
</pre>
</pre>


==== pander ====
'''Windows'''
Try pandoc[1] with a minimal reproducible example, you might give a try to my "[http://cran.r-project.org/web/packages/pander/ pander]" package [2] too:


If you launch R by clicking its icon from Windows Desktop, the R starts in '''C:\User\$USER\Documents''' directory. So we can create a new file '''.Rprofile''' in this directory.
<pre>
<pre>
library(pander)
if (interactive()) {
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
            output = tempfile(), convert = 'docx')
}
</pre>
</pre>
Where the content of the "minimal.brew" file is something you might have
got used to with Sweave - although it's using "brew" syntax instead. See
the examples of pander [3] for more details. Please note that pandoc should
be installed first, which is pretty easy on Windows.


# http://johnmacfarlane.net/pandoc/
== Disable "Save workspace image?" prompt when exit R? ==
# http://rapporter.github.com/pander/
[https://stackoverflow.com/a/4996252 How to disable "Save workspace image?" prompt in R?]
# http://rapporter.github.com/pander/#examples
 
== R release versions ==
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.


==== R2wd ====
== getRversion() ==
Use [http://cran.r-project.org/web/packages/R2wd/ R2wd] package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.
<pre>
<pre>
> library(R2wd)
getRversion()
> wdGet()
[1] ‘4.3.0’
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
    installstatconnDCOM()
 
This will download and install the current version of statconnDCOM
 
You will need a working Internet connection
because installation needs to download a file.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() :
  argument is of length zero
</pre>
</pre>


The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
== Detect number of running R instances in Windows ==
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
<pre>
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
INFO: No tasks are running which match the specified criteria.


==== Convert from pdf to word ====
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert


==== rtf ====
Image Name                    PID Session Name        Session#    Mem Usage
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.
============================================================================
Rgui.exe                      1096 Console                    1    44,712 K


==== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ====
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
Package xtable will produce html output. <syntaxhighlight lang='rsplus'>print(xtable(X), type="html")</syntaxhighlight>


If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.
Image Name                    PID Session Name        Session#    Mem Usage
============================================================================
Rserve.exe                    6108 Console                    1    381,796 K
</pre>
In R, we can use
<pre>
> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
[1] ""                                                                           
[2] "Image Name                    PID Session Name        Session#    Mem Usage"
[3] "============================================================================"
[4] "Rgui.exe                      1096 Console                    1    44,804 K"


==== [http://cran.r-project.org/web/packages/ReporteRs/index.html ReporteRs] ====
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
Microsoft Word, Microsoft Powerpoint and HTML documents generation from R. The source code is hosted on https://github.com/davidgohel/ReporteRs
</pre>


[https://statbandit.wordpress.com/2016/10/28/a-quick-exploration-of-reporters/ A quick exploration]
== Editor ==
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs


=== R Graphs Gallery ===
<ul>
* [https://www.facebook.com/pages/R-Graph-Gallery/169231589826661 Romain François]
<li>Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
* [http://shinyapps.stat.ubc.ca/r-graph-catalog/ R Graph Catalog] written using R + Shiny. The source code is available on [https://github.com/jennybc/r-graph-catalog Github].
* Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
* Forest plot. See the packages [https://cran.r-project.org/web/packages/rmeta/index.html rmeta] and [https://cran.r-project.org/web/packages/forestplot/ forestplot]. The forest plot can be used to plot the quantities like relative risk (with 95% CI) in survival data.
<pre>
(setq-default inferior-R-program-name
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
</pre>
* [https://blog.rwhitedwarf.com/post/use_emacs_for_r/ Using Emacs for R] 2022
</ul>
* [http://www.rstudio.com/ Rstudio] - editor/R terminal/R graphics/file browser/package manager. The new version (0.98) also provides a new feature for debugging step-by-step. See also [https://www.rstudio.com/rviews/2016/11/11/easy-tricks-you-mightve-missed/ RStudio Tricks]
* [http://www.geany.org/ geany] - I like the feature that it shows defined functions on the side panel even for R code. RStudio can also do this (see the bottom of the code panel).
* [http://rgedit.sourceforge.net/ Rgedit] which includes a feature of splitting screen into two panes and run R in the bottom panel. See [http://www.stattler.com/article/using-gedit-or-rgedit-r here].
* Komodo IDE with browser preview http://www.youtube.com/watch?v=wv89OOw9roI at 4:06 and http://docs.activestate.com/komodo/4.4/editor.html


=== COM client or server ===
== GUI for Data Analysis ==
[https://www.r-bloggers.com/2023/06/update-to-data-science-software-popularity/ Update to Data Science Software Popularity] 6/7/2023


==== Client ====
=== BlueSky Statistics ===
* https://www.blueskystatistics.com/Default.asp
* [https://r4stats.com/articles/software-reviews/bluesky/ A Comparative Review of the BlueSky Statistics GUI for R]


[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
=== Rcmdr ===
http://cran.r-project.org/web/packages/Rcmdr/index.html. After loading a dataset, click Statistics -> Fit models. Then select Linear regression, Linear model, GLM, Multinomial logit model, Ordinal regression model, Linear mixed model, and Generalized linear mixed model. However, Rcmdr does not include, e.g. random forest, SVM, glmnet, et al.


==== Server ====
=== Deducer ===
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]
http://cran.r-project.org/web/packages/Deducer/index.html


=== Use R under proxy ===
=== jamovi ===
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* https://www.jamovi.org/
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]


=== RStudio ===
== Scope ==
* [https://github.com/rstudio/rstudio Github]
See
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]


==== rstudio.cloud ====
=== source() ===
https://rstudio.cloud/
* [https://twitter.com/henrikbengtsson/status/1563849697084809219?s=20&t=nStcqVabAQ_HvJ2FaBloNQ source() assigns to the global environment, not the calling environment, which might not be what you want/expect]. Instead, use source("file.R", local = TRUE) to avoid assigning functions and variables to the global environment.
* [[#How_to_exit_a_sourced_R_script|source()]] does not work like C's preprocessor where statements in header files will be literally inserted into the code. It does not work when you define a variable in a function but want to use it outside the function (even through '''source()''')


==== Launch RStudio ====
{{Pre}}
[[Rstudio#Multiple_versions_of_R|Multiple versions of R]]
## foo.R ##
cat(ArrayTools, "\n")
## End of foo.R


==== Create .Rproj file ====
# 1. Error
If you have an existing package that doesn't have an .Rproj file, you can use devtools::use_rstudio("path/to/package") to add it.
predict <- function() {
 
  ArrayTools <- "C:/Program Files" # or through load() function
With an RStudio project file, you can
  source("foo.R")                  # or through a function call; foo()
* Restore .RData into workspace at startup
}
* Save workspace to .RData on exit
predict()   # Object ArrayTools not found
* Always save history (even if no saving .RData)
* etc


==== package search ====
# 2. OK. Make the variable global
https://github.com/RhoInc/CRANsearcher
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict() 
ArrayTools


==== Git ====
# 3. OK. Create a global variable
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
ArrayTools <- "C:/Program Files"
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]
predict <- function() {
 
  source("foo.R")
=== Visual Studio ===
}
[http://blog.revolutionanalytics.com/2017/05/r-and-python-support-now-built-in-to-visual-studio-2017.html R and Python support now built in to Visual Studio 2017]
predict()
 
=== List files using regular expression ===
* Extension
<pre>
list.files(pattern = "\\.txt$")
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
<pre>
list.files(pattern = "^Something")
</pre>
</pre>


Using '''Sys.glob()"' as
'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
<pre>
> Sys.glob("~/Downloads/*.txt")
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
</pre>


=== Hidden tool: rsync in Rtools ===
Example 1.
<pre>
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
sending incremental file list
> ttt
a.exe
  type JpnTest
 
1    a    999
sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
2    b    999
total size is 1198416  speedup is 3.71
3    c     999
 
4    d    999
c:\Rtools\bin>
5    e    999
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1    a     N5
2    b    999
3    c    999
4    d    999
5    e    999
> ttt
  type JpnTest
1    a    999
2    b    999
3   c    999
4    d    999
5    e    999
</pre>
</pre>
And rsync works best when we need to sync folder.
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
binary/
binary/Eula.txt
binary/cherrytree.lnk
binary/depends64.chm
binary/depends64.dll
binary/depends64.exe
binary/mtputty.exe
binary/procexp.chm
binary/procexp.exe
binary/pscp.exe
binary/putty.exe
binary/sqlite3.exe
binary/wget.exe


sent 4115294 bytes  received 244 bytes  1175868.00 bytes/sec
Example 2. [http://stackoverflow.com/questions/1236620/global-variables-in-r How can we set global variables inside a function?] The answer is to use the "<<-" operator or '''assign(, , envir = .GlobalEnv)''' function.
total size is 8036311  speedup is 1.95


c:\Rtools\bin>rm c:\users\limingc\Documents\binary\procexp.exe
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
cygwin warning:
  MS-DOS style path detected: c:\users\limingc\Documents\binary\procexp.exe
  Preferred POSIX equivalent is: /cygdrive/c/users/limingc/Documents/binary/procexp.exe
  CYGWIN environment variable option "nodosfilewarning" turns off this warning.
  Consult the user's guide for more details about POSIX paths:
    http://cygwin.com/cygwin-ug-net/using.html#using-pathnames


c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/binary" "/cygdrive/c/users/limingc/Documents/"
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
sending incremental file list
binary/
binary/procexp.exe


sent 1767277 bytes  received 35 bytes  3534624.00 bytes/sec
=== New environment ===
total size is 8036311  speedup is 4.55
* http://adv-r.had.co.nz/Environments.html.
* [https://www.r-bloggers.com/2011/06/environments-in-r/ Environments in R]
* load(), attach(), with().
* [https://stackoverflow.com/questions/33109379/how-to-switch-to-a-new-environment-and-stick-into-it How to switch to a new environment and stick into it?] seems not possible!


c:\Rtools\bin>
Run the same function on a bunch of R objects
{{Pre}}
mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(<nowiki>mye[[n]]</nowiki>)
</pre>
</pre>


Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also http://superuser.com/questions/69620/rsync-file-permissions-on-windows
Just look at the contents of rda file without saving to anywhere (?load)
 
<pre>
=== Install rgdal package (geospatial Data) on ubuntu ===
local({
Terminal
  load("myfile.rda")
<syntaxhighlight lang='bash'>
  ls()
sudo apt-get install libgdal1-dev libproj-dev
})
</syntaxhighlight>
</pre>
 
Or use '''attach()''' which is a wrapper of load(). It creates an environment and slots it into the list right after the global environment, then populates it with the objects we're attaching.
R
{{Pre}}
<syntaxhighlight lang='rsplus'>
attach("all.rda") # safer and will warn about masked objects w/ same name in .GlobalEnv
install.packages("rgdal")
ls(pos = 2)
</syntaxhighlight>
##  also typically need to cleanup the search path:
 
detach("file:all.rda")
=== Set up Emacs on Windows ===
</pre>
Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
If we want to read data from internet, '''load()''' works but not attach().
<pre>
con <- url("http://some.where.net/R/data/example.rda")
## print the value to see what objects were created.
print(load(con))
close(con)
# Github example
# https://stackoverflow.com/a/62954840
</pre>
[https://stackoverflow.com/a/39621091 source() case].  
<pre>
<pre>
(setq-default inferior-R-program-name
myEnv <- new.env()   
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
source("some_other_script.R", local=myEnv)
attach(myEnv, name="sourced_scripts")
search()
ls(2)
ls(myEnv)
with(myEnv, print(x))
</pre>
</pre>


=== Database ===
=== str( , max) function ===
* https://cran.r-project.org/web/views/Databases.html
Use '''max.level''' parameter to avoid a long display of the structure of a complex R object. Use '''give.head = FALSE''' to hide the attributes. See [https://www.rdocumentation.org/packages/utils/versions/3.6.1/topics/str ?str]
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]


==== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ====
If we use str() on a function like str(lm), it is equivalent to args(lm)
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
* https://github.com/rstats-db/RSQLite


'''Creating a new database''':
For a complicated list object, it is useful to use the '''max.level''' argument; e.g. str(, max.level = 1)
<syntaxhighlight lang='rsplus'>
library(DBI)


mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
For a large data frame, we can use the '''tibble()''' function; e.g. mydf %>% tibble()
dbDisconnect(mydb)
unlink("my-db.sqlite")


# temporary database
=== tidy() function ===
mydb <- dbConnect(RSQLite::SQLite(), "")
broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See [[Tidyverse#broom|here]].
dbDisconnect(mydb)
</syntaxhighlight>


'''Loading data''':
=== View all objects present in a package, ls() ===
<syntaxhighlight lang='rsplus'>
https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get
mydb <- dbConnect(RSQLite::SQLite(), "")
{{Pre}}
dbWriteTable(mydb, "mtcars", mtcars)
> devtools::load_all("mypackage")
dbWriteTable(mydb, "iris", iris)
> search()
[1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
[4] "package:stats"    "package:graphics"  "package:grDevices"
[7] "package:utils"    "package:datasets" "package:methods"
[10] "Autoloads"         "package:base"


dbListTables(mydb)
> ls("package:mypackage")
[1] "_mypackage_rcpp_hello_world" "evalCpp"                    "library.dynam.unload"     
[4] "rcpp_hello_world"            "system.file"
</pre>


dbListFields(con, "mtcars")
Note that the first argument of ls() (or detach()) is used to specify the environment. It can be
* an integer (the position in the ‘search’ list);
* the character string name of an element in the search list;
* an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).


dbReadTable(con, "mtcars")
== Speedup R code ==
</syntaxhighlight>
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+


'''Queries''':
=== Profiler ===
<syntaxhighlight lang='rsplus'>
* [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler] (Video)
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')
* [https://github.com/atheriel/xrprof-package xrprof] package, [https://www.infoworld.com/article/3604688/top-r-tips-and-news-from-rstudio-global-2021.amp.html Top R tips and news from RStudio Global 2021]


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
== && vs & ==
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.  


dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
* The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The return is a vector.
* The longer form evaluates left to right examining only the first element of each vector. The return is one value.
* '''The longer form''' evaluates left to right examining only the first element of each vector. '''Evaluation proceeds only until the result is determined.'''
* The idea of the longer form && in R seems to be the same as the && operator in linux shell; see [https://youtu.be/AVXYq8aL47Q?t=1475 here].
* [https://medium.com/biosyntax/single-or-double-and-operator-and-or-operator-in-r-442f00332d5b Single or double?: AND operator and OR operator in R]. The confusion might come from the inconsistency when choosing these operators in different languages. For example, in C, & performs bitwise AND, while && does Boolean logical AND.
* [https://www.tjmahr.com/think-of-stricter-logical-operators/ Think of && as a stricter &]


res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
<pre>
dbFetch(res)
c(T,F,T) & c(T,T,T)
</syntaxhighlight>
# [1]  TRUE FALSE  TRUE
c(T,F,T) && c(T,T,T)
# [1] TRUE
c(T,F,T) && c(F,T,T)
# [1] FALSE
c(T,F,T) && c(NA,T,T)
# [1] NA
</pre>
<pre>
# Assume 'b' is not defined
> if (TRUE && b==3) cat("end")
Error: object 'b' not found
> if (FALSE && b==3) cat("end")
> # No error since the 2nd condition is never evaluated
</pre>
It's useful in functions(). We don't need nested if statements. In this case if 'arg' is missing, the argument 'L' is not needed so there is not syntax error.
<pre>
> foo <- function(arg, L) {
  # Suppose 'L' is meaningful only if 'arg' is provided
  #
  # Evaluate 'L' only if 'arg' is provided
  #
  if (!missing(arg) && L) {
    print("L is true")
  } else {
    print("Either arg is missing or L is FALSE")
  }
}
> foo()
[1] "arg is missing or L is FALSE"
> foo("a", F)
[1] "arg is missing or L is FALSE"
> foo("a", T)
[1] "L is true"
</pre>
Other examples: '''&&''' is more flexible than '''&'''.
<pre>
nspot <- ifelse(missing(rvm) || !rvm, nrow(exprTrain), sum(filter))


'''Batched queries''':
if (!is.null(exprTest) && any(is.na(exprTest))) { ... }
<syntaxhighlight lang='rsplus'>
</pre>
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


dbClearResult(rs)
== for-loop, control flow ==
</syntaxhighlight>
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Control ?Control]
* '''next''' can be used to skip the rest of the inner-most loop
* [https://www.programiz.com/r/ifelse-function ifelse() Function]


'''Multiple parameterised queries''':
== Vectorization ==
<syntaxhighlight lang='rsplus'>
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
nrow(dbFetch(rs))
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
#> [1] 4
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html
dbClearResult(rs)
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
</syntaxhighlight>
* [https://statcompute.wordpress.com/2018/09/16/why-vectorize/ Why Vectorize?] statcompute.wordpress.com
* [https://www.jimhester.com/2018/04/12/vectorize/ Beware of Vectorize] from Jim Hester
* [https://github.com/henrikbengtsson/matrixstats matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors). E.g. col / rowMedians(), col / rowRanks(), and col / rowSds(). [https://github.com/HenrikBengtsson/matrixStats/wiki/Benchmark-reports Benchmark reports].
 
=== sapply vs vectorization ===
[http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs vectorization]


'''Statements''':
=== lapply vs for loop ===
<syntaxhighlight lang='rsplus'>
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
* https://code-examples.net/en/q/286e03a
#> [1] 0
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)
</syntaxhighlight>


==== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ====
=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
Manipulate R data frames using SQL. Depends on RSQLite. [http://datascienceplus.com/a-use-of-gsub-reshape2-and-sqldf-with-healthcare-data/ A use of gsub, reshape2 and sqldf with healthcare data]
split() can be used to split a vector, columns or rows. See [https://stackoverflow.com/questions/3302356/how-to-split-a-data-frame How to split a data frame?]
<ul>
<li>Split divides the data in the '''vector''' or '''data frame''' x into the groups defined by f. The syntax is
{{Pre}}
split(x, f, drop = FALSE, …)
</pre>


==== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ====
<li>split() + cut(). [https://www.r-bloggers.com/2024/10/how-to-split-data-into-equal-sized-groups-in-r-a-comprehensive-guide-for-beginners/ How to Split Data into Equal Sized Groups in R: A Comprehensive Guide for Beginners]
<li>[https://stackoverflow.com/a/3321659 Split a vector into chunks]. split() returns a vector/indices and the indices can be used in lapply() to subset the data. Useful for the '''split() + lapply() + do.call()''' or '''split() + sapply()''' operations.
<pre>
d <- 1:10
chunksize <- 4
ceiling(1:10/4)
# [1] 1 1 1 1 2 2 2 2 3 3
split(d, ceiling(seq_along(d)/chunksize))
# $`1`
# [1] 1 2 3 4
#
# $`2`
# [1] 5 6 7 8
#
# $`3`
# [1]  9 10
do.call(c, lapply(split(d, ceiling(seq_along(d)/4)), function(x) sum(x)) )
#  1  2  3
# 10 26 19


==== [[MySQL#Use_through_R|RMySQL]] ====
# bigmemory vignette
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
planeindices <- split(1:nrow(x), x[,'TailNum'])
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.
planeStart <- sapply(planeindices,
                    function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
</pre>


==== MongoDB ====
<li>Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.
* http://www.r-bloggers.com/r-and-mongodb/
{{Pre}}
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/
split(mtcars,mtcars$cyl)


==== odbc ====
split(data.frame(matrix(1:20, nr=10) ), ceiling(1:10/chunksize)) # data.frame/tibble works
split.data.frame(matrix(1:20, nr=10), ceiling(1:10/chunksize))  # split.data.frame() works for matrices
</pre>


==== RODBC ====
<li>Split columns of a data frame/matrix.
{{Pre}}
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
</pre>


==== DBI ====
<li>split() + sapply() to merge columns. See below [[#Mean_of_duplicated_columns:_rowMeans.3B_compute_Means_by_each_row|Mean of duplicated columns]] for more detail.


==== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ====
<li>split() + sapply() to split a vector. See [https://www.rdocumentation.org/packages/genefilter/versions/1.54.2/topics/nsFilter nsFilter()] function which can remove duplicated probesets/rows using unique Entrez Gene IDs ('''genefilter''' package). The source code of [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/nsFilter.R#L224 nsFilter()] and [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/all.R#L170 findLargest()].
* To use databases with dplyr, you need to first install dbplyr
{{Pre}}
* https://db.rstudio.com/dplyr/
tSsp = split.default(testStat, lls)
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
# testStat is a vector of numerics including probeset IDs as names
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html
# lls is a vector of entrez IDs (same length as testStat)
# tSSp is a list of the same length as unique elements of lls.


'''Create a new SQLite database''':
sapply(tSsp, function(x) names(which.max(x)))
<syntaxhighlight lang='rsplus'>
# return a vector of probset IDs of length of unique entrez IDs
surveys <- read.csv("data/surveys.csv")
</pre>
plots <- read.csv("data/plots.csv")
</ul>


my_db_file <- "portal-database.sqlite"
=== strsplit and sapply ===
my_db <- src_sqlite(my_db_file, create = TRUE)
{{Pre}}
> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
> strsplit(namedf, " ")
[[1]]
[1] "John" "ABC"


copy_to(my_db, surveys)
[[2]]
copy_to(my_db, plots)
[1] "Mary" "CDE"
my_db
</syntaxhighlight>


'''Connect to a database''':
[[3]]
<syntaxhighlight lang='rsplus'>
[1] "Kat" "FGH"
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


library(dbplyr)
> sapply(strsplit(namedf, " "), "[", 1)
library(dplyr)
[1] "John" "Mary" "Kat"
mammals <- src_sqlite("portal_mammals.sqlite")
> sapply(strsplit(namedf, " "), "[", 2)
</syntaxhighlight>
[1] "ABC" "CDE" "FGH"
</pre>


'''Querying the database with the SQL syntax''':
=== Mean of duplicated columns: rowMeans; compute Means by each row ===
<syntaxhighlight lang='rsplus'>
<ul>
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
<li>[https://stackoverflow.com/questions/35925529/reduce-columns-of-a-matrix-by-a-function-in-r Reduce columns of a matrix by a function in R]. To use rowMedians() instead of rowMeans(), we need to install [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats] from CRAN.
</syntaxhighlight>
<syntaxhighlight lang='r'>
set.seed(1)
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
colnames(x) <- c("b", "b", "b", "c", "a", "a"); x
res <- sapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
res  # notice the sorting of columns
      a  b  c
[1,] 46  1 31
[2,] 47 12 32
[3,] 48 13 33
[4,] 49 14 34
[5,] 50 15 35
[6,] 51 16 36
[7,] 52 17 37
[8,] 53 18 38
[9,] 54 19 39
[10,] 55 20 40


'''Querying the database with the dplyr syntax''':
# vapply() is safter than sapply().
<syntaxhighlight lang='rsplus'>
# The 3rd arg in vapply() is a template of the return value.
surveys <- tbl(mammals, "surveys")
res2 <- vapply(split(1:ncol(x), colnames(x)),
surveys %>%
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
    select(year, species_id, plot_id)
              rep(0, nrow(x)))
head(surveys, n = 10)
 
show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database
</syntaxhighlight>
</syntaxhighlight>
</li>
<li>[https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/colSums colSums, rowSums, colMeans, rowMeans] (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster.
{{Pre}}
rowMeans(x, na.rm=T)
# [1] 31 27 28 29 30 31 32 33 34 35


'''Simple database queries''':
apply(x, 1, mean, na.rm=T)
<syntaxhighlight lang='rsplus'>
# [1] 31 27 28 29 30 31 32 33 34 35
surveys %>%
</pre>
  filter(weight < 5) %>%
</li>
  select(species_id, sex, weight)
<li>[https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
</syntaxhighlight>
</li>
<li>[https://www.statforbiology.com/2020/stat_r_tidyverse_columnwise/ From ''for()'' loops to the ''split-apply-combine'' paradigm for column-wise tasks: the transition for a dinosaur]
</li>
</ul>


'''Laziness''' (instruct R to stop being lazy):
=== Mean of duplicated rows: colMeans and rowsum ===
<syntaxhighlight lang='rsplus'>
<ul>
data_subset <- surveys %>%
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/colSums colMeans(x, na.rm = FALSE, dims = 1)], take mean per columns & sum over rows. It returns a vector. Other similar idea functions include '''colSums, rowSums, rowMeans'''.
  filter(weight < 5) %>%
{{Pre}}
   select(species_id, sex, weight) %>%
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
  collect()
rownames(x) <- c(rep("b", 2), rep("c", 3), rep("d", 4), "a") # move 'a' to the last
</syntaxhighlight>
res <- sapply(split(1:nrow(x), rownames(x)),
              function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
res  # still a matrix, rows are ordered
#   [,1] [,2] [,3] [,4] [,5] [,6]
# a 10.0 20.0 30.0 40.0 50.0 60.0
# b  1.5 12.0 22.0 31.5 41.5 51.5
# c  4.0 14.0 24.0 34.0 44.0 54.0
# d  7.5 17.5 27.5 37.5 47.5 57.5
table(rownames(x))
# a b c d
# 1 2 3 4


'''Complex database queries''':
aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
<syntaxhighlight lang='rsplus'>
#  Group.1  V1  V2  V3  V4  V5  V6
plots <- tbl(mammals, "plots")
# 1      a 10.0 20.0 30.0 40.0 50.0 60.0
plots # # The plot_id column features in the plots table
# 2      b  1.5 12.0 22.0 31.5 41.5 51.5
# 3      c  4.0 14.0 24.0 34.0 44.0 54.0
# 4      d  7.5 17.5 27.5 37.5 47.5 57.5
</pre>
<li>[[Arraytools#Reducing_multiple_probes.2Fprobe_sets_to_one_per_gene_symbol|Reduce multiple probes by the maximally expressed probe (set) measured by average intensity across arrays]]


surveys # The plot_id column also features in the surveys table
</li>
 
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/rowsum rowsum(x, group, reorder = TRUE, …)]. Sum over rows. It returns a matrix. This is very special. It's not the same as rowSums. There is no "colsum" function. ''It has the speed advantage over sapply+colSums OR aggregate.''
# Join databases method 1
{{Pre}}
plots %>%
group <- rownames(x)
  filter(plot_id == 1) %>%
rowsum(x, group, na.rm=T)/as.vector(table(group))
  inner_join(surveys) %>%
#   [,1] [,2] [,3] [,4] [,5] [,6]
   collect()
# a 10.0 20.0 30.0 40.0 50.0 60.0
</syntaxhighlight>
# b  1.5  6.0 11.0 31.5 41.5 51.5
# c  4.0 14.0 24.0 34.0 44.0 54.0
# d  7.5 17.5 27.5 37.5 47.5 57.5
</pre>
</li>
</ul>
* [https://stackoverflow.com/questions/25198442/how-to-calculate-mean-median-per-group-in-a-dataframe-in-r How to calculate mean/median per group in a dataframe in r] where '''doBy''' and '''dplyr''' are recommended.
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
* [http://stackoverflow.com/questions/7881660/finding-the-mean-of-all-duplicates use ave() and unique()]
* [http://stackoverflow.com/questions/17383635/average-between-duplicated-rows-in-r data.table package]
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
<ul>
<li>'''by()''' function. [https://thomasadventure.blog/posts/calculating-change-from-baseline-in-r/ Calculating change from baseline in R]
</li>
<li>See [https://finnstats.com/index.php/2021/06/20/aggregate-function-in-r/ '''aggregate''' Function in R- A powerful tool for data frames] & [https://finnstats.com/index.php/2021/06/01/summarize-in-r-data-summarization-in-r/ summarize in r, Data Summarization In R] </li>
<li>[http://www.statmethods.net/management/aggregate.html aggregate()] function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. [http://www.win-vector.com/blog/2015/10/dont-use-statsaggregate/ Don't use aggregate] post.
{{Pre}}
> attach(mtcars)
dim(mtcars)
[1] 32 11
> head(mtcars)
                  mpg cyl disp  hp drat    wt  qsec vs am gear carb
Mazda RX4        21.0  6  160 110 3.90 2.620 16.46  0  1    4    4
Mazda RX4 Wag    21.0  6  160 110 3.90 2.875 17.02  0  1    4    4
Datsun 710        22.8  4  108  93 3.85 2.320 18.61  1  1    4    1
Hornet 4 Drive    21.4   6  258 110 3.08 3.215 19.44  1  0    3    1
Hornet Sportabout 18.7  8  360 175 3.15 3.440 17.02  0  0    3    2
Valiant          18.1  6  225 105 2.76 3.460 20.22  1  0    3    1
> with(mtcars, table(cyl, vs))
  vs
cyl  0  1
  4  1 10
  6  3  4
  8 14  0
> aggdata <-aggregate(mtcars, by=list(cyl,vs),  FUN=mean, na.rm=TRUE)
> print(aggdata)
  Group.1 Group.2      mpg cyl  disp      hp    drat      wt    qsec vs
1      4      0 26.00000  4 120.30  91.0000 4.430000 2.140000 16.70000  0
2      6      0 20.56667  6 155.00 131.6667 3.806667 2.755000 16.32667  0
3      8      0 15.10000  8 353.10 209.2143 3.229286 3.999214 16.77214  0
4      4      1 26.73000  4 103.62  81.8000 4.035000 2.300300 19.38100  1
5      6      1 19.12500  6 204.55 115.2500 3.420000 3.388750 19.21500  1
        am    gear    carb
1 1.0000000 5.000000 2.000000
2 1.0000000 4.333333 4.666667
3 0.1428571 3.285714 3.500000
4 0.7000000 4.000000 1.500000
5 0.0000000 3.500000 2.500000
> detach(mtcars)


==== NoSQL ====
# Another example: select rows with a minimum value from a certain column (yval in this case)
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
> mydf <- read.table(header=T, text='
 
id xval yval
=== Github ===
A 1  1
 
A -2  2
==== R source  ====
B 3  3
https://github.com/wch/r-source/  Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.
B 4  4
C 5  5
')
> x = mydf$xval
> y = mydf$yval
> aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
  id xval yval
1  A    1    1
2  B    3    3
3  C    5    5
</pre>
</li>
</ul>
 
=== Mean by Group ===
[https://statisticsglobe.com/mean-by-group-in-r Mean by Group in R (2 Examples) | dplyr Package vs. Base R]
<pre>
aggregate(x = iris$Sepal.Length,                # Specify data column
          by = list(iris$Species),              # Specify group indicator
          FUN = mean)                          # Specify function (i.e. mean)
</pre>
<pre>
library(dplyr)
iris %>%                                        # Specify data frame
  group_by(Species) %>%                        # Specify group indicator
  summarise_at(vars(Sepal.Length),              # Specify column
              list(name = mean))              # Specify function
</pre>
* [https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/ave ave(x, ..., FUN)],
* aggregate(x, by, FUN),
* by(x, INDICES, FUN): return is a list
* tapply(): return results as a matrix or array. Useful for [https://en.wikipedia.org/wiki/Jagged_array ragged array].
 
== Apply family ==
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.
 
The following list gives a hierarchical relationship among these functions.
* '''apply'''(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
* '''lapply'''(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
** '''sapply'''(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
*** '''replicate'''(n, expr, simplify = "array")
** '''mapply'''(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
*** '''Vectorize'''(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
*** '''Map'''(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
** '''vapply'''(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
** '''rapply'''(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
* '''tapply'''(V, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a [https://en.wikipedia.org/wiki/Jagged_array "Ragged" Array]. V is typically a vector where split() will be applied. INDEX is a list of one or more factors.
** '''aggregate'''(D, by, FUN, ..., simplify = TRUE, drop = TRUE) - Apply a function to each '''columns''' of subset data frame split by factors. FUN (such as mean(), weighted.mean(), sum()) is a simple function applied to a vector. D is typically a data frame. This is used to '''summarize''' data.
** '''by'''(D, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to each '''subset data frame''' split by factors. FUN (such as summary(), lm()) is applied to a data frame. D is typically a data frame.
* '''eapply'''(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment


If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
[https://www.queryhome.com/tech/76799/r-difference-between-apply-vs-sapply-vs-lapply-vs-tapply Difference between apply vs sapply vs lapply vs tapply?]
* apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
* lapply - When you want to apply a function to each element of a list in turn and get a list back.
* sapply - When you want to apply a function to each element of a list in turn, but you want a vector back, rather than a list.
* tapply - When you want to apply a function to subsets of a vector and the subsets are defined by some other vector, usually a factor.


==== R packages (only) source (metacran) ====
Some short examples:
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.
* [http://people.stern.nyu.edu/ylin/r_apply_family.html stern.nyu.edu].
* [http://www.datasciencemadesimple.com/apply-function-r/ Apply Function in R – apply vs lapply vs sapply vs mapply vs tapply vs rapply vs vapply] from datasciencemadesimple.com.
* [https://stackoverflow.com/a/7141669 How to use which one (apply family) when?]


==== Bioconductor packages source ====
=== Apply vs for loop ===
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>
Note that, apply's performance is not always better than a for loop. See
* http://tolstoy.newcastle.edu.au/R/help/06/05/27255.html (answered by Brian Ripley)
* https://stat.ethz.ch/pipermail/r-help/2014-October/422455.html (has one example)
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]. The author said '' 'an important reason for using *apply() functions may instead be that they fit the functional programming paradigm better, where everything is done using function calls and side effects are reduced'... The scope of the variables defined within f is limited to f, and variables defined outside f cannot be modified inside f (except using the special scoping assignment operator <<-).  ''
** [http://adv-r.had.co.nz/Functional-programming.html Functional programming]
* [https://privefl.github.io/blog/why-loops-are-slow-in-r/ Why loops are slow in R]
* [https://stackoverflow.com/a/18763102 Why is `unlist(lapply)` faster than `sapply`?]


==== Send local repository to Github in R by using reports package ====
=== Progress bar ===
http://www.youtube.com/watch?v=WdOI_-aZV0Y
[http://peter.solymos.org/code/2016/09/11/what-is-the-cost-of-a-progress-bar-in-r.html What is the cost of a progress bar in R?]


==== My collection ====
The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out [http://www.theanalystatlarge.com/for-loop-tracking-windows-progress-bar/ this post]. It uses  winProgressBar() and setWinProgressBar() functions.
* https://github.com/arraytools
* https://gist.github.com/4383351 heatmap using leukemia data
* https://gist.github.com/4382774 heatmap using sequential data
* https://gist.github.com/4484270 biocLite


==== How to download ====
[https://www.jottr.org/2020/07/04/progressr-erum2020-slides/ e-Rum 2020 Slides on Progressr] by Henrik Bengtsson. [https://www.jottr.org/2021/06/11/progressr-0.8.0/ progressr 0.8.0: RStudio's progress bar, Shiny progress updates, and absolute progress], [https://www.r-bloggers.com/2022/06/progressr-0-10-1-plyr-now-supports-progress-updates-also-in-parallel/ progressr 0.10.1: Plyr Now Supports Progress Updates also in Parallel]


Clone ~ Download.
=== simplify option in sapply() ===
* Command line
<pre>
<pre>
git clone https://gist.github.com/4484270.git
library(KEGGREST)
 
names1 <- keggGet(c("hsa05340", "hsa05410"))
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2
 
names3 <- keggGet(c("hsa05340"))
names4 <- sapply(names3, function(x) x$GENE)
length(names4)  # may or may not be what we expect
# [1] 76
names4 <- sapply(names3, function(x) x$GENE, simplify = FALSE)
length(names4)  # same if we use lapply() w/o simplify
# [1] 1
</pre>
</pre>
This will create a subdirectory called '4484270' with all cloned files there.


* Within R
=== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ===
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
<ul>
<li>mapply() [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/mapply documentation]. [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Use mapply() to merge lists].
<pre>
<pre>
library(devtools)
mapply(rep, 1:4, 4:1)
source_gist("4484270")
mapply(rep, times = 1:4, x = 4:1)
mapply(function(x, y) seq_len(x) + y,
      c(a =  1, b = 2, c = 3),  # names from first
      c(A = 10, B = 0, C = -10))
mapply(c, firstList, secondList, SIMPLIFY=FALSE)
</pre>
</pre>
or
</li>
First download the json file from
<li>[https://bensstats.wordpress.com/2020/10/06/robservations-3-finding-the-expected-value-of-the-maximum-of-two-bivariate-normal-variables-with-simulation/ Finding the Expected value of the maximum of two Bivariate Normal variables with simulation] sapply + mapply.
https://api.github.com/users/MYUSERLOGIN/gists
and then
<pre>
<pre>
library(RJSONIO)
z <- mapply(function(u, v) { max(u, v) },
x <- fromJSON("~/Downloads/gists.json")
            u = x[, 1], v = x[, 2])
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})
</pre>
</pre>
</li>
<li>[http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming </li>
<li>Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
<ul>
<li>If you have two or more lists (or data frames) that you need to process in <span style="color: red">parallel</span>, use '''Map()'''. One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to '''mapply()''' function and is more concise than '''lapply()'''. [http://adv-r.had.co.nz/Functionals.html#functionals-loop Advanced R] has a comment that Map() is better than mapply().
{{Pre}}
# Syntax: Map(f, ...)


==== Jekyll ====
xs <- replicate(5, runif(10), simplify = FALSE)
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
Map(weighted.mean, xs, ws)


=== Connect R with Arduino ===
# instead of a more clumsy way
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
lapply(seq_along(xs), function(i) {
* http://jean-robert.github.io/2012/11/11/thermometer-R-using-Arduino-Java.html
  weighted.mean(xs[[i]], ws[[i]])
* http://bio7.org/?p=2049
})
* http://www.rforge.net/Arduino/svn.html
</pre>
</li>
<li>Reduce() reduces a vector, x, to a single value by <span style="color: red">recursively</span> calling a function, f, two arguments at a time. A good example of using '''Reduce()''' function is to read a list of matrix files and merge them. See [https://stackoverflow.com/questions/29820029/how-to-combine-multiple-matrix-frames-into-one-using-r How to combine multiple matrix frames into one using R?]
{{Pre}}
# Syntax: Reduce(f, x, ...)


=== Android App ===
> m1 <- data.frame(id=letters[1:4], val=1:4)
* [https://play.google.com/store/apps/details?id=appinventor.ai_RInstructor.R2&hl=zh_TW R Instructor] $4.84
> m2 <- data.frame(id=letters[2:6], val=2:6)
* [http://realxyapp.blogspot.tw/2010/12/statistical-distribution.html Statistical Distribution] (Not R related app)
> merge(m1, m2, "id", all = T)
* [https://datascienceplus.com/data-driven-introspection-of-my-android-mobile-usage-in-r/ Data-driven Introspection of my Android Mobile usage in R]
  id val.x val.y
 
1  a    1    NA
=== Common plots tips ===
2  b    2    2
==== Grouped boxplots ====
3  c    3    3
* [http://sphaerula.com/legacy/R/boxplotTwoWay.html Box Plots of Two-Way Layout]
4  d    4    4
* [http://r-video-tutorial.blogspot.com/2013/06/box-plot-with-r-tutorial.html Step by step to create a grouped boxplots]
5  e    NA    5
** 'at' parameter in boxplot() to change the equal spaced boxplots
6  f    NA    6
** embed par(mar=) in boxplot()
> m <- list(m1, m2)
** mtext(line=) to solve the problem the xlab overlapped with labels.
> Reduce(function(x,y) merge(x,y, "id",all=T), m)
* [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others ggplot2 approach] (Hint: '''facet_grid''' is used)
  id val.x val.y
1  a    1    NA
2  b    2    2
3  c    3    3
4  d    4    4
5  e    NA    5
6  f    NA    6
</pre>
</li>
</ul>
</li>
</ul>
* [https://statcompute.wordpress.com/2018/09/08/playing-map-and-reduce-in-r-subsetting/ Playing Map() and Reduce() in R – Subsetting] - using parallel and future packages. [https://statcompute.wordpress.com/2018/09/22/union-multiple-data-frames-with-different-column-names/ Union Multiple Data.Frames with Different Column Names]


==== [https://www.samruston.co.uk/ Weather Time Line] ====
=== sapply & vapply ===
The plot looks similar to a boxplot though it is not. See a [https://www.samruston.co.uk/images/screens/screen_2.png screenshot] on Android by [https://www.samruston.co.uk/ Sam Ruston].
* [http://stackoverflow.com/questions/12339650/why-is-vapply-safer-than-sapply This] discusses why '''vapply''' is safer and faster than sapply.
* [http://adv-r.had.co.nz/Functionals.html#functionals-loop Vector output: sapply and vapply] from Advanced R (Hadley Wickham).
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
* [http://theautomatic.net/2019/03/13/speed-test-sapply-vs-vectorization/ Speed test: sapply vs. vectorization]
* sapply can be used in plotting; for example, [https://cran.r-project.org/web/packages/glmnet/vignettes/relax.pdf#page=13 glmnet relax vignette] uses '''sapply(myList, lines, col="grey") ''' to draw multiple lines simultaneously on a list of matrices.


==== Horizontal bar plot ====
See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up [https://github.com/SRTRdevhub/C_Statistic_Github/blob/master/Simulation_Demonstration.Rmd#L115 this example].
<syntaxhighlight lang='rsplus'>
library(ggplot2)
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA",
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) +
  coord_flip() + xlab("") + ylab("Fold Change") 
</syntaxhighlight>


[[File:Ggplot2bar.svg|300px]]
=== rapply - recursive version of lapply ===
* http://4dpiecharts.com/tag/recursive/
* [https://github.com/wch/r-source/search?utf8=%E2%9C%93&q=rapply Search in R source code]. Mainly [https://github.com/wch/r-source/blob/trunk/src/library/stats/R/dendrogram.R r-source/src/library/stats/R/dendrogram.R].


==== Include bar values in a barplot ====
=== replicate ===
* https://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r.
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
* [http://stackoverflow.com/questions/12481430/how-to-display-the-frequency-at-the-top-of-each-factor-in-a-barplot-in-r barplot(), text() and axis()] functions. The data can be from a table() object.
{{Pre}}
* [https://stackoverflow.com/questions/11938293/how-to-label-a-barplot-bar-with-positive-and-negative-bars-with-ggplot2 How to label a barplot bar with positive and negative bars with ggplot2]
> replicate(5, rnorm(3))
 
          [,1]      [,2]      [,3]      [,4]        [,5]
Use text().
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</pre>


Or use geom_text() if we are using the ggplot2 package. See an example [http://dsgeek.com/2014/09/19/Customizingggplot2charts.html here] or [https://rpubs.com/escott8908/RGC_Ch3_Gar_Graphs this].
See [[#parallel_package|parSapply()]] for a parallel version of replicate().


For stacked barplot, see [http://t-redactyl.io/blog/2016/01/creating-plots-in-r-using-ggplot2-part-4-stacked-bar-plots.html this] post.
=== Vectorize ===
* [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/Vectorize Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE)]: creates a function wrapper that vectorizes a scalar function. Its value is a list or vector or array. It calls '''mapply()'''.
{{Pre}}
> rep(1:4, 4:1)
[1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
[[1]]
[1] 1 1 1 1


==== Grouped barplots ====
[[2]]
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)<syntaxhighlight lang='rsplus'>
[1] 2 2 2
library(ggplot2)
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) +
      geom_bar(position="dodge", stat="identity")
</syntaxhighlight>
* https://datascienceplus.com/building-barplots-with-error-bars/. The error bars define 2 se (95% interval) for the black-and-white version and 1 se (68% interval) for ggplots. Be careful.<syntaxhighlight lang='rsplus'>
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042
</syntaxhighlight>
* [http://stackoverflow.com/questions/27466035/adding-values-to-barplot-of-table-in-r two bars in one factor] (stack). The data can be a 2-dim matrix with numerical values.
* [http://stats.stackexchange.com/questions/3879/how-to-put-values-over-bars-in-barplot-in-r two bars in one factor], [https://stats.stackexchange.com/questions/14118/drawing-multiple-barplots-on-a-graph-in-r Drawing multiple barplots on a graph in R] (next to each other)
** [https://datascienceplus.com/building-barplots-with-error-bars/ Include error bars]
* [http://bl.ocks.org/patilv/raw/7360425/ Three variables] barplots
* [https://peltiertech.com/stacked-bar-chart-alternatives/ More alternatives] (not done by R)


==== Math expression ====
[[3]]
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
[1] 3 3
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* http://vis.supstat.com/2013/04/mathematical-annotation-in-r/
* https://andyphilips.github.io/blog/2017/08/16/mathematical-symbols-in-r-plots.html


<syntaxhighlight lang='rsplus'>
[[4]]
# Expressions
[1] 4
plot(x,y, xlab = expression(hat(x)[t]),
</pre>
    ylab = expression(phi^{rho + a}),
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy]
    main = "Pure Expressions")
{{Pre}}
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))  
[1] 1.6491761 0.9610109
</pre>
* https://blogs.msdn.microsoft.com/gpalem/2013/03/28/make-vectorize-your-friend-in-r/ 
{{Pre}}
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15


# Expressions with Spacing
myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
# '~' is to add space and '*' is to squish characters together
myfunc2(1, 2) # 2
plot(1:10, xlab= expression(Delta * 'C'))
myfunc2(3, 5) # 15
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
myfunc2(c(1,3), c(2,5)) # NA
    ylab = expression(phi^{rho + a} * z * w),
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
    main = "Pure Expressions with Spacing")
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
                                        # parameter will be re-used
</pre>
 
== plyr and dplyr packages ==
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
 
[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.


# Expressions with Text
[http://seananderson.ca/courses/12-plyr/plyr_2012.pdf A quick introduction to plyr] with a summary of apply functions in R and compare them with functions in plyr package.
plot(x,y,
    xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")),
    ylab = expression(paste("Here is some text of ", phi^{rho})),
    main = "Expressions with Text")


# Substituting Expressions
# plyr has a common syntax -- easier to remember
plot(x,y,
# plyr requires less code since it takes care of the input and output format
    xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)),
# plyr can easily be run in parallel -- faster
    ylab = substitute(paste("e is = ", e ), list(e = ee)),
    main = "Substituted Expressions")
</syntaxhighlight>


==== Impose a line to a scatter plot ====
Tutorials
* abline + lsfit # least squares
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
: <syntaxhighlight lang='rsplus'>
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
plot(cars)
* [http://www.dataschool.io/dplyr-tutorial-for-faster-data-manipulation-in-r/ Hands-on dplyr tutorial for faster data manipulation in R] from dataschool.io.
abline(lsfit(cars[, 1], cars[, 2]))
 
# OR
Examples of using dplyr:
abline(lm(cars[,2] ~ cars[,1]))
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]  
</syntaxhighlight>
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
* abline + line # robust line fitting
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
: <syntaxhighlight lang='rsplus'>
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
</syntaxhighlight>
* lines
: <syntaxhighlight lang='rsplus'>
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')
</syntaxhighlight>
==== Rotating x axis labels for barplot ====
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
<syntaxhighlight lang='rsplus'>
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
</syntaxhighlight>


==== Set R plots x axis to show at y=0 ====
=== [https://cran.r-project.org/web/packages/tibble/ tibble] ===
https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0
[https://www.r-bloggers.com/2024/08/tidy-dataframes-but-not-tibbles/ Tidy DataFrames but not Tibbles]
<syntaxhighlight lang='rsplus'>
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</syntaxhighlight>


==== Different colors of axis labels in barplot ====
Tibble objects
See [https://stackoverflow.com/questions/18839731/vary-colors-of-axis-labels-in-r-based-on-another-variable Vary colors of axis labels in R based on another variable]
* it does not have row names (cf data frame),
* it never changes the type of the inputs (e.g. it never converts strings to factors!),
* it never changes the names of variables


Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
To show all rows or columns of a tibble object,
<syntaxhighlight lang='rsplus'>
<pre>
tN <- table(Ni <- stats::rpois(100, lambda = 5))
print(tbObj, n= Inf)
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")
</syntaxhighlight>


Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
print(tbObj, width = Inf)
<syntaxhighlight lang='rsplus'>
</pre>
barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)
</syntaxhighlight>


==== Use [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/text text()] to draw labels on X/Y-axis including rotation ====
If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.
* adj = 1 means top/rigth alignment. The default is to center the text.
 
* [https://www.rdocumentation.org/packages/graphics/versions/3.4.3/topics/par par("usr")] gives the extremes of the user coordinates of the plotting region of the form c(x1, x2, y1, y2).
'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
** par("usr") is determined *after* a plot has been created
{{Pre}}
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
TibbleObject$VarName
* https://datascienceplus.com/building-barplots-with-error-bars/
# OR
<syntaxhighlight lang='rsplus'>
TibbleObject[["VarName"]]
par(mar = c(5, 6, 4, 5) + 0.1)
# OR
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
pull(TibbleObject, VarName) # won't be a tibble object anymore
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
    adj = 1, labels = myData$names, xpd = TRUE)
</syntaxhighlight>
* https://www.r-bloggers.com/rotated-axis-labels-in-r-plots/


==== Vertically stacked plots with the same x axis ====
# For multiple columns, use select()
https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
# OR
dplyr::select(TibbleObject, 2:5) #
</pre>


==== Increase/decrease legend font size ====
'''Convert a data frame to a tibble''' See [http://www.sthda.com/english/wiki/tibble-data-format-in-r-best-and-modern-way-to-work-with-your-data Tibble Data Format in R: Best and Modern Way to Work with Your Data]
https://stackoverflow.com/a/36842578
<pre>
<syntaxhighlight lang='rsplus'>
my_data <- as_tibble(iris)
plot(rnorm(100))
class(my_data)
op <- par(cex=2)
</pre>
legend("topleft", legend = 1:4, col=1:4, pch=1)
par(op)
</syntaxhighlight>


==== Superimpose a density plot or any curves ====
=== llply() ===
Use '''lines()'''.  
llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.
<pre>
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
</pre>
where rLLID is a list of entrez ID. For example,
<pre>
get("org.Hs.egGO")[["6772"]]
</pre>
returns a list of 49 GOs.


Example 1
=== ddply() ===
<syntaxhighlight lang='rsplus'>
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))
</syntaxhighlight>


Example 2
=== ldply() ===
<syntaxhighlight lang='rsplus'>
[http://rpsychologist.com/an-r-script-to-automatically-look-at-pubmed-citation-counts-by-year-of-publication/ An R Script to Automatically download PubMed Citation Counts By Year of Publication]
require(survival)
 
n = 10000
=== Performance/speed comparison ===
beta1 = 2; beta2 = -1
[https://www.r-bloggers.com/2023/01/performance-comparison-of-converting-list-to-data-frame-with-r-language/ Performance comparison of converting list to data.frame with R language]
lambdaT = 1 # baseline hazard
 
lambdaC = 2  # hazard of censoring
== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ==
set.seed(1234)
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
x1 = rnorm(n,0)
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2))
C <- rweibull(n, shape=1, scale=lambdaC) 
time = pmin(T,C) 
status <- 1*(T <= C)
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1),
    ylab="Survival probability",
    main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull
</syntaxhighlight>


==== Custom scales ====
=== get_seed() ===
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
See the same blog
{{Pre}}
get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}
</pre>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


=== Time series ===
=== Random seeds ===
* [https://www.amazon.com/Applied-Time-Analysis-R-Second/dp/1498734227 Applied Time Series Analysis with R]
By default, R uses the exact time in milliseconds of the computer's clock when R starts up to generate a seed. See [https://stat.ethz.ch/R-manual/R-patched/library/base/html/Random.html ?Random].
* [http://www.springer.com/us/book/9780387759586 Time Series Analysis With Applications in R]
<pre>
set.seed(as.numeric(Sys.time()))


==== Time series stock price plot ====
set.seed(as.numeric(Sys.Date()))  # same seed for each day
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
</pre>
* [https://datascienceplus.com/visualize-your-portfolios-performance-and-generate-a-nice-report-with-r/ Visualize your Portfolio’s Performance and Generate a Nice Report with R]
* https://timelyportfolio.github.io/rCharts_time_series/history.html


<syntaxhighlight lang='rsplus'>
=== .Machine and the largest integer, double ===
library(quantmod)
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].
getSymbols("AAPL")
{{Pre}}
getSymbols("IBM") # similar to AAPL
                          Linux/Mac  32-bit Windows 64-bit Windows
getSymbols("CSCO") # much smaller than AAPL, IBM
double.eps              2.220446e-16  2.220446e-16  2.220446e-16
getSymbols("DJI") # Dow Jones, huge
double.neg.eps          1.110223e-16  1.110223e-16  1.110223e-16
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)",
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
    col='orange', subset = '2017::2017-08')
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
double.base            2.000000e+00  2.000000e+00  2.000000e+00
double.digits          5.300000e+01  5.300000e+01  5.300000e+01
double.rounding        5.000000e+00  5.000000e+00  5.000000e+00
double.guard            0.000000e+00  0.000000e+00  0.000000e+00
double.ulp.digits      -5.200000e+01  -5.200000e+01  -5.200000e+01
double.neg.ulp.digits  -5.300000e+01  -5.300000e+01  -5.300000e+01
double.exponent        1.100000e+01  1.100000e+01  1.100000e+01
double.min.exp        -1.022000e+03  -1.022000e+03  -1.022000e+03
double.max.exp          1.024000e+03  1.024000e+03  1.024000e+03
integer.max            2.147484e+09  2.147484e+09  2.147484e+09
sizeof.long            8.000000e+00  4.000000e+00  4.000000e+00
sizeof.longlong        8.000000e+00  8.000000e+00  8.000000e+00
sizeof.longdouble      1.600000e+01  1.200000e+01  1.600000e+01
sizeof.pointer          8.000000e+00  4.000000e+00  8.000000e+00
</pre>


tail(Cl(DJI))
=== NA when overflow ===
</syntaxhighlight>
<pre>
tmp <- 156287L
tmp*tmp
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647
</pre>


==== Timeline plot ====
== How to select a seed for simulation or randomization ==
https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
* [https://sciprincess.wordpress.com/2019/03/14/how-to-select-a-seed-for-simulation-or-randomization/ How to select a seed for simulation or randomization]
* [https://www.makeuseof.com/tag/lesson-gamers-rng/ What Is RNG? A Lesson for Gamers ]


=== Circular plot ===
== set.seed() allow alphanumeric seeds ==
* http://freakonometrics.hypotheses.org/20667 which uses https://cran.r-project.org/web/packages/circlize/ circlize] package.
https://stackoverflow.com/a/10913336
* https://www.biostars.org/p/17728/
* [https://cran.r-project.org/web/packages/RCircos/ RCircos] package from CRAN.
* [http://www.bioconductor.org/packages/release/bioc/html/OmicCircos.html OmicCircos] from Bioconductor.


=== Word cloud ===
== set.seed(), for loop and saving random seeds ==
* [http://www.sthda.com/english/wiki/text-mining-and-word-cloud-fundamentals-in-r-5-simple-steps-you-should-know Text mining and word cloud fundamentals in R : 5 simple steps you should know]
<ul>
* [https://www.displayr.com/alternatives-word-cloud/ 7 Alternatives to Word Clouds for Visualizing Long Lists of Data]
<li>[https://www.jottr.org/2020/09/21/detect-when-the-random-number-generator-was-used/ Detect When the Random Number Generator Was Used]
* [https://www.littlemissdata.com/blog/steam-data-art1 Data + Art STEAM Project: Initial Results]
<pre>
if (interactive()) {
  invisible(addTaskCallback(local({
    last <- .GlobalEnv$.Random.seed
   
    function(...) {
      curr <- .GlobalEnv$.Random.seed
      if (!identical(curr, last)) {
        msg <- "NOTE: .Random.seed changed"
        if (requireNamespace("crayon", quietly=TRUE)) msg <- crayon::blurred(msg)
        message(msg)
        last <<- curr
      }
      TRUE
    }
  }), name = "RNG tracker"))
}
</pre>
</li>
<li>http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
<pre>
set.seed(1001)
data <- vector("list", 30)
seeds <- vector("list", 30)
for(i in 1:30) {
  seeds[[i]] <- .Random.seed
  data[[i]] <- runif(5)
}
# If we save and load .Random.seed from a file using scan(), make
# sure to convert its type from doubles to integers.
# Otherwise, .Random.seed will complain!


=== World map ===
.Random.seed <- seeds[[23]]  # restore
[https://www.enchufa2.es/archives/visualising-ssh-attacks-with-r.html Visualising SSH attacks with R] ([https://cran.r-project.org/package=rworldmap rworldmap] and [https://cran.r-project.org/package=rgeolocate rgeolocate] packages)
data.23 <- runif(5)
data.23
data[[23]]
</pre>
</li>
</ul>
* [https://www.rdocumentation.org/packages/impute/versions/1.46.0/topics/impute.knn impute.knn]
* Duncan Murdoch: ''This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed.  See ?.Random.seed for details.''
* Uwe Ligges's comment: ''set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.''
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
* [http://www.questionflow.org/2019/08/13/local-randomness-in-r/ Local randomness in R].


=== Diagram/flowchart/Directed acyclic diagrams (DAGs) ===
== sample() ==
=== sample() inaccurate on very large populations, fixed in R 3.6.0 ===
* [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17494 The default method for generating from a discrete uniform distribution (used in ‘sample()’, for instance) has been changed]. In prior versions, the probability of generating each integer could vary from equal by up to 0.04% (or possibly more if generating more than a million different integers). See also [https://www.r-bloggers.com/whats-new-in-r-3-6-0/amp/ What's new in R 3.6.0] by David Smith.
{{Pre}}
# R 3.5.3
set.seed(123)
m <- (2/5)*2^32
m > 2^31
# [1] FALSE
log10(m)
# [1] 9.23502
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1
# 400070 599930
</pre>
* [https://blog.daqana.com/en/fast-sampling-support-in-dqrng/ Fast sampling support in dqrng]
* Differences of the output of sample()
{{Pre}}
# R 3.5.3
# docker run --net=host -it --rm r-base:3.5.3
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5


==== [https://cran.r-project.org/web/packages/DiagrammeR/index.html DiagrammeR] ====
# R 3.6.0
* http://rich-iannone.github.io/DiagrammeR/
# docker run --net=host -it --rm r-base:3.6.0
* https://donlelek.github.io/2015-03-31-dags-with-r/
> set.seed(1234)
> sample(5)
[1] 4 5 2 3 1
> RNGkind(sample.kind = "Rounding")
Warning message:
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5
</pre>


==== [https://cran.r-project.org/web/packages/diagram/ diagram] ====
=== Getting different results with set.seed() in RStudio ===
Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams
[https://community.rstudio.com/t/getting-different-results-with-set-seed/31624/2 Getting different results with set.seed()].  ''It's possible that you're loading an R package that is changing the requested random number generator; RNGkind().''


==== DAGitty (browser-based and R package) ====
=== dplyr::sample_n() ===
* http://dagitty.net/
The function has a parameter [https://dplyr.tidyverse.org/reference/sample.html weight]. For example if we have some download statistics for each day and we want to do sampling based on their download numbers, we can use this function.
* https://cran.r-project.org/web/packages/dagitty/index.html


==== dagR ====
== Regular Expression ==
* https://cran.r-project.org/web/packages/dagR
See [[Regular_expression|here]].


=== Venn Diagram ===
== Read rrd file ==
* limma http://www.ats.ucla.edu/stat/r/faq/venn.htm - only black and white?
* https://en.wikipedia.org/wiki/RRDtool
* VennDiagram - input has to be the numbers instead of the original vector?
* http://oss.oetiker.ch/rrdtool/
* http://manuals.bioinformatics.ucr.edu/home/R_BioCondManual#TOC-Venn-Diagrams and the [http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R R code] or the [http://www.bioconductor.org/packages/release/bioc/html/systemPipeR.html Bioc package systemPipeR]
* https://github.com/pldimitrov/Rrd
<syntaxhighlight lang='rsplus'>
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/
# systemPipeR package method
library(systemPipeR)
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))
OLlist <- overLapper(setlist[1:3], type="vennsets")
vennPlot(list(OLlist))                           


# R script source method
== on.exit() ==
source("http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/overLapper.R")  
Examples of using on.exit(). In all these examples, '''add = TRUE''' is used in the on.exit() call to ensure that each exit action is added to the list of actions to be performed when the function exits, rather than replacing the previous actions.
setlist <- list(A=sample(letters, 18), B=sample(letters, 16), C=sample(letters, 20), D=sample(letters, 22), E=sample(letters, 18))  
<ul>
# or (obtained by dput(setlist))
<li>Database connections
setlist <- structure(list(A = c("o", "h", "u", "p", "i", "s", "a", "w",
<pre>
"b", "z", "n", "c", "k", "j", "y", "m", "t", "q"), B = c("h",  
library(RSQLite)
"r", "x", "y", "b", "t", "d", "o", "m", "q", "g", "v", "c", "u",
sqlite_get_query <- function(db, sql) {
"f", "z"), C = c("b", "e", "t", "u", "s", "j", "o", "k", "d",
  conn <- dbConnect(RSQLite::SQLite(), db)
"l", "g", "i", "w", "n", "p", "a", "y", "x", "m", "z"), D = c("f",
  on.exit(dbDisconnect(conn), add = TRUE)
"g", "b", "k", "j", "m", "e", "q", "i", "d", "o", "l", "c", "t",
  dbGetQuery(conn, sql)
"x", "r", "s", "u", "w", "a", "z", "n"), E = c("u", "w", "o",
}
"k", "n", "h", "p", "z", "l", "m", "r", "d", "q", "s", "x", "b",
</pre>
"v", "t"), F = c("o", "j", "r", "c", "l", "l", "u", "b", "f",
<li>File connections
"d", "u", "m", "y", "t", "y", "s", "a", "g", "t", "m", "x", "m"
<pre>
)), .Names = c("A", "B", "C", "D", "E", "F"))
read_chars <- function(file_name) {
  conn <- file(file_name, "r")
  on.exit(close(conn), add = TRUE)
  readChar(conn, file.info(file_name)$size)
}
</pre>
<li>Temporary files
<pre>
history_lines <- function() {
  f <- tempfile()
  on.exit(unlink(f), add = TRUE)
  savehistory(f)
  readLines(f, encoding = "UTF-8")
}
</pre>
<li>Printing messages
<pre>
myfun = function(x) {
  on.exit(print("first"))
  on.exit(print("second"), add = TRUE)
  return(x)
}
</pre>
</ul>


OLlist <- overLapper(setlist[1:3], type="vennsets")
== file, connection ==
counts <- list(sapply(OLlist$Venn_List, length))
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/cat cat()] and [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/scan scan()] (read data into a vector or list from the console or file)
vennPlot(counts=counts)                          
* read() and write()
</syntaxhighlight>
* read.table() and write.table()
{{Pre}}
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)


[[File:Vennplot.png|250px]]
foo <- function() {
  con <- file()
  ...
  on.exit(close(con))
  ...
}
</pre>
[https://r.789695.n4.nabble.com/Why-I-get-this-error-Error-in-close-connection-f-invalid-connection-td904413.html Error in close.connection(f) : invalid connection]. If we want to use '''close(con)''', we have to specify how to '''open''' the connection; such as
<pre>
con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)
</pre>


=== Bump chart/Metro map ===
=== withr package ===
https://dominikkoch.github.io/Bump-Chart/
https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by [https://cran.r-project.org/web/packages/languageserver/index.html languageserver].


=== Amazing plots ===
== Clipboard (?connections), textConnection(), pipe() ==
==== New R logo 2/11/2016 ====
<ul>
* http://rud.is/b/2016/02/11/plot-the-new-svg-r-logo-with-ggplot2/
<li>On Windows, we can use readClipboard() and writeClipboard().
* https://www.stat.auckland.ac.nz/~paul/Reports/Rlogo/Rlogo.html
{{Pre}}
<syntaxhighlight lang='rsplus'>
source("clipboard")
library(sp)
read.table("clipboard")
library(maptools)
</pre></li>
library(ggplot2)
<li>Clipboard -> R. Reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function:
library(ggthemes)
{{Pre}}
# rgeos requires the installation of GEOS from http://trac.osgeo.org/geos/
x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
system("curl http://download.osgeo.org/geos/geos-3.5.0.tar.bz2 | tar jx")
# Or on Mac
system("cd geos-3.5.0; ./configure; make; sudo make install")
x <- read.delim(pipe("pbpaste"))
library(rgeos)
# safely ignore the warning: incomplete final line found by readTableHeader on 'pbpaste'
</pre>
r_wkt_gist_file <- "https://gist.githubusercontent.com/hrbrmstr/07d0ccf14c2ff109f55a/raw/db274a39b8f024468f8550d7aeaabb83c576f7ef/rlogo.wkt"
An example is to copy data from [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others?answertab=active#tab-top this post]. In this case we need to use read.table() instead of read.delim().
if (!file.exists("rlogo.wkt")) download.file(r_wkt_gist_file, "rlogo.wkt")
</li>
rlogo <- readWKT(paste0(readLines("rlogo.wkt", warn=FALSE))) # rgeos
<li>R -> clipboard on Mac. Note: '''pbcopy''' and '''pbpaste''' are macOS terminal commands. See [http://osxdaily.com/2007/03/05/manipulating-the-clipboard-from-the-command-line/ pbcopy & pbpaste: Manipulating the Clipboard from the Command Line].  
* pbcopy: takes standard input and places it in the clipboard buffer
rlogo_shp <- SpatialPolygonsDataFrame(rlogo, data.frame(poly=c("halo", "r"))) # sp
* pbpaste: takes data from the clipboard buffer and writes it to the standard output
rlogo_poly <- fortify(rlogo_shp, region="poly") # ggplot2
{{Pre}}
clip <- pipe("pbcopy", "w")
ggplot(rlogo_poly) +
write.table(apply(x, 1, mean), file = clip, row.names=F, col.names=F)
  geom_polygon(aes(x=long, y=lat, group=id, fill=id)) +
# write.table(data.frame(Var1, Var2), file = clip, row.names=F, quote=F, sep="\t")
  scale_fill_manual(values=c(halo="#b8babf", r="#1e63b5")) +
close(clip)
  coord_equal() +
</pre>
  theme_map() +
<li>
  theme(legend.position="none")
<li>Clipboard -> Excel.
</syntaxhighlight>
* Method 1: Paste icon -> Text import wizard -> Delimit (Tab, uncheck Space) or Fixed width depending on the situation -> Finish.
* Method 2: Ctrl+v first. Then choose Data -> Text to Columns. Fixed width -> Next -> Next -> Finish.
</li>
<li>On Linux, we need to install "xclip". See [https://stackoverflow.com/questions/45799496/r-copy-from-clipboard-in-ubuntu-linux R Copy from Clipboard in Ubuntu Linux]. It seems to work.
{{Pre}}
# sudo apt-get install xclip
read.table(pipe("xclip -selection clipboard -o",open="r"))
</pre>
</li>
</ul>


==== 3D plot ====
=== clipr ===
Using [https://chitchatr.wordpress.com/2010/06/28/fun-with-persp-function/ persp] function to create the following plot.
[https://cran.rstudio.com/web/packages/clipr/ clipr]: Read and Write from the System Clipboard


[[File:3dpersp.png|200px]]
== read/manipulate binary data ==
<syntaxhighlight lang='rsplus'>
* x <- readBin(fn, raw(), file.info(fn)$size)
### Random pattern
* rawToChar(x[1:16])
# Create matrix with random values with dimension of final grid
* See Biostrings C API
  rand <- rnorm(441, mean=0.3, sd=0.1)
  mat.rand <- matrix(rand, nrow=21)
# Create another matrix for the colors. Start by making all cells green
  fill <- matrix("green3", nr = 21, nc = 21)
# Change colors in each cell based on corresponding mat.rand value
  fcol <- fill
  fcol[] <- terrain.colors(40)[cut(mat.rand,
    stats::quantile(mat.rand, seq(0,1, len = 41),
    na.rm=T), include.lowest = TRUE)]
# Create concave surface using expontential function
  x <- -10:10
  y <- x^2
  y <- as.matrix(y)
  y1 <- y
  for(i in 1:20){tmp <- cbind(y,y1); y1 <- tmp[,1]; y <- tmp;}
  mat <- tmp[1:21, 1:21]
# Plot it up!
  persp(1:21, 1:21, t(mat)/10, theta = 90, phi = 35,col=fcol,
    scale = FALSE, axes = FALSE, box = FALSE)


### Organized pattern
== String Manipulation ==
# Same as before
* [https://www.gastonsanchez.com/r4strings/ Handling Strings with R](ebook) by Gaston Sanchez.
  rand <- rnorm(441, mean=0.3, sd=0.1)
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
# Create concave surface using expontential function
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
  x <- -10:10
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
  y <- x^2
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''
  y <- as.matrix(y)
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]
  for(i in 1:20){tmp <- cbind(y,y); y1 <- tmp[,1]; y <- tmp;}
* [https://statisticaloddsandends.wordpress.com/2022/05/05/a-short-note-on-the-startswith-function/ A short note on the startsWith function]
  mat <- tmp[1:21, 1:21]
###Organize rand by y and put into matrix form
  o <- order(rand,as.vector(mat))
  o.tmp <- cbind(rand[o], rev(sort(as.vector(mat))))
  mat.org <- matrix(o.tmp[,1], nrow=21)
  half.1 <- mat.org[,seq(1,21,2)]
  half.2 <- mat.org[,rev(seq(2,20,2))]
  full <- cbind(half.1, half.2)
  full <- t(full)
# Again, create color matrix and populate using rand values
zi <- full[-1, -1] + full[-1, -21] + full[-21,-1] + full[-21, -21]
fill <- matrix("green3", nr = 20, nc = 20)
fcol <- fill
fcol[] <- terrain.colors(40)[cut(zi,
        stats::quantile(zi, seq(0,1, len = 41), na.rm=T),
        include.lowest = TRUE)]
# Plot it up!       
persp(1:21, 1:21, t(mat)/10, theta = 90, phi = 35,col=t(fcol),
    scale = FALSE, axes = FALSE, box = FALSE)
</syntaxhighlight>


==== Christmas tree ====
=== format(): padding with zero ===
http://wiekvoet.blogspot.com/2014/12/merry-christmas.html
<pre>
<syntaxhighlight lang='rsplus'>
ngenes <- 10
# http://blogs.sas.com/content/iml/2012/12/14/a-fractal-christmas-tree/
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
# Each row is a 2x2 linear transformation
# [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
# Christmas tree
</pre>
L <- matrix(
    c(0.03, 0,     0  ,  0.1,
        0.85,  0.00,  0.00, 0.85,
        0.8,  0.00,  0.00, 0.8,
        0.2,  -0.08,  0.15, 0.22,
        -0.2,  0.08,  0.15, 0.22,
        0.25, -0.1,  0.12, 0.25,
        -0.2,  0.1,  0.12, 0.2),
    nrow=4)
# ... and each row is a translation vector
B <- matrix(
    c(0, 0,
        0, 1.5,
        0, 1.5,
        0, 0.85,
        0, 0.85,
        0, 0.3,
        0, 0.4),
    nrow=2)


prob = c(0.02, 0.6,.08, 0.07, 0.07, 0.07, 0.07)
=== noquote() ===
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/noquote noqute] Print character strings without quotes.


# Iterate the discrete stochastic map
=== stringr package ===
N = 1e5 #5  #  number of iterations
* https://stringr.tidyverse.org/index.html
x = matrix(NA,nrow=2,ncol=N)
* [https://stringr.tidyverse.org/articles/from-base.html Vignette compares stringr functions to their base R equivalents]
x[,1] = c(0,2)   # initial point
* When I try to use trimws() on data obtained from readxl::read_excell(), I find trimws() does not work but [https://stringr.tidyverse.org/reference/str_trim.html stringr::str_trim()] works. [https://stackoverflow.com/questions/45050617/trimws-bug-leading-whitespace-not-removed trimws bug? leading whitespace not removed].
k <- sample(1:7,N,prob,replace=TRUE) # values 1-7


for (i in 2:N)
=== glue package ===
  x[,i] = crossprod(matrix(L[,k[i]],nrow=2),x[,i-1]) + B[,k[i]] # iterate
<ul>
<li>[https://cran.r-project.org/web/packages/glue/index.html glue]. Useful in a loop and some function like ggtitle() or ggsave(). Inside the curly braces {R-Expression}, the expression is evaluated.
<syntaxhighlight lang='r'>
library(glue)
name <- "John"
age <- 30
glue("My name is {name} and I am {age} years old.")
# My name is John and I am 30 years old.


# Plot the iteration history
price <- 9.99
png('card.png')
quantity <- 3
par(bg='darkblue',mar=rep(0,4))  
total <- glue("The total cost is {round(price * quantity, 2)}.")
plot(x=x[1,],y=x[2,],
# Inside the curly braces {}, the expression round(price * quantity, 2) is evaluated.
    col=grep('green',colors(),value=TRUE),
print(total)
    axes=FALSE,
# The total cost is 29.97.
    cex=.1,
</syntaxhighlight>
    xlab='',
The syntax of glue() in R is quite similar to Python's print() function when using formatted strings. In Python, you typically use [https://www.pythontutorial.net/python-basics/python-f-strings/ f-strings] to embed variables inside strings.
    ylab='' )#,pch='.')
<syntaxhighlight lang='python'>
name = "John"
age = 30
print(f"My name is {name} and I am {age} years old.")
# My name is John and I am 30 years old.


bals <- sample(N,20)
price = 9.99
points(x=x[1,bals],y=x[2,bals]-.1,
quantity = 3
    col=c('red','blue','yellow','orange'),
total = f"The total cost is {price * quantity:.2f}."
    cex=2,
print(total)
    pch=19
# The total cost is 29.97.
)
text(x=-.7,y=8,
    labels='Merry',
    adj=c(.5,.5),
    srt=45,
    vfont=c('script','plain'),
    cex=3,
    col='gold'
)
text(x=0.7,y=8,
    labels='Christmas',
    adj=c(.5,.5),
    srt=-45,
    vfont=c('script','plain'),
    cex=3,
    col='gold'
)
</syntaxhighlight>
</syntaxhighlight>
[[File:XMastree.png|150px]]


==== Happy Thanksgiving ====
</li>
[http://blog.revolutionanalytics.com/2015/11/happy-thanksgiving.html Turkey]
<li>[https://en.wikipedia.org/wiki/String_interpolation String interpolation] </li>
</ul>


[[File:Turkey.png|150px]]
=== Raw data type ===
[https://twitter.com/hadleywickham/status/1387747735441395712 Fun with strings], [https://en.wikipedia.org/wiki/Cyrillic_alphabets Cyrillic alphabets]
<pre>
a1 <- "А"
a2 <- "A"
a1 == a2
# [1] FALSE
charToRaw("А")
# [1] d0 90
charToRaw("A")
# [1] 41
</pre>


==== Happy Valentine's Day ====
=== number of characters limit ===
* [https://rud.is/b/2017/02/14/geom%E2%9D%A4%EF%B8%8F/  Geom❤️] 2017
[https://twitter.com/eddelbuettel/status/1438326822635180036 It's a limit on a (single) input line in the REPL]
* [http://www.theanalyticslab.nl/2019/02/14/nerds-on-valentines-day/ Happy Valentines day by Nerds] 2019


==== treemap ====
=== Comparing strings to numeric ===
http://ipub.com/treemap/
[https://stackoverflow.com/a/57348393 ">" coerces the number to a string before comparing].
<syntaxhighlight lang='r' inline>"10" < 2 # TRUE</syntaxhighlight>


[[File:TreemapPop.png|150px]]
== HTTPs connection ==
HTTPS connection becomes default in R 3.2.2. See
* http://blog.rstudio.org/2015/08/17/secure-https-connections-for-r/
* http://blog.revolutionanalytics.com/2015/08/good-advice-for-security-with-r.html


==== [https://en.wikipedia.org/wiki/Voronoi_diagram Voronoi diagram] ====
[http://developer.r-project.org/blosxom.cgi/R-devel/2016/12/15#n2016-12-15 R 3.3.2 patched] The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)
* https://www.stat.auckland.ac.nz/~paul/Reports/VoronoiTreemap/voronoiTreeMap.html
* http://letstalkdata.com/2014/05/creating-voronoi-diagrams-with-ggplot/


==== Silent Night ====
== setInternet2 ==
[[File:Silentnight.png|200px]]
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.


<syntaxhighlight lang='rsplus'>
Read the [https://stat.ethz.ch/pipermail/r-devel/2015-August/071595.html discussion] reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in [https://stat.ethz.ch/pipermail/r-devel/2015-August/071623.html this post]. The following demonstrated the original problem.
# https://aschinchon.wordpress.com/2014/03/13/the-lonely-acacia-is-rocked-by-the-wind-of-the-african-night/
<pre>
depth <- 9
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
angle<-30 #Between branches division
              "GCF_000001405.13.assembly.txt")
L <- 0.90 #Decreasing rate of branches by depth
f1 <- tempfile()
nstars <- 300 #Number of stars to draw
download.file(url, f1)
mstars <- matrix(runif(2*nstars), ncol=2)
</pre>
branches <- rbind(c(1,0,0,abs(jitter(0)),1,jitter(5, amount = 5)), data.frame())
It seems the bug was fixed in R 3.2-branch. See [https://github.com/wch/r-source/commit/3a02ed3a50ba17d9a093b315bf5f31ffc0e21b89 8/16/2015] patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385098%28v=vs.85%29.aspx InternetOpenUrl()] function of [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385473%28v=vs.85%29.aspx wininet] library. [http://slacksite.com/other/ftp.html This article] and [http://stackoverflow.com/questions/1699145/what-is-the-difference-between-active-and-passive-ftp this post] explain differences of active and passive FTP.  
colnames(branches) <- c("depth", "x1", "y1", "x2", "y2", "inertia")
for(i in 1:depth)
{
  df <- branches[branches$depth==i,]
  for(j in 1:nrow(df))
  {
    branches <- rbind(branches, c(df[j,1]+1, df[j,4], df[j,5], df[j,4]+L^(2*i+1)*sin(pi*(df[j,6]+angle)/180),
                                  df[j,5]+L^(2*i+1)*cos(pi*(df[j,6]+angle)/180), df[j,6]+angle+jitter(10, amount = 8)))
    branches <- rbind(branches, c(df[j,1]+1, df[j,4], df[j,5], df[j,4]+L^(2*i+1)*sin(pi*(df[j,6]-angle)/180),
                                  df[j,5]+L^(2*i+1)*cos(pi*(df[j,6]-angle)/180), df[j,6]-angle+jitter(10, amount = 8)))
  }
}
nodes <- rbind(as.matrix(branches[,2:3]), as.matrix(branches[,4:5]))
png("image.png", width = 1200, height = 600)
plot.new()
par(mai = rep(0, 4), bg = "gray12")
plot(nodes, type="n", xlim=c(-7, 3), ylim=c(0, 5))
for (i in 1:nrow(mstars))  
{
  points(x=10*mstars[i,1]-7, y=5*mstars[i,2], col = "blue4", cex=.7, pch=16)
  points(x=10*mstars[i,1]-7, y=5*mstars[i,2], col = "blue",  cex=.3, pch=16)
  points(x=10*mstars[i,1]-7, y=5*mstars[i,2], col = "white", cex=.1, pch=16)
}
# The moon
points(x=-5, y=3.5, cex=40, pch=16, col="lightyellow")
# The tree
for (i in 1:nrow(branches)) {
  lines(x=branches[i,c(2,4)], y=branches[i,c(3,5)],
    col = paste("gray", as.character(sample(seq(from=50, to=round(50+5*branches[i,1]), by=1), 1)), sep = ""),
    lwd=(65/(1+3*branches[i,1])))
}
rm(branches)
dev.off()
</syntaxhighlight>


==== The Travelling Salesman Portrait ====
The following R command will show the exact svn revision for the R you are currently using.
https://fronkonstin.com/2018/04/04/the-travelling-salesman-portrait/
<pre>
R.Version()$"svn rev"
</pre>


==== Moon phase calendar ====
If setInternet2(T), then https protocol is supported in download.file().  
https://chichacha.netlify.com/2018/05/26/making-calendar-with-ggplot-moon-phase-calendar/


==== Chaos ====
When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.
[https://fronkonstin.com/2019/01/10/rcpp-camaron-de-la-isla-and-the-beauty-of-maths/ Rcpp, Camarón de la Isla and the Beauty of Maths]


=== Google Analytics ===
The setInternet2() function is defined in [https://github.com/wch/r-source/commits/trunk/src/library/utils/R/windows/sysutils.R R> src> library> utils > R > windows > sysutils.R].
==== GAR package ====
http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html


=== Linear Programming ===
'''R up to 3.2.2'''
http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/
<pre>
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
</pre>
See also
* <src/include/Internal.h> (declare do_setInternet2()),
* <src/main/names.c> (show do_setInternet2() in C)
* <src/main/internet.c>  (define do_setInternet2() in C).


=== Amazon Alexa ===
Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file.  If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2,  it is also possible to download from https by setIneternet2(T).
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html


=== R and Singularity ===
'''R 3.3.0'''
https://www.rstudio.com/rviews/2017/03/29/r-and-singularity/
<pre>
setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}
</pre>


=== Teach kids about R with Minecraft ===
Note that setInternet2.Rd says As from \R 3.3.0 it changes nothing, and only \code{use = NA} is accepted. Also NEWS.Rd says setInternet2() has no effect and will be removed in due course.
http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html


=== Secure API keys ===
== Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan() ==
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, '''Inf''' and '''NaN''' as input or output.


=== Vision and image recognition ===
See [https://stat.ethz.ch/R-manual/R-devel/library/base/html/is.finite.html ?is.finite].
* https://www.stoltzmaniac.com/google-vision-api-in-r-rooglevision/ Google vision API IN R] – RoogleVision
* [http://www.bnosac.be/index.php/blog/66-computer-vision-algorithms-for-r-users Computer Vision Algorithms for R users] and https://github.com/bnosac/image


=== Turn pictures into coloring pages ===
[https://datasciencetut.com/how-to-replace-inf-values-with-na-in-r/ How to replace Inf with NA in All or Specific Columns of the Data Frame]
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0


=== Numerical optimization ===
== replace() function ==
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/uniroot.html uniroot]: One Dimensional Root (Zero) Finding. This is used in [http://onlinelibrary.wiley.com/doi/10.1002/sim.7178/full simulating survival data for predefined censoring rate]
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/replace replace](vector, index, values)
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
* https://stackoverflow.com/a/11811147
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optim.html optim]: General-purpose optimization based on Nelder–Mead, quasi-Newton and conjugate-gradient algorithms.
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/constrOptim.html constrOptim]: Linearly Constrained Optimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nlm.html nlm]: Non-Linear Minimization
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/nls.html nls]: Nonlinear Least Squares


== R packages ==
== File/path operations ==
[[R_packages|R packages]]
* list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
* file.info()
* dir.create()
* file.create()
* file.copy()
* file.exists()
<ul>
<li>'''basename'''() - remove the parent path, '''dirname'''() - returns the part of the path up to but excluding the last path separator
<pre>
> file.path("~", "Downloads")
[1] "~/Downloads"
> dirname(file.path("~", "Downloads"))
[1] "/home/brb"
> basename(file.path("~", "Downloads"))
[1] "Downloads"
</pre>
</li></ul>
* '''path.expand'''("~/.Renviron")  # "/home/brb/.Renviron"
<ul>
<li> '''normalizePath'''() # Express File Paths in Canonical Form
<pre>
> cat(normalizePath(c(R.home(), tempdir())), sep = "\n")
/usr/lib/R
/tmp/RtmpzvDhAe
</pre>
</li>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/system.file system.file()] - Finds the full file names of files in packages etc
<pre>
> system.file("extdata", "ex1.bam", package="Rsamtools")
[1] "/home/brb/R/x86_64-pc-linux-gnu-library/4.0/Rsamtools/extdata/ex1.bam"
</pre>
</li></ul>
* tools::file_path_sans_ext() - [https://stackoverflow.com/a/29114021 remove the file extension] or the sub() function.


== Tricks ==
== read/download/source a file from internet ==
=== Simple text file http ===
<pre>
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
</pre>


=== Getting help ===
=== Zip, RData, gz file and url() function ===
* http://stackoverflow.com/questions/tagged/r and [https://stackoverflow.com/tags/r/info R page] contains resources.  
<pre>
* https://stat.ethz.ch/pipermail/r-help/
x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
* https://stat.ethz.ch/pipermail/r-devel/
</pre>
<pre>
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
source(con)
close(con)
</pre>
Here url() function is like file(),  gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.


=== Better Coder/coding, best practices ===
Another example is [https://stackoverflow.com/a/9548672 Read gzipped csv directly from a url in R]
* http://www.mango-solutions.com/wp/2015/10/10-top-tips-for-becoming-a-better-coder/
<pre>
* [https://www.rstudio.com/rviews/2016/12/02/writing-good-r-code-and-writing-well/ Writing Good R Code and Writing Well]
con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
* [http://www.thertrader.com/2018/09/01/r-code-best-practices/ R Code – Best practices]
                      "aa_combined-20110321.csv.gz", sep="")))
txt <- readLines(con)
dat <- read.csv(textConnection(txt))
</pre>


=== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ===
Another example of using url() is
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
<pre>
load(url("http:/www.example.com/example.RData"))
</pre>
 
This does not work with load(), dget(), read.table() for files on '''OneDrive'''. In fact, I cannot use wget with shared files from OneDrive. The following trick works: [https://mangolassi.it/topic/19276/how-to-configure-a-onedrive-file-for-use-with-wget How to configure a OneDrive file for use with wget].


=== Change default R repository ===
'''Dropbox''' is easy and works for load(), wget, ...
Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local .Rprofile settings take precedence.


For example, I can specify the R mirror I like by creating a single line <.Rprofile> file under my home directory.
[https://stackoverflow.com/a/46875562 R download .RData] or [https://stackoverflow.com/a/56670130 Directly loading .RData from github] from Github.
 
=== zip function ===
This will include 'hallmarkFiles' root folder in the files inside zip.
<pre>
<pre>
local({
zip(zipfile = 'myFile.zip',
  r = getOption("repos")
    files = dir('hallmarkFiles', full.names = TRUE))
  r["CRAN"] = "https://cran.rstudio.com/"
 
  options(repos = r)
# Verify/view the files. 'list = TRUE' won't extract
})
unzip('testZip.zip', list = TRUE)  
options(continue = "  ")
</pre>
message("Hi MC, loading ~/.Rprofile")
 
if (interactive()) {
=== [http://cran.r-project.org/web/packages/downloader/index.html downloader] package ===
  .Last <- function() try(savehistory("~/.Rhistory"))
This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.
}


=== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
{{Pre}}
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))
</pre>
</pre>


=== Change the default web browser ===
=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey).
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]
<syntaxhighlight lang='rsplus'>
> help.start()
If the browser launched by 'xdg-open' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...
> /usr/bin/xdg-open: 461: /usr/bin/xdg-open: x-www-browser: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: firefox: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: mozilla: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: epiphany: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: konqueror: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: chromium-browser: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: google-chrome: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links2: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: links: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: lynx: not found
/usr/bin/xdg-open: 461: /usr/bin/xdg-open: w3m: not found
xdg-open: no method available for opening 'http://127.0.0.1:27919/doc/html/index.html'
</syntaxhighlight>


The solution is to put
=== Github files https using RCurl package ===
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
<pre>
<pre>
options(browser='seamonkey')
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
</pre>
in the '''.Rprofile''' of your home directory. If the browser is not in the global PATH, we need to put the full path above.
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package
 
== data summary table ==
=== summarytools: create summary tables for vectors and data frames ===
https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and data frames.


For one-time only purpose, we can use the ''browser'' option in help.start() function:
=== skimr: A frictionless, pipeable approach to dealing with summary statistics ===
<syntaxhighlight lang='rsplus'>
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]
> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...
</syntaxhighlight>


We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See
=== modelsummary ===
* [https://stat.ethz.ch/pipermail/r-help/2003-August/037484.html Changing default browser in options()].
[https://cloud.r-project.org/web/packages/modelsummary/index.html modelsummary]: Summary Tables and Plots for Statistical Models and Data: Beautiful, Customizable, and Publication-Ready
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/browseURL.html
 
=== broom ===
[[Tidyverse#broom|Tidyverse->broom]]


=== Rconsole, Rprofile.site, Renviron.site files ===
=== Create publication tables using '''tables''' package ===
* https://cran.r-project.org/doc/manuals/r-release/R-admin.html ('''Rprofile.site''')
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]
* https://cran.r-project.org/doc/manuals/r-release/R-intro.html ('''Rprofile.site, Renviron.site, Rconsole''' (Windows only))
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html  ('''Renviron.site''')
* [http://blog.revolutionanalytics.com/2015/11/how-to-store-and-use-authentication-details-with-r.html How to store and use webservice keys and authentication details]
* [http://itsalocke.com/use-rprofile-give-important-notifications/ Use your .Rprofile to give you important notifications]


If we like to install R packages to a personal directory, follow [https://stat.ethz.ch/pipermail/r-devel/2015-July/071562.html this]. Just add the line
R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,
<pre>
{{Pre}}
R_LIBS_SITE=F:/R/library
> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
                                                 
                Sepal.Length      Sepal.Width   
Species    n  mean        sd  mean        sd 
setosa      50 5.01        0.35 3.43        0.38
versicolor  50 5.94        0.52 2.77        0.31
virginica  50 6.59        0.64 2.97        0.32
All        150 5.84        0.83 3.06        0.44
> str(iris)
'data.frame':  150 obs. of  5 variables:
$ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
$ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
$ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
$ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
$ Species    : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...
</pre>
</pre>
to the file '''R_HOME/etc/x64/Renviron.site'''.
and
 
Note that on Windows OS, R/etc contains
<pre>
<pre>
$ ls -l /c/Progra~1/r/r-3.2.0/etc
# This example shows some of the less common options       
total 142
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
-rw-r--r--    1  Administ    1043 Jun 20  2013 Rcmd_environ
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
-rw-r--r--    1  Administ    1924 Mar 17  2010 Rconsole
> z <- rnorm(100)+5
-rw-r--r--    1   Administ      943 Oct  3  2011 Rdevga
> fmt <- function(x) {
-rw-r--r--    1  Administ      589 May 20  2013 Rprofile.site
   s <- format(x, digits=2)
-rw-r--r--    1  Administ   251894 Jan 17  2015 curl-ca-bundle.crt
   even <- ((1:length(s)) %% 2) == 0
drwxr-xr-x    1   Administ        0 Jun  8 10:30 i386
   s[even] <- sprintf("(%s)", s[even])
-rw-r--r--    1   Administ    1160 Dec 31  2014 repositories
   s
-rw-r--r--    1  Administ    30188 Mar 17  2010 rgb.txt
}
drwxr-xr-x    3   Administ        0 Jun  8 10:30 x64
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
 
                  Status             
$ ls /c/Progra~1/r/r-3.2.0/etc/i386
Sex    Statistic high  low    medium
Makeconf
Female mean      4.88  4.96  5.17
 
        sd        (1.20) (0.82) (1.35)
$ cat /c/Progra~1/r/r-3.2.0/etc/Rconsole
Male  mean      4.45  4.31   5.05
# Optional parameters for the console and the pager
        sd        (1.01) (0.93) (0.75)
# The system-wide copy is in R_HOME/etc.
</pre>
# A user copy can be installed in `R_USER'.
 
## Style
# This can be `yes' (for MDI) or `no' (for SDI).
   MDI = yes
# MDI = no
 
# the next two are only relevant for MDI
toolbar = yes
statusbar = no
 
## Font.
# Please use only fixed width font.
# If font=FixedFont the system fixed font is used; in this case
# points and style are ignored. If font begins with "TT ", only
# True Type fonts are searched for.
font = TT Courier New
points = 10
style = normal # Style can be normal, bold, italic


# Dimensions (in characters) of the console.
=== fgsea example ===
rows = 25
[http://www.bioconductor.org/packages/release/bioc/vignettes/fgsea/inst/doc/fgsea-tutorial.html  vignette] & [https://github.com/ctlab/fgsea/blob/master/R/plot.R#L28 source code]
columns = 80
# Dimensions (in characters) of the internal pager.
pgrows = 25
pgcolumns = 80
# should options(width=) be set to the console width?
setwidthonresize = yes


# memory limits for the console scrolling buffer, in chars and lines
=== (archived) ClinReport: Statistical Reporting in Clinical Trials ===
# NB: bufbytes is in bytes for R < 2.7.0, chars thereafter.
https://cran.r-project.org/web/packages/ClinReport/index.html
bufbytes = 250000
buflines = 8000


# Initial position of the console (pixels, relative to the workspace for MDI)
== Append figures to PDF files ==
# xconsole = 0
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
# yconsole = 0


# Dimension of MDI frame in pixels
== Save base graphics as pseudo-objects ==
# Format (w*h+xorg+yorg) or use -ve w and h for offsets from right bottom
[https://www.andrewheiss.com/blog/2016/12/08/save-base-graphics-as-pseudo-objects-in-r/ Save base graphics as pseudo-objects in R]. Note there are some cons with this approach.
# This will come up maximized if w==0
<pre>
# MDIsize = 0*0+0+0
pdf(NULL)
# MDIsize = 1000*800+100+0
dev.control(displaylist="enable")
# MDIsize = -50*-50+50+50  # 50 pixels space all round
plot(df$x, df$y)
text(40, 0, "Random")
text(60, 2, "Text")
lines(stats::lowess(df$x, df$y))
p1.base <- recordPlot()
invisible(dev.off())


# The internal pager can displays help in a single window
# Display the saved plot
# or in multiple windows (one for each topic)
grid::grid.newpage()
# pagerstyle can be set to `singlewindow' or `multiplewindows'
p1.base
pagerstyle = multiplewindows
</pre>


## Colours for console and pager(s)
== Extracting tables from PDFs ==
# (see rwxxxx/etc/rgb.txt for the known colours).
<ul>
background = White
<li>[http://datascienceplus.com/extracting-tables-from-pdfs-in-r-using-the-tabulizer-package/ extracting Tables from PDFs in R] using Tabulizer. This needs the [https://cran.r-project.org/web/packages/rJava/index.html rJava] package. Linux works fine. Some issue came out on my macOS 10.12 Sierra. '''Library not loaded: /Library/Java/JavaVirtualMachines/jdk-9.jdk/Contents/Home/lib/server/libjvm.dylib. Referenced from: /Users/XXXXXXX/Library/R/3.5/library/rJava/libs/rJava.so'''.
normaltext = NavyBlue
</li>
usertext = Red
<li>
highlight = DarkRed
[https://docs.ropensci.org/pdftools/ pdftools] - Text Extraction, Rendering and Converting of PDF Documents. [https://ropensci.org/technotes/2018/12/14/pdftools-20/ pdf_text() and pdf_data()] functions.
{{Pre}}
library(pdftools)
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
txt <- pdf_text(pdf_file) # length = number of pages
# Suppose the table we are interested in is on page 1
cat(txt[1]) # Good but not in a data frame format


## Initial position of the graphics window
pdf_data(pdf_file)[[1]]  # data frame/tibble format
## (pixels, <0 values from opposite edge)
</pre>
xgraphics = -25
However, it seems it does not work on [http://www.bloodjournal.org/content/109/8/3177/tab-figures-only Table S6]. Tabulizer package is better at this case.
ygraphics = 0
 
## Language for messages
language =


## Default setting for console buffering: 'yes' or 'no'
This is another example. [https://mp.weixin.qq.com/s?__biz=MzAxMDkxODM1Ng==&mid=2247490327&idx=1&sn=cca7d4423426318e0c23adb098cf0ad7&chksm=9b485bacac3fd2ba2196b380c59b5eab9d29795d3334b040f50a2fa58124ec6e3be9472829e0&scene=21#wechat_redirect 神技能-自动化批量从PDF里面提取表格]
buffered = yes
</li>
<li>[https://www.linuxuprising.com/2019/05/how-to-convert-pdf-to-text-on-linux-gui.html?m=1 How To Convert PDF To Text On Linux (GUI And Command Line)]. It works when I tested my PDF file.
{{Pre}}
sudo apt install poppler-utils
pdftotext -layout input.pdf output.txt
pdftotext -layout -f 3 -l 4 input.pdf output.txt # from page 3 to 4.
</pre>
</pre>
and on Linux
</li>
<pre>
<li>[https://www.adobe.com/acrobat/how-to/pdf-to-excel-xlsx-converter.html Convert PDF files into Excel spreadsheets] using Adobe Acrobat. See [https://helpx.adobe.com/acrobat/how-to/extract-pages-from-pdf.html How to extract pages from a PDF]. Note the PDF file should not be opened by Excel since it is binary format Excel can't recognize.
brb@brb-T3500:~$ whereis R
<li>I found it is easier to use copy the column (it works) from PDF and paste them to Excel </li>
R: /usr/bin/R /etc/R /usr/lib/R /usr/bin/X11/R /usr/local/lib/R /usr/share/R /usr/share/man/man1/R.1.gz
<li>[https://www.r-bloggers.com/2024/04/tabulapdf-extract-tables-from-pdf-documents/ tabulapdf: Extract Tables from PDF Documents]
</ul>


brb@brb-T3500:~$ ls /usr/lib/R
== Print tables ==
bin  COPYING  etc  lib  library  modules  site-library  SVN-REVISION


brb@brb-T3500:~$ ls /usr/lib/R/etc
=== addmargins() ===
javaconf  ldpaths  Makeconf  Renviron  Renviron.orig  Renviron.site  Renviron.ucf  repositories  Rprofile.site
* [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins]. Puts Arbitrary Margins On Multidimensional Tables Or Arrays.
* [https://datasciencetut.com/how-to-put-margins-on-tables-or-arrays-in-r/ How to put margins on tables or arrays in R?]


brb@brb-T3500:~$ ls /usr/local/lib/R
=== tableone ===
site-library
* https://cran.r-project.org/web/packages/tableone/
</pre>
* [https://datascienceplus.com/table-1-and-the-characteristics-of-study-population/ Table 1 and the Characteristics of Study Population]
and  
* [https://www.jianshu.com/p/e76f2b708d45 如何快速绘制论文的表1(基本特征三线表)?]
<pre>
* See Table 1 from [https://boiled-data.github.io/ClassificationDiabetes.html Tidymodels Machine Learning: Diabetes Classification]
brb@brb-T3500:~$ cat /usr/lib/R/etc/Rprofile.site
##                                              Emacs please make this -*- R -*-
## empty Rprofile.site for R on Debian
##
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Rprofile and Rprofile.site


# ## Example of .Rprofile
=== Some examples ===
# options(width=65, digits=5)
Cox models
# options(show.signif.stars=FALSE)
* [https://aacrjournals.org/clincancerres/article/27/12/3383/671420/Integrative-Genomic-Analysis-of-Gemcitabine Integrative Genomic Analysis of Gemcitabine Resistance in Pancreatic Cancer by Patient-derived Xenograft Models]
# setHook(packageEvent("grDevices", "onLoad"),
#        function(...) grDevices::ps.options(horizontal=FALSE))
# set.seed(1234)
# .First <- function() cat("\n  Welcome to R!\n\n")
# .Last <- function()  cat("\n  Goodbye!\n\n")


# ## Example of Rprofile.site
=== finalfit package ===
# local({
* https://cran.r-project.org/web/packages/finalfit/index.html. Lots of vignettes.
#  # add MASS to the default packages, set a CRAN mirror
** [https://cran.r-project.org/web/packages/finalfit/vignettes/survival.html Survival]. It fits both univariate and multivariate regressions and reports the results for both of them.
#  old <- getOption("defaultPackages"); r <- getOption("repos")
* [https://finalfit.org/index.html summary_factorlist()] from the finalfit package.
#  r["CRAN"] <- "http://my.local.cran"
* [https://www.r-bloggers.com/2018/05/elegant-regression-results-tables-and-plots-in-r-the-finalfit-package/ Elegant regression results tables and plots in R: the finalfit package]
#  options(defaultPackages = c(old, "MASS"), repos = r)
#})
brb@brb-T3500:~$ cat /usr/lib/R/etc/Renviron.site
##                                              Emacs please make this -*- R -*-
## empty Renviron.site for R on Debian
##
## Copyright (C) 2008 Dirk Eddelbuettel and GPL'ed
##
## see help(Startup) for documentation on ~/.Renviron and Renviron.site


# ## Example ~/.Renviron on Unix
=== table1 ===
# R_LIBS=~/R/library
* https://cran.r-project.org/web/packages/table1/
# PAGER=/usr/local/bin/less
* [https://www.rdatagen.net/post/2023-09-26-nice-looking-table-1-with-standardized-mean-difference/ Creating a nice looking Table 1 with standardized mean differences (SMD)]. SMD is the difference in group means divided by the pooled standard deviation (and is defined differently for categorical measures). Note that the pooled standard deviation defined here is different from we see on the '''[[T-test#Two_sample_test_assuming_equal_variance|t.test]]''' when we assume equivalent variance in two samples.


# ## Example .Renviron on Windows
=== gtsummary ===
# R_LIBS=C:/R/library
* [https://education.rstudio.com/blog/2020/07/gtsummary/ Presentation-Ready Summary Tables with gtsummary]
# MY_TCLTK="c:/Program Files/Tcl/bin"
* [https://www.danieldsjoberg.com/gtsummary/ gtsummary] & on [https://cloud.r-project.org/web/packages/gtsummary/index.html CRAN]
** [https://www.danieldsjoberg.com/gtsummary/articles/tbl_summary.html tbl_summary()]. The output is in the "Viewer" window.
* An example: [https://boiled-data.github.io/ClassificationDiabetes.html Tidymodels Machine Learning: Diabetes Classification]. The table is saved in a png file. The column variable is response.


# ## Example of setting R_DEFAULT_PACKAGES (from R CMD check)
=== gt* ===
# R_DEFAULT_PACKAGES='utils,grDevices,graphics,stats'
* [https://cran.r-project.org/web/packages/gt/index.html gt]: Easily Create Presentation-Ready Display Tables
# # this loads the packages in the order given, so they appear on
* [https://www.r-bloggers.com/2024/02/introduction-to-clinical-tables-with-the-gt-package/ Introduction to Clinical Tables with the {gt} Package]
# # the search path in reverse order.
* [https://www.youtube.com/watch?v=qFOFMed18T4 Add any Plot to your {gt} table]
brb@brb-T3500:~$
</pre>


==== What is the best place to save Rconsole on Windows platform ====
=== dplyr ===
Put/create the file <Rconsole> under ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
https://stackoverflow.com/a/34587522. The output includes counts and proportions in a publication like fashion.


My preferred settings:
=== tables::tabular() ===
* Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
* Size: 12
* background: black
* normaltext: white
* usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow


and others (default options)
=== gmodels::CrossTable() ===
* pagebg: white
https://www.statmethods.net/stats/frequencies.html
* pagetext: navy
* highlight: DarkRed
* dataeditbg: white
* dataedittext: navy (View() function)
* dataedituser: red
* editorbg: white (edit() function)
* editortext: black


=== Saving and loading history automatically: .Rprofile & local() ===
=== base::prop.table(x, margin) ===
* http://stat.ethz.ch/R-manual/R-patched/library/utils/html/savehistory.html
[http://developer.r-project.org/blosxom.cgi/R-devel/2020/02/13#n2020-02-13 New function ‘proportions()’ and ‘marginSums()’. These should replace the unfortunately named ‘prop.table()’ and ‘margin.table()’.] for R 4.0.0.
* '''.Rprofile''' will automatically be loaded when R has started from that directory
* '''.Rprofile''' has been created/used by the '''packrat''' package to restore a packrat environment. See the packrat/init.R file.
* [http://www.statmethods.net/interface/customizing.html Customizing Startup] from R in Action, [http://www.onthelambda.com/2014/09/17/fun-with-rprofile-and-customizing-r-startup/ Fun with .Rprofile and customizing R startup]
** You can also place a '''.Rprofile''' file in any directory that you are going to run R from or in the user home directory.  
** At startup, R will source the '''Rprofile.site''' file. It will then look for a '''.Rprofile''' file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
<pre>
<pre>
options(continue="  ") # default is "+ "
R> m <- matrix(1:4, 2)
options(editor="nano") # default is "vi" on Linux
R> prop.table(m, 1) # row percentage
# options(htmlhelp=TRUE)
          [,1]      [,2]
 
[1,] 0.2500000 0.7500000
local((r <- getOption("repos")
[2,] 0.3333333 0.6666667
  r["CRAN"] <- "http://cran.rstudio.com"
R> prop.table(m, 2) # column percentage
  options(repos = r)))
          [,1]      [,2]
 
[1,] 0.3333333 0.4285714
.First <- function(){
[2,] 0.6666667 0.5714286
# library(Hmisc)
cat("\nWelcome at", date(), "\n")
}
 
.Last <- function(){
cat("\nGoodbye at ", date(), "\n")
</pre>
* https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
* The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define '''R_HISTFILE''' system variable.
* [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/eval local()] function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)
 
'''Linux''' or '''Mac'''
 
In '''~/.profile''' or '''~/.bashrc''' I put:
<pre>
export R_HISTFILE=~/.Rhistory
</pre>
In '''~/.Rprofile''' I put:
<pre>
if (interactive()) {
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory"))
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>
</pre>


'''Windows'''
=== stats::xtabs() ===


If you launch R by clicking its icon from Windows Desktop, the R starts in '''C:\User\$USER\Documents''' directory. So we can create a new file '''.Rprofile''' in this directory.
=== stats::ftable() ===
<pre>
{{Pre}}
if (interactive()) {
> ftable(Titanic, row.vars = 1:3)
   .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
                  Survived  No Yes
}
Class Sex    Age                 
</pre>
1st  Male  Child            0  5
 
            Adult          118  57
=== R release versions ===
      Female Child            0  1
[http://cran.r-project.org/web/packages/rversions/index.html rversions]: Query the main 'R' 'SVN' repository to find the released versions & dates.
            Adult            4 140
 
2nd  Male  Child            0  11
=== Detect number of running R instances in Windows ===
            Adult          154  14
* http://stackoverflow.com/questions/15935931/detect-number-of-running-r-instances-in-windows-within-r
      Female Child            0  13
<pre>
            Adult          13  80
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
3rd  Male  Child          35  13
INFO: No tasks are running which match the specified criteria.
            Adult          387  75
 
      Female Child          17  14
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"
            Adult          89  76
 
Crew  Male  Child            0  0
Image Name                    PID Session Name        Session#   Mem Usage
            Adult          670 192
========================= ======== ================ =========== ============
      Female Child            0   0
Rgui.exe                      1096 Console                    1    44,712 K
            Adult            3  20
 
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"
            Survived  No Yes
 
Class Sex                   
Image Name                    PID Session Name       Session#    Mem Usage
1st  Male            118  62
========================= ======== ================ =========== ============
      Female            4 141
Rserve.exe                    6108 Console                    1    381,796 K
2nd  Male            154  25
</pre>
      Female          13  93
In R, we can use
3rd  Male            422  88
<pre>
      Female          106  90
> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
Crew  Male            670 192
[1] ""                                                                          
      Female            3  20
[2] "Image Name                    PID Session Name        Session#    Mem Usage"
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
[3] "========================= ======== ================ =========== ============"
            Survived  No Yes
[4] "Rgui.exe                      1096 Console                    1    44,804 K"
Sex   Class               
 
Male  1st            118  62
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
      2nd            154  25
</pre>
      3rd            422  88
 
      Crew          670 192
=== Editor ===
Female 1st              4 141
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
       2nd            13  93
 
      3rd            106  90
* Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
      Crew            3  20
* [http://www.rstudio.com/ Rstudio] - editor/R terminal/R graphics/file browser/package manager. The new version (0.98) also provides a new feature for debugging step-by-step. See also [https://www.rstudio.com/rviews/2016/11/11/easy-tricks-you-mightve-missed/ RStudio Tricks]
> str(Titanic)
* [http://www.geany.org/ geany] - I like the feature that it shows defined functions on the side panel even for R code. RStudio can also do this (see the bottom of the code panel).
table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
* [http://rgedit.sourceforge.net/ Rgedit] which includes a feature of splitting screen into two panes and run R in the bottom panel. See [http://www.stattler.com/article/using-gedit-or-rgedit-r here].
- attr(*, "dimnames")=List of 4
* Komodo IDE with browser preview http://www.youtube.com/watch?v=wv89OOw9roI at 4:06 and http://docs.activestate.com/komodo/4.4/editor.html
  ..$ Class  : chr [1:4] "1st" "2nd" "3rd" "Crew"
 
  ..$ Sex    : chr [1:2] "Male" "Female"
=== GUI for Data Analysis ===
  ..$ Age    : chr [1:2] "Child" "Adult"
 
  ..$ Survived: chr [1:2] "No" "Yes"
==== Rcmdr ====
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
http://cran.r-project.org/web/packages/Rcmdr/index.html
> x
 
          gear  3  4  5
==== Deducer ====
cyl vs am             
http://cran.r-project.org/web/packages/Deducer/index.html
4  0  0        0  0  0
 
      1        0  0  1
==== jamovi ====
    1  0        1  2  0
* https://www.jamovi.org/
      1        0  6  1
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
6  0  0        0  0  0
 
      1        0  2  1
=== Scope ===
    1  0        2  2  0
See
      1        0  0  0
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
8  0  0      12  0  0
* [[#How_to_exit_a_sourced_R_script|source()]] does not work like C's preprocessor where statements in header files will be literally inserted into the code. It does not work when you define a variable in a function but want to use it outside the function (even through '''source()''')
      1        0  0  2
    1  0        0  0  0
      1        0  0  0
> ftable(x, row.vars = c(2, 4))
        cyl  4    6    8 
        am  0  1  0  1  0  1
vs gear                     
0  3        0  0  0  0 12  0
  4        0  0  0  2  0  0
  5        0  1  0  1  0  2
1  3        1  0  2  0  0  0
  4        2  6  2  0  0  0
  5        0  1  0  0  0  0
>
> ## Start with expressions, use table()'s "dnn" to change labels
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
        dnn = c("Cylinders", "V/S", "Transmission", "Gears"))


<syntaxhighlight lang='rsplus'>
          Cylinders    4    6    8 
## foo.R ##
          Transmission  0  1  0  1  0  1
cat(ArrayTools, "\n")
V/S Gears                             
## End of foo.R
0  3                  0  0  0  0 12  0
 
    4                  0  0  0  2  0  0
# 1. Error
    5                  0  1  0  1  0  2
predict <- function() {
1  3                  1  0  2  0  0  0
  ArrayTools <- "C:/Program Files" # or through load() function
    4                  2  6  2  0  0  0
  source("foo.R")                  # or through a function call; foo()
    5                  0  1  0  0  0  0
}
</pre>
predict()   # Object ArrayTools not found
 
== tracemem, data type, copy ==
[http://stackoverflow.com/questions/18359940/r-programming-vector-a1-2-avoid-copying-the-whole-vector/18361181#18361181 How to avoid copying a long vector]
 
== Tell if the current R is running in 32-bit or 64-bit mode ==
<pre>
8 * .Machine$sizeof.pointer
</pre>
where '''sizeof.pointer''' returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.
 
== 32- and 64-bit ==
See [http://cran.r-project.org/doc/manuals/R-admin.html#Choosing-between-32_002d-and-64_002dbit-builds R-admin.html].
* For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
* Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
* Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).


# 2. OK. Make the variable global
== Handling length 2^31 and more in R 3.0.0 ==
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict() 
ArrayTools


# 3. OK. Create a global variable
From R News for 3.0.0 release:
ArrayTools <- "C:/Program Files"
predict <- function() {
  source("foo.R")
}
predict()
</syntaxhighlight>


'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
''There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.  
''


Example 1.  
In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error
<pre>
<pre>
> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
> x <- seq(1, 2^31)
> ttt
Error in from:to : result would be too long a vector
  type JpnTest
1    a    999
2   b    999
3    c    999
4    d    999
5    e    999
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1    a      N5
2    b    999
3    c    999
4    d    999
5    e    999
> ttt
  type JpnTest
1    a     999
2    b    999
3    c    999
4    d    999
5    e    999
</pre>
</pre>


Example 2. [http://stackoverflow.com/questions/1236620/global-variables-in-r How can we set global variables inside a function?] The answer is to use the "<<-" operator or '''assign(, , envir = .GlobalEnv)''' function.
However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):
 
<pre>
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
> system.time(x <- seq(1,2^31))
 
  user  system elapsed
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
  8.604  11.060 120.815
 
> length(x)
==== New environment ====
[1] 2147483648
http://adv-r.had.co.nz/Environments.html
> length(x)/2^20
 
[1] 2048
Run the same function on a bunch of R objects
> gc()
<syntaxhighlight lang='rsplus'>
            used    (Mb) gc trigger    (Mb)  max used    (Mb)
mye = new.env()
Ncells    183823    9.9    407500    21.8    350000    18.7
load(<filename>, mye)
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
for(n in names(mye)) n = as_tibble(mye[[n]])
>
</syntaxhighlight>
</pre>
Note:
# 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
# On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
# My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
# For large dataset, it makes sense to use database or specially crafted packages like [http://cran.r-project.org/web/packages/bigmemory/ bigmemory] or [http://cran.r-project.org/web/packages/ff/ ff] or [https://privefl.github.io/bigstatsr/ bigstatsr].
# [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17330 [[<- for index 2^31 fails]
 
== NA in index ==
* Question: what is seq(1, 3)[c(1, 2, NA)]?


=== Speedup R code ===
Answer: It will reserve the element with NA in indexing and return the value NA for it.
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+


=== Profiler ===
* Question: What is TRUE & NA?
(Video) [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler]
Answer: NA


=== && vs & ===
* Question: What is FALSE & NA?
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.
Answer: FALSE


The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The longer form evaluates left to right examining only the first element of each vector.
* Question: c("A", "B", NA) != "" ?
Answer: TRUE TRUE NA


=== Vectorization ===
* Question: which(c("A", "B", NA) != "") ?
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
Answer: 1 2
* [https://en.wikipedia.org/wiki/Array_programming Array programming] from wikipedia
* [https://en.wikipedia.org/wiki/SIMD Single instruction, multiple data (SIMD)] from wikipedia
* [https://stackoverflow.com/a/1422181 What is vectorization] stackoverflow
* http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html
* https://github.com/vsbuffalo/devnotes/wiki/R-and-Vectorization
* [https://github.com/henrikbengtsson/matrixstats matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors). E.g. col / rowMedians(), col / rowRanks(), and col / rowSds(). [https://github.com/HenrikBengtsson/matrixStats/wiki/Benchmark-reports Benchmark reports].


==== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ====
* Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?
split() can be used to split a vector, columns or rows. See [https://stackoverflow.com/questions/3302356/how-to-split-a-data-frame How to split a data frame?]
Answer: TRUE TRUE FALSE
* Split rows of a data frame/matrix <syntaxhighlight lang='rsplus'>
 
split(mtcars,mtcars$cyl)
* Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?
</syntaxhighlight>
Answer: TRUE TRUE FALSE
* Split columns of a data frame/matrix. <syntaxhighlight lang='rsplus'>
ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10)))
# $`1`
#  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
#
# $`2`
#  [1] 16  9  4  1  0  1  4  9 16 25
</syntaxhighlight>
* split() + sapply() to merge columns. See below 'Mean of duplicated columns' for more detail.
* split() + sapply() to split a vector. See [https://www.rdocumentation.org/packages/genefilter/versions/1.54.2/topics/nsFilter nsFilter()] function which can remove duplicated probesets/rows using unique Entrez Gene IDs ('''genefilter''' package). The source code of [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/nsFilter.R#L224 nsFilter()] and [https://github.com/Bioconductor/genefilter/blob/b86f2cf47cf420b1444188bfe970714a7cc7f33b/R/all.R#L170 findLargest()]. <syntaxhighlight lang='rsplus'>
tSsp = split.default(testStat, lls)  
# testStat is a vector of numerics including probeset IDs as names
# lls is a vector of entrez IDs (same length as testStat)
# tSSp is a list of the same length as unique elements of lls.


sapply(tSsp, function(x) names(which.max(x)))
'''Conclusion''': In order to exclude empty or NA for numerical or character data type, we can use '''which()''' or a convenience function '''keep.complete(x) <- function(x) x != "" & !is.na(x)'''. This will guarantee return logical values and not contain NAs.
# return a vector of probset IDs of length of unique entrez IDs
</syntaxhighlight>
: And here is another example from the [https://cran.r-project.org/web/packages/bigmemory/vignettes/Overview.pdf bigmemory] vignette,
: <syntaxhighlight lang='rsplus'>
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
                    function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
</syntaxhighlight>


==== Mean of duplicated columns: rowMeans ====
Don't just use x != "" OR !is.na(x).
* [https://stackoverflow.com/questions/35925529/reduce-columns-of-a-matrix-by-a-function-in-r Reduce columns of a matrix by a function in R]  <syntaxhighlight lang='rsplus'>
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
colnames(x) <- c("A","A", "b", "b", "b", "c"); x
res <- sapply(split(1:ncol(x), colnames(x)),
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
res


# vapply() is safter than sapply().  
=== Some functions ===
# The 3rd arg in vapply() is a template of the return value.
* X %>% [https://tidyr.tidyverse.org/reference/drop_na.html tidyr::drop_na()]
res2 <- vapply(split(1:ncol(x), colnames(x)),
* '''stats::na.omit()''' and '''stats::complete.cases()'''. [https://statisticsglobe.com/na-omit-r-example/ NA Omit in R | 3 Example Codes for na.omit (Data Frame, Vector & by Column)]
              function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
              rep(0, nrow(x)))
</syntaxhighlight>
* [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/colSums colSums, rowSums, colMeans, rowMeans] (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster. <syntaxhighlight lang='rsplus'>
rowMeans(x, na.rm=T)
# [1] 31 27 28 29 30 31 32 33 34 35


apply(x, 1, mean, na.rm=T)
== Constant and 'L' ==
# [1] 31 27 28 29 30 31 32 33 34 35
Add 'L' after a constant. For example,
</syntaxhighlight>
{{Pre}}
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
for(i in 1L:n) { }


==== Mean of duplicated rows: colMeans and rowsum ====
if (max.lines > 0L) { }
* colMeans(x, na.rm = FALSE, dims = 1)
 
: <syntaxhighlight lang='rsplus'>
label <- paste0(n-i+1L, ": ")
x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
 
rownames(x) <- c(rep("a", 2), rep("b", 3), rep("c", 4), "d")
n <- length(x);  if(n == 0L) { }
res <- sapply(split(1:nrow(x), rownames(x)),
</pre>
              function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
 
res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
== Vector/Arrays ==
R indexes arrays from 1 like Fortran, not from 0 like C or Python.
 
=== remove integer(0) ===
[https://stackoverflow.com/a/27980810 How to remove integer(0) from a vector?]
 
=== Append some elements ===
[https://www.r-bloggers.com/2023/09/3-r-functions-that-i-enjoy/ append() and its after argument]
 
=== setNames() ===
Assign names to a vector
 
<pre>
z <- setNames(1:3, c("a", "b", "c"))
# OR
z <- 1:3; names(z) <- c("a", "b", "c")
# OR
z <- c("a"=1, "b"=2, "c"=3) # not work if "a", "b", "c" is like x[1], x[2], x[3].
</pre>
 
== Factor ==
=== labels argument ===
We can specify the factor levels and new labels using the factor() function.
 
{{Pre}}
sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))
 
factor(rev(letters[1:3]), labels = c("A", "B", "C"))
# C B A
# Levels: A B C
</pre>
 
=== Create a factor/categorical variable from a continuous variable: cut() and dplyr::case_when() ===
* [https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html Mastering Data Segmentation: A Guide to Using the cut() Function in R]
:<syntaxhighlight lang='r'>
cut(
    c(0, 10, 30),  
    breaks = c(0, 30, 50, Inf),  
    labels = c("Young", "Middle-aged", "Elderly")
) # Default include.lowest = FALSE
# [1] <NA>  Young Young
</syntaxhighlight>
</syntaxhighlight>
* rowsum(x, group, reorder = TRUE, …)
* https://dplyr.tidyverse.org/reference/case_when.html
: <syntaxhighlight lang='rsplus'>
* [https://rpubs.com/DaveRosenman/ifelsealternative Using dplyr’s mutate and case_when functions as alternative for if else statement]
x <- matrix(runif(100), ncol = 5) # 20 x 5
* [http://www.datasciencemadesimple.com/case-statement-r-using-case_when-dplyr/ Case when in R using case_when() Dplyr – case_when in R]
group <- sample(1:8, 20, TRUE)
* [https://predictivehacks.com/how-to-convert-continuous-variables-into-categorical-by-creating-bins/ How To Convert Continuous Variables Into Categorical By Creating Bins]
(xsum <- rowsum(x, group)) # 8 x 5
<ul>
</syntaxhighlight>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/cut ?cut]
* [https://stackoverflow.com/questions/25198442/how-to-calculate-mean-median-per-group-in-a-dataframe-in-r How to calculate mean/median per group in a dataframe in r] where '''doBy''' and '''dplyr''' are recommended.
{{Pre}}
* [https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
set.seed(1)
* [https://cran.r-project.org/web/packages/doBy/ doBy] package
x <- rnorm(100)
* [http://stackoverflow.com/questions/7881660/finding-the-mean-of-all-duplicates use ave() and unique()]
facVar <- cut(x, c(min(x), -1, 1, max(x)), labels = c("low", "medium", "high"))
* [http://stackoverflow.com/questions/17383635/average-between-duplicated-rows-in-r data.table package]
table(facVar, useNA = "ifany")
* [http://stackoverflow.com/questions/10180132/consolidate-duplicate-rows plyr package]
facVar
* [http://www.statmethods.net/management/aggregate.html aggregate()] function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. [http://www.win-vector.com/blog/2015/10/dont-use-statsaggregate/ Don't use aggregate] post. <syntaxhighlight lang='rsplus'>
#  low medium  high  <NA>
> attach(mtcars)
#    10    74    15      1
dim(mtcars)
</pre>
[1] 32 11
Note the option '''include.lowest = TRUE''' is needed when we use cut() + quantile(); otherwise the smallest data will become NA since the intervals have the format '''(a, b]'''.
> head(mtcars)
<pre>
                  mpg cyl disp  hp drat    wt  qsec vs am gear carb
x2 <- cut(x, quantile(x, 0:2/2), include.lowest = TRUE) # split x into 2 levels
Mazda RX4        21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
x2 <- cut(x, quantile(x, 0:3/3), include.lowest = TRUE) # split x into 3 levels
Mazda RX4 Wag    21.0  6  160 110 3.90 2.875 17.02  0  1    4    4
 
Datsun 710        22.8  4  108  93 3.85 2.320 18.61  1  1    4    1
library(tidyverse); library(magrittr)
Hornet 4 Drive    21.4  6  258 110 3.08 3.215 19.44  1  0    3   1
set.seed(1)
Hornet Sportabout 18.7  8  360 175 3.15 3.440 17.02  0  0    3    2
breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
Valiant          18.1   6  225 105 2.76 3.460 20.22  1  0   3    1
x <- runif(50)
> aggdata <-aggregate(mtcars, by=list(cyl,vs), FUN=mean, na.rm=TRUE)
bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)
> print(aggdata)
 
   Group.1 Group.2      mpg cyl  disp      hp     drat      wt     qsec vs
data.frame(sc=x, bins=bins) %>%
1      4      0 26.00000  4 120.30  91.0000 4.430000 2.140000 16.70000  0
  group_by(bins) %>%
2       6      0 20.56667   6 155.00 131.6667 3.806667 2.755000 16.32667  0
  summarise(n=n()) %>%
3      8      0 15.10000   8 353.10 209.2143 3.229286 3.999214 16.77214  0
   ggplot(aes(x = bins, y = n)) +
4      4      1 26.73000   4 103.62  81.8000 4.035000 2.300300 19.38100  1
     geom_col(color = "black", fill = "#90AACB") +
5      6      1 19.12500  6 204.55 115.2500 3.420000 3.388750 19.21500  1
     theme_minimal() +
        am    gear    carb
    theme(axis.text.x = element_text(angle = 90)) +
1 1.0000000 5.000000 2.000000
    theme(legend.position = "none") + coord_flip()
2 1.0000000 4.333333 4.666667
</pre>
3 0.1428571 3.285714 3.500000
<li>[https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html A Guide to Using the cut() Function in R]
4 0.7000000 4.000000 1.500000
<li>[https://youtu.be/7oyiPBjLAWY?t=2480 tibble object]
5 0.0000000 3.500000 2.500000
{{Pre}}
> detach(mtcars)
library(tidyverse)
tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
      age_cat = case_when(
          age_yrs < 2 ~ "baby",
          age_yrs < 13 ~ "kid",
          age_yrs < 20 ~ "teen",
          TRUE        ~ "adult")
)
</pre>
</li>
<li>[https://youtu.be/JsNqXLl3eFc?t=96 R tip: Learn dplyr’s case_when() function]
<pre>
case_when(
   condition1 ~ value1,
   condition2 ~ value2,
   TRUE ~ ValueAnythingElse
)
# Example
case_when(
  x %%2 == 0 ~ "even",
  x %%2 == 1 ~ "odd",
  TRUE ~ "Neither even or odd"
)
</pre>
<li>
</ul>


# Another example: select rows with a minimum value from a certain column (yval in this case)
=== How to change one of the level to NA ===
> mydf <- read.table(header=T, text='
https://stackoverflow.com/a/25354985. Note that the factor level is removed.
id xval yval
<pre>
A 1  1
x <- factor(c("a", "b", "c", "NotPerformed"))
A -2  2
levels(x)[levels(x) == 'NotPerformed'] <- NA
B 3  3
</pre>
B 4  4
C 5  5
')
> x = mydf$xval
> y = mydf$yval
> aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
  id xval yval
1  A    1    1
2  B    3    3
3  C    5    5
</syntaxhighlight>


=== Apply family ===
[https://webbedfeet.netlify.app/post/creating-missing-values-in-factors/ Creating missing values in factors]
Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply. Check out [http://people.stern.nyu.edu/ylin/r_apply_family.html this].


The following list gives a hierarchical relationship among these functions.
=== Concatenating two factor vectors ===
* apply(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
Not trivial. [https://stackoverflow.com/a/5068939 How to concatenate factors, without them being converted to integer level?].
* tapply(X, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a [https://en.wikipedia.org/wiki/Jagged_array "Ragged" Array]
<pre>
** by(data, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to a Data Frame Split by Factors
unlist(list(f1, f2))
** aggregate(x, by, FUN, ..., simplify = TRUE, drop = TRUE) - Compute Summary Statistics of Data Subsets
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))
* lapply(X, FUN, ...) – Apply a Function over a List or Vector
</pre>
** sapply(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
 
*** replicate(n, expr, simplify = "array")
=== droplevels() ===
** mapply(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/droplevels droplevels()]: drop unused levels from a factor or, more commonly, from factors in a data frame.
*** Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
 
** vapply(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
=== factor(x , levels = ...) vs levels(x) <- ===
* rapply(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
<span style="color: red">Note [https://stat.ethz.ch/R-manual/R-devel/library/base/html/levels.html levels(x)] is to set/rename levels, not reorder.</span> Use <s>'''relevel()'''</s> or '''factor()''' to reorder.  
* eapply(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment


Note that, apply's performance is not always better than a for loop. See
{| class="wikitable"
* http://tolstoy.newcastle.edu.au/R/help/06/05/27255.html (answered by Brian Ripley)
|-
* https://stat.ethz.ch/pipermail/r-help/2014-October/422455.html (has one example)
| [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/levels levels()]</br>[https://www.rdocumentation.org/packages/plyr/versions/1.8.9/topics/revalue plyr::revalue()]</br>[https://rdocumentation.org/packages/forcats/versions/1.0.0/topics/fct_recode forcats::fct_recode()]
| rename levels
|-
| [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/factor factor(, levels)]
| reorder levels
|}


The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out [http://www.theanalystatlarge.com/for-loop-tracking-windows-progress-bar/ this post]. It uses winProgressBar() and setWinProgressBar() functions.
<syntaxhighlight lang='rsplus'>
sizes <- factor(c("small", "large", "large", "small", "medium"))
sizes
#> [1] small  large large  small  medium
#> Levels: large medium small


==== Progress bar ====
sizes2 <- factor(sizes, levels = c("small", "medium", "large")) # reorder levels but data is not changed
[http://peter.solymos.org/code/2016/09/11/what-is-the-cost-of-a-progress-bar-in-r.html What is the cost of a progress bar in R?]
sizes2
# [1] small  large  large  small  medium
# Levels: small medium large


==== lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists ====
sizes3 <- sizes
* Examples of using lapply() + split() on a data frame. See [http://rollingyours.wordpress.com/category/r-programming-apply-lapply-tapply/ rollingyours.wordpress.com].
levels(sizes3) <- c("small", "medium", "large") # rename, not reorder
* mapply() [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/mapply documentation]. [https://stackoverflow.com/questions/9519543/merge-two-lists-in-r Use mapply() to merge lists].
                                                # large -> small
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/fprog.html Map() and Reduce()] in functional programming
                                                # medium -> medium
* Map(), Reduce(), and Filter() from [http://adv-r.had.co.nz/Functionals.html#functionals-fp Advanced R] by Hadley
                                                # small -> large
** If you have two or more lists (or data frames) that you need to process in <span style="color: red">parallel</span>, use '''Map()'''. One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to '''mapply()''' function and is more concise than '''lapply()'''. [http://adv-r.had.co.nz/Functionals.html#functionals-loop Advanced R] has a comment that Map() is better than mapply(). <syntaxhighlight lang='rsplus'>
sizes3
# Syntax: Map(f, ...)
# [1] large  small  small  large  medium
# Levels: small medium large
</syntaxhighlight>
A regression example.
<syntaxhighlight lang='rsplus'>
set.seed(1)
x <- sample(1:2, 500, replace = TRUE)
y <- round(x + rnorm(500), 3)
x <- as.factor(x)
sample_data <- data.frame(x, y)
# create linear model
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)  
# (Intercept) 0.96804    0.06610  14.65  <2e-16 ***
# x2          0.99620    0.09462  10.53  <2e-16 ***


xs <- replicate(5, runif(10), simplify = FALSE)
# Wrong way when we want to change the baseline level to '2'
ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
# No change on the model fitting except the apparent change on the variable name in the printout
Map(weighted.mean, xs, ws)
levels(sample_data$x) <- c("2", "1")
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)   
# (Intercept)  0.96804    0.06610  14.65  <2e-16 ***
# x1          0.99620    0.09462  10.53  <2e-16 ***


# instead of a more clumsy way
# Correct way if we want to change the baseline level to '2'
lapply(seq_along(xs), function(i) {
# The estimate was changed by flipping the sign from the original data
  weighted.mean(xs[[i]], ws[[i]])
sample_data$x <- relevel(x, ref = "2")
})
summary(lm( y~x, sample_data))
# Coefficients:
#            Estimate Std. Error t value Pr(>|t|)  
# (Intercept) 1.96425    0.06770  29.01  <2e-16 ***
# x1          -0.99620    0.09462  -10.53  <2e-16 ***
</syntaxhighlight>
</syntaxhighlight>
** Reduce() reduces a vector, x, to a single value by <span style="color: red">recursively</span> calling a function, f, two arguments at a time. A good example of using '''Reduce()''' function is to read a list of matrix files and merge them. See [https://stackoverflow.com/questions/29820029/how-to-combine-multiple-matrix-frames-into-one-using-r How to combine multiple matrix frames into one using R?] <syntaxhighlight lang='rsplus'>
# Syntax: Reduce(f, x, ...)


> m1 <- data.frame(id=letters[1:4], val=1:4)
=== stats::relevel() ===
> m2 <- data.frame(id=letters[2:6], val=2:6)
[https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/relevel relevel]. This function can only be used to change the '''reference level''' of a factor variable. '''It does not directly create an arbitrary order of levels'''. That is, it is useful in lm() or aov(), etc.
> merge(m1, m2, "id", all = T)
 
  id val.x val.y
=== reorder(), levels() and boxplot() ===
a     1    NA
<ul>
2  b    2    2
<li>[https://www.r-bloggers.com/2023/09/how-to-reorder-boxplots-in-r-a-comprehensive-guide/ How to Reorder Boxplots in R: A Comprehensive Guide] (tapply() method, simple & effective)
3  c    3    3
<li>[https://stat.ethz.ch/R-manual/R-devel/library/stats/html/reorder.factor.html reorder()].This is useful in barplot (ggplot2::geom_col()) where we want to sort the bars by a numerical variable.
4  d    4    4
<pre>
5  e    NA    5
# Syntax:
6  f    NA    6
# newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
> m <- list(m1, m2)
 
> Reduce(function(x,y) merge(x,y, "id",all=T), m)
(bymedian <- with(InsectSprays, reorder(spray, count, median)) )
  id val.x val.y
class(bymedian)
1  a    1    NA
levels(bymedian)
2  b    2    2
boxplot(count ~ bymedian, data = InsectSprays,
3  c    3    3
        xlab = "Type of spray", ylab = "Insect count",
4  d    4    4
        main = "InsectSprays data", varwidth = TRUE,
5  e    NA    5
        col = "lightgray") # boxplots are sorted according to the new levels
6  f    NA    6
boxplot(count ~ spray, data = InsectSprays,
</syntaxhighlight>
        xlab = "Type of spray", ylab = "Insect count",
* [https://statcompute.wordpress.com/2018/09/08/playing-map-and-reduce-in-r-subsetting/ Playing Map() and Reduce() in R – Subsetting] - using parallel and future packages. [https://statcompute.wordpress.com/2018/09/22/union-multiple-data-frames-with-different-column-names/ Union Multiple Data.Frames with Different Column Names]
        main = "InsectSprays data", varwidth = TRUE,
        col = "lightgray") # not sorted
</pre>
<li>[http://www.deeplytrivial.com/2020/05/statistics-sunday-my-2019-reading.html Statistics Sunday: My 2019 Reading] (reorder function)
</ul>
 
=== factor() vs ordered() ===
<pre>
factor(levels=c("a", "b", "c"), ordered=TRUE)
# ordered(0)
# Levels: a < b < c
 
factor(levels=c("a", "b", "c"))
# factor(0)
# Levels: a b c


==== sapply & vapply ====
ordered(levels=c("a", "b", "c"))
* [http://stackoverflow.com/questions/12339650/why-is-vapply-safer-than-sapply This] discusses why '''vapply''' is safer and faster than sapply.
# Error in factor(x, ..., ordered = TRUE) :
* [http://adv-r.had.co.nz/Functionals.html#functionals-loop Vector output: sapply and vapply] from Advanced R (Hadley Wickham).
#  argument "x" is missing, with no default
* [http://theautomatic.net/2018/11/13/those-other-apply-functions/ THOSE “OTHER” APPLY FUNCTIONS…]. rapply(), vapply() and eapply() are covered.
</pre>


See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up [https://github.com/SRTRdevhub/C_Statistic_Github/blob/master/Simulation_Demonstration.Rmd#L115 this example].
== Data frame ==
* http://adv-r.had.co.nz/Data-structures.html#data-frames. '''A data frame is a list of equal-length vectors'''. So a data frame is not a vector nor a matrix though it looks like a matrix.
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/


==== rapply - recursive version of lapply ====
=== stringsAsFactors = FALSE ===
* http://4dpiecharts.com/tag/recursive/
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/
* [https://github.com/wch/r-source/search?utf8=%E2%9C%93&q=rapply Search in R source code]. Mainly [https://github.com/wch/r-source/blob/trunk/src/library/stats/R/dendrogram.R r-source/src/library/stats/R/dendrogram.R].


==== replicate ====
We can use '''options(stringsAsFactors=FALSE)''' forces R to import character data as character objects.
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
<syntaxhighlight lang='rsplus'>
> replicate(5, rnorm(3))
          [,1]      [,2]      [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477
</syntaxhighlight>


See [[#parallel_package|parSapply()]] for a parallel version of replicate().
In R 4.0.0, [https://developer.r-project.org/Blog/public/2020/02/16/stringsasfactors/ stringAsFactors=FALSE] will be default. This also affects read.table() function.


==== Vectorize ====
=== check.names = FALSE ===
<syntaxhighlight lang='rsplus'>
Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
<pre>
[1] 2.17123
> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2))
  1a 2a
[1] 1.6491761 0.9610109
1  1  1
</syntaxhighlight>
2  2  2
> data.frame("1a"=1:2, "2a"=1:2) # default
  X1a X2a
1   1  1
2  2  2
</pre>


https://blogs.msdn.microsoft.com/gpalem/2013/03/28/make-vectorize-your-friend-in-r/
=== Create unique rownames: make.unique() ===
<syntaxhighlight lang='rsplus'>
<pre>
myfunc <- function(a, b) a*b
groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
myfunc(1, 2) # 2
rownames(mydf) <- make.unique(groupCodes)
myfunc(3, 5) # 15
</pre>
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15


myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
=== data.frame() will change rownames ===
myfunc2(1, 2) # 2
<pre>
myfunc2(3, 5) # 15
class(df2)
myfunc2(c(1,3), c(2,5)) # NA
# [1] "matrix" "array"
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
rownames(df2)[c(9109, 44999)]
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
# [1] "A1CF"    "A1BG-AS1"
                                        # parameter will be re-used
rownames(data.frame(df2))[c(9109, 44999)]
</syntaxhighlight>
# [1] "A1CF"    "A1BG.AS1"
</pre>


=== plyr and dplyr packages ===
=== Print a data frame without rownames ===
[https://peerj.com/collections/50-practicaldatascistats/ Practical Data Science for Stats - a PeerJ Collection]
<pre>
# Method 1.  
rownames(df1) <- NULL


[http://www.jstatsoft.org/v40/i01/paper The Split-Apply-Combine Strategy for Data Analysis] (plyr package) in J. Stat Software.
# Method 2.  
print(df1, row.names = FALSE)
</pre>


[http://seananderson.ca/courses/12-plyr/plyr_2012.pdf A quick introduction to plyr] with a summary of apply functions in R and compare them with functions in plyr package.
=== Convert data frame factor columns to characters ===
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
{{Pre}}
# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)


# plyr has a common syntax -- easier to remember
# Method 2:
# plyr requires less code since it takes care of the input and output format
bob[] <- lapply(bob, as.character)
# plyr can easily be run in parallel -- faster
</pre>


Tutorials
[https://stackoverflow.com/a/2853231 To replace only factor columns]:
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
<pre>
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
# Method 1:
* [http://www.dataschool.io/dplyr-tutorial-for-faster-data-manipulation-in-r/ Hands-on dplyr tutorial for faster data manipulation in R] from dataschool.io.
i <- sapply(bob, is.factor)
bob[i] <- lapply(bob[i], as.character)


Examples of using dplyr:  
# Method 2:
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
library(dplyr)
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
bob %>% mutate_if(is.factor, as.character) -> bob
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.  
</pre>
* [http://rpubs.com/turnersd/plot-deseq-results-multipage-pdf DESeq2 post analysis] select(), gather(), arrange() and %>%.


==== [https://cran.r-project.org/web/packages/tibble/ tibble] ====
=== Sort Or Order A Data Frame ===
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
[https://howtoprogram.xyz/2018/01/07/r-how-to-order-a-data-frame/ How To Sort Or Order A Data Frame In R]
# df[order(df$x), ], df[order(df$x, decreasing = TRUE), ], df[order(df$x, df$y), ]
# library(plyr); arrange(df, x), arrange(df, desc(x)), arrange(df, x, y)
# library(dplyr); df %>% arrange(x),df %>% arrange(x, desc(x)), df %>% arrange(x, y)
# library(doBy); order(~x, df), order(~ -x, df), order(~ x+y, df)


<syntaxhighlight lang='rsplus'>
=== data.frame to vector ===
> data(pew, package = "efficient")
<pre>
> dim(pew)  
df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))
[1] 18 10
 
> class(pew) # tibble is also a data frame!!
class(df)
[1] "tbl_df"     "tbl"       "data.frame"
# [1] "data.frame"
class(t(df))
# [1] "matrix" "array"
class(unlist(df))
# [1] "numeric"
 
# Method 1: Convert data frame to matrix using as.matrix()
# and then Convert matrix to vector using as.vector() or c()
mat <- as.matrix(df)
vec1 <- as.vector(mat)  # [1] 1 2 3 4 5 6
vec2 <- c(mat)
 
# Method 2: Convert data frame to matrix using t()/transpose
# and then Convert matrix to vector using as.vector() or c()
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
vec4 <- c(t(df))
 
# Not working
as.vector(df)
# $x
# [1] 1 2 3
# $y
# [1] 4 5 6
 
# Method 3: unlist() - easiest solution
unlist(df)
# x1 x2 x3 y1 y2 y3
#  1  2  3  4  5  6
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
# [1] 1 2 3 4 5 6
</pre>
Q: Why as.vector(df) cannot convert a data frame into a vector?
 
A: The as.vector function cannot be used directly on a data frame to convert it into a vector because a data frame is a list of vectors (i.e., its columns) and '''as.vector only removes the attributes of an object to create a vector'''. When you apply as.vector to a data frame, R does not know how to concatenate these independent columns (which could be of different types) into a single vector. Therefore, it doesn’t perform the operation. Therefore as.vector() returns the underlying list structure of the data frame instead of converting it into a vector.
 
However, when you transpose the data frame using t(), it gets converted into a matrix. A matrix in R is a vector with dimensions. Therefore, all elements of the matrix must be of the same type. If they are not, R will coerce them to be so. Once you have a matrix, as.vector() can easily convert it into a vector because all elements are of the same type.


> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
=== Using cbind() to merge vectors together? ===
# A tibble: 162 x 3
It’s a common mistake to try and create a data frame by cbind()ing vectors together. This doesn’t work because cbind() will create a matrix unless one of the arguments is already a data frame. Instead use data.frame() directly. See [http://adv-r.had.co.nz/Data-structures.html#data-frames Advanced R -> Data structures] chapter.  
                                                      religion Income Count
                                                          <chr>  <chr> <int>
1                                                    Agnostic  <$10k    27
2                                                      Atheist  <$10k    12
...
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[, 3])
[1] NA
Warning message:
In mean.default(tidyr::gather(pew, key = Income, value = Count,  :
  argument is not numeric or logical: returning NA
> mean(tidyr::gather(pew, key=Income, value = Count, -religion)[[3]])
[1] 181.6975
</syntaxhighlight>


If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.
=== cbind NULL and data.frame ===
[https://9to5tutorial.com/cbind-can-t-combine-null-with-dataframe cbind can't combine NULL with dataframe]. Add as.matrix() will fix the problem.


'''Subsetting''': to [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull().
=== merge ===
<syntaxhighlight lang='rsplus'>
* [https://thomasadventure.blog/posts/r-merging-datasets/ All You Need To Know About Merging (Joining) Datasets in R]. If we like to merge/join by the rownames, we can use '''dplyr::rownames_to_column()'''; see [https://stackoverflow.com/a/42418771 dplyr left_join() by rownames].
TibbleObject$VarName
* [https://www.geeksforgeeks.org/merge-dataframes-by-row-names-in-r/ Merge DataFrames by Row Names in R]
# OR
* [https://jozefhajnala.gitlab.io/r/r006-merge/ How to perform merges (joins) on two or more data frames with base R, tidyverse and data.table]
TibbleObject[["VarName"]]
* [https://www.dummies.com/programming/r/how-to-use-the-merge-function-with-data-sets-in-r/ How to understand the different types of merge]
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore
</syntaxhighlight>


==== llply() ====
Special character in the matched variable can create a trouble when we use merge() or dplyr::inner_join(). I guess R internally turns df2 (a matrix but not a data frame) to a data frame (so rownames are changed if they contain special character like "-"). This still does not explain the situation when I
llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.
<pre>
<pre>
LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])
class(df1); class(df2)
# [1] "data.frame"  # 2 x 2
# [1] "matrix" "array" # 52439 x 2
rownames(df1)
# [1] "A1CF"    "A1BG-AS1"
merge(df1, df2[c(9109, 44999), ], by=0)
#  Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BG-AS1    0    0  7.321358  6.908333
# 2      A1CF    0    0  3.011470  1.189578
merge(df1, df2[c(9109, 38959:44999), ], by= 0) # still correct
merge(df1, df2[c(9109, 38958:44999), ], by= 0) # same as merge(df1, df2, by=0)
#  Row.names 786-0 A498 ACH-000001 ACH-000002
# 1      A1CF    0    0    3.01147  1.189578
rownames(df2)[38958:38959]
# [1] "ITFG2-AS1" "ADGRD1-AS1"
 
rownames(df1)[2] <- "A1BGAS1"
rownames(df2)[44999] <- "A1BGAS1"
merge(df1, df2, by= 0)
#  Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BGAS1    0    0  7.321358  6.908333
# 2      A1CF    0    0  3.011470  1.189578
</pre>
</pre>
where rLLID is a list of entrez ID. For example,
 
=== is.matrix: data.frame is not necessarily a matrix ===
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/matrix ?matrix]. is.matrix returns TRUE '''if x is a vector and has a "dim" attribute of length 2''' and FALSE otherwise.  
 
An example that is a data frame (is.data.frame() returns TRUE) but not a matrix (is.matrix() returns FALSE) is an object returned by
<pre>
<pre>
get("org.Hs.egGO")[["6772"]]
X <- data.frame(x=1:2, y=3:4)
</pre>  
</pre>
returns a list of 49 GOs.
The 'X' object is NOT a vector and it does NOT have the "dim" attribute. It has only 3 attributes: "names", "row.names" & "class". Note that dim() function works fine and returns correctly though there is not "dim" attribute.  


==== ddply() ====
Another example that is a data frame but not a matrix is the built-in object ''cars''; see ?matrix. It is not a vector
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html


==== ldply() ====
=== Convert a data frame to a matrix: as.matrix() vs data.matrix() ===
[http://rpsychologist.com/an-r-script-to-automatically-look-at-pubmed-citation-counts-by-year-of-publication/ An R Script to Automatically download PubMed Citation Counts By Year of Publication]
If I have a data frame X which recorded the time of some files.


=== Using R's set.seed() to set seeds for use in C/C++ (including Rcpp) ===
* is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
http://rorynolan.rbind.io/2018/09/30/rcsetseed/
* as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/data.matrix data.matrix(X)] will convert the time to numerical values. So use data.matrix() if the data is numeric. The returned object is not a data frame anymore.


==== get_seed() ====
<syntaxhighlight lang='r'>
See the same blog
# latex directory contains cache files from knitting an rmarkdown file
<syntaxhighlight lang='rsplus'>
X <- list.files("latex/", full.names = T) %>%
get_seed <- function() {
    grep("RData", ., value=T) %>%
  sample.int(.Machine$integer.max, 1)
    file.info() %>% 
}
    `[`("mtime")
X %>% is.data.frame() # TRUE
X %>% is.matrix() # FALSE
X %>% as.matrix() %>% is.matrix() # TRUE
X %>% data.matrix() %>% is.matrix() # TRUE
X %>% as.matrix() %>% "["(1:2, ) # timestamps
X %>% data.matrix() %>% "["(1:2, ) # numeric
</syntaxhighlight>
</syntaxhighlight>


=== set.seed(), for loop and saving random seeds ===
* The '''as.matrix()''' function is used to coerce an object into a matrix. It can be used with various types of R objects, such as vectors, data frames, and arrays.
http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
* The '''data.matrix()''' function is specifically designed for converting a data frame into a matrix by coercing all columns to numeric values. If the data frame contains non-numeric columns, such as character or factor columns, data.matrix() will convert them to numeric values if possible (e.g., by converting factors to their integer codes).
* See the following example where as.matrix() and data.matrix() return different resuls.
<syntaxhighlight lang='r'>
df <- data.frame(a = c(1, 2, 3), b = c("x", "y", "z"))
mat <- as.matrix(df)
mat
#      a  b 
# [1,] "1" "x"
# [2,] "2" "y"
# [3,] "3" "z"
class(mat)
# [1] "matrix" "array"
mat2 <- data.matrix(df)
mat2
#      a b
# [1,] 1 1
# [2,] 2 2
# [3,] 3 3
class(mat2)
# [1] "matrix" "array"
typeof(mat)
# [1] "character"
typeof(mat2)
# [1] "double"
</syntaxhighlight>


<syntaxhighlight lang='rsplus'>
=== matrix vs data.frame ===
set.seed(1001)  
Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.
data <- vector("list", 30)  
<pre>
seeds <- vector("list", 30)  
Browse[2]> names(res2$surv.data.new[[index]])
for(i in 1:30) {
NULL
  seeds[[i]] <- .Random.seed
Browse[2]> colnames(res2$surv.data.new[[index]])
  data[[i]] <- runif(5)  
[1] "time"  "status" "treat"  "AKT1"  "BRAF"  "FLOT2"  "MTOR"  "PCK2"  "PIK3CA"
}
[10] "RAF1"  
   
Browse[2]> mode(res2$surv.data.new[[index]])
.Random.seed <- seeds[[23]] # restore
[1] "numeric"
data.23 <- runif(5)
Browse[2]> is.matrix(res2$surv.data.new[[index]])
data.23
[1] TRUE
data[[23]]  
Browse[2]> dim(res2$surv.data.new[[index]])
</syntaxhighlight>
[1] 991  10
* Duncan Murdoch: ''This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed.  See ?.Random.seed for details.''
</pre>
* Uwe Ligges's comment: ''set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.''
* Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.


=== Regular Expression ===
Case 2:
See [[Regular_expression|here]].
{{Pre}}
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])  # OK


=== Read rrd file ===
ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
* https://en.wikipedia.org/wiki/RRDtool
unique(ip2$Priority)    # OK
* http://oss.oetiker.ch/rrdtool/
</pre>
* https://github.com/pldimitrov/Rrd
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/


=== file, connection ===
The length of a matrix and a data frame is different.
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/cat cat()] and [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/scan scan()] (read data into a vector or list from the console or file)
{{Pre}}
* read() and write()
> length(matrix(1:6, 3, 2))
* read.table() and write.table()
[1] 6
> length(data.frame(matrix(1:6, 3, 2)))
[1] 2
> x[1]
  X1
1  1
2  2
3
4  4
5
6  6
> x[[1]]
[1] 1 2 3 4 5 6
</pre>
So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.


=== Clipboard (?connections) & textConnection() ===
=== How to Remove Duplicates ===
<syntaxhighlight lang='rsplus'>
[https://www.r-bloggers.com/2021/08/how-to-remove-duplicates-in-r-with-example/ How to Remove Duplicates in R with Example]
source("clipboard")
read.table("clipboard")
</syntaxhighlight>


* On Windows, we can use readClipboard() and writeClipboard().
=== Convert a matrix (not data frame) of characters to numeric ===
* reading/writing clipboard method seems not quite stable on Linux/macOS. So the alternative is to use the [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function: <syntaxhighlight lang='rsplus'>
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
{{Pre}}
</syntaxhighlight> An example is to copy data from [https://stackoverflow.com/questions/28426026/plotting-boxplots-of-multiple-y-variables-using-ggplot2-qplot-or-others?answertab=active#tab-top this post]. In this case we need to use read.table() instead of read.delim().
tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
    a    b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)    # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp) # "character"


=== read/manipulate binary data ===
> mode(tmp) <- "numeric"
* x <- readBin(fn, raw(), file.info(fn)$size)
> sum(tmp)
* rawToChar(x[1:16])
[1] 1.917
* See Biostrings C API
</pre>
 
=== Convert Data Frame Row to Vector ===
as.numeric() or '''c()'''


=== String Manipulation ===
=== Convert characters to integers ===
* [http://gastonsanchez.com/blog/resources/how-to/2013/09/22/Handling-and-Processing-Strings-in-R.html ebook] by Gaston Sanchez.
mode(x) <- "integer"
* [http://blog.revolutionanalytics.com/2018/06/handling-strings-with-r.html A guide to working with character data in R] (6/22/2018)
* Chapter 7 of the book 'Data Manipulation with R' by Phil Spector.
* Chapter 7 of the book 'R Cookbook' by Paul Teetor.
* Chapter 2 of the book 'Using R for Data Management, Statistical Analysis and Graphics' by Horton and Kleinman.
* http://www.endmemo.com/program/R/deparse.php. '''It includes lots of examples for each R function it lists.'''


=== HTTPs connection ===  
=== Non-Standard Evaluation ===
HTTPS connection becomes default in R 3.2.2. See
[https://thomasadventure.blog/posts/understanding-nse-part1/ Understanding Non-Standard Evaluation. Part 1: The Basics]
* http://blog.rstudio.org/2015/08/17/secure-https-connections-for-r/  
* http://blog.revolutionanalytics.com/2015/08/good-advice-for-security-with-r.html


[http://developer.r-project.org/blosxom.cgi/R-devel/2016/12/15#n2016-12-15 R 3.3.2 patched] The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)
=== Select Data Frame Columns in R ===
This is part of series of DATA MANIPULATION IN R from [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ datanovia.com]


=== setInternet2 ===
* pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.
* select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
* select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
* Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names


Read the [https://stat.ethz.ch/pipermail/r-devel/2015-August/071595.html discussion] reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in [https://stat.ethz.ch/pipermail/r-devel/2015-August/071623.html this post]. The following demonstrated the original problem.
Another way is to the dollar sign '''$''' operator (?"$") to extract rows or column from a data frame.
<pre>
<pre>
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
class(USArrests)  # "data.frame"
              "GCF_000001405.13.assembly.txt")
USArrests$"Assault"
f1 <- tempfile()
download.file(url, f1)
</pre>
</pre>
It seems the bug was fixed in R 3.2-branch. See [https://github.com/wch/r-source/commit/3a02ed3a50ba17d9a093b315bf5f31ffc0e21b89 8/16/2015] patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385098%28v=vs.85%29.aspx InternetOpenUrl()] function of [https://msdn.microsoft.com/en-us/library/windows/desktop/aa385473%28v=vs.85%29.aspx wininet] library. [http://slacksite.com/other/ftp.html This article] and [http://stackoverflow.com/questions/1699145/what-is-the-difference-between-active-and-passive-ftp this post] explain differences of active and passive FTP.
Note that for both data frame and matrix objects, we need to use the '''[''' operator to extract columns and/or rows.
<pre>
USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
#        Murder Assault
# Alabama  13.2    236
# Alaska    10.0    263
USArrests[c("Murder", "Assault")] # all rows


The following R command will show the exact svn revision for the R you are currently using.
tmp <- data(package="datasets")
class(tmp$results)  # "matrix" "array"
tmp$results[, "Item"]
# Same method can be used if rownames are available in a matrix
</pre>
Note for a '''data.table''' object, we can extract columns using the column names without double quotes.
<pre>
<pre>
R.Version()$"svn rev"
data.table(USArrests)[1:2, list(Murder, Assault)]
</pre>
</pre>


If setInternet2(T), then https protocol is supported in download.file().
=== Add columns to a data frame ===
[https://datasciencetut.com/how-to-add-columns-to-a-data-frame-in-r/ How to add columns to a data frame in R]


When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.
=== Exclude/drop/remove data frame columns ===
* [https://datasciencetut.com/remove-columns-from-a-data-frame/ How to Remove Columns from a data frame in R]
* [https://www.listendata.com/2015/06/r-keep-drop-columns-from-data-frame.html R: keep / drop columns from data frame]
<pre>
# method 1
df = subset(mydata, select = -c(x,z) )


The setInternet2() function is defined in [https://github.com/wch/r-source/commits/trunk/src/library/utils/R/windows/sysutils.R R> src> library> utils > R > windows > sysutils.R].
# method 2
drop <- c("x","z")
df = mydata[,!(names(mydata) %in% drop)]


'''R up to 3.2.2'''
# method 3: dplyr
<pre>
mydata2 = select(mydata, -a, -x, -y)
setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))
mydata2 = select(mydata, -c(a, x, y))
mydata2 = select(mydata, -a:-y)
mydata2 = mydata[,!grepl("^INC",names(mydata))]
</pre>
</pre>
See also
* <src/include/Internal.h> (declare do_setInternet2()),
* <src/main/names.c> (show do_setInternet2() in C)
* <src/main/internet.c>  (define do_setInternet2() in C).


Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file.  If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2,  it is also possible to download from https by setIneternet2(T).
=== Remove Rows from the data frame ===
[https://datasciencetut.com/remove-rows-from-the-data-frame-in-r/ Remove Rows from the data frame in R]


'''R 3.3.0'''
=== Danger of selecting rows from a data frame ===
<pre>
<pre>
setInternet2 <- function(use = TRUE) {
> dim(cars)
     if(!is.na(use)) stop("use != NA is defunct")
[1] 50  2
    NA
> data.frame(a=cars[1,], b=cars[2, ])
}
  a.speed a.dist b.speed b.dist
1      4      2      4     10
> dim(data.frame(a=cars[1,], b=cars[2, ]))
[1] 1 4
> cars2 = as.matrix(cars)
> data.frame(a=cars2[1,], b=cars2[2, ])
      a  b
speed 4  4
dist  2 10
</pre>
</pre>


Note that setInternet2.Rd says As from \R 3.3.0 it changes nothing, and only \code{use = NA} is accepted. Also NEWS.Rd says setInternet2() has no effect and will be removed in due course.
=== Creating data frame using structure() function ===
[https://tomaztsql.wordpress.com/2019/05/27/creating-data-frame-using-structure-function-in-r/ Creating data frame using structure() function in R]


=== File operation ===
=== Create an empty data.frame ===
* list.files()
https://stackoverflow.com/questions/10689055/create-an-empty-data-frame
* file.info()
<pre>
* dir.create()
# the column types default as logical per vector(), but are then overridden
* file.create()
a = data.frame(matrix(vector(), 5, 3,
* file.copy()
              dimnames=list(c(), c("Date", "File", "User"))),
              stringsAsFactors=F)
str(a) # NA but they are logical , not numeric.
a[1,1] <- rnorm(1)
str(a)
 
# similar to above
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))
 
# different data type
a <- data.frame(x1 = character(),
                x2 = numeric(),
                x3 = factor(),
                stringsAsFactors = FALSE)
</pre>


=== read/download/source a file from internet ===
=== Objects from subsetting a row in a data frame vs matrix ===
==== Simple text file http ====
* [https://stackoverflow.com/a/23534617 Warning: row names were found from a short variable and have been discarded]
<ul>
<li>Subsetting creates repeated rows. This will create unexpected rownames.
<pre>
<pre>
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
R> z <- data.frame(x=1:3, y=2:4)
R> rownames(z) <- letters[1:3]
R> rownames(z)[c(1,1)]
[1] "a" "a"
R> rownames(z[c(1,1),])
[1] "a"  "a.1"
R> z[c(1,1), ]
    x y
a  1 2
a.1 1 2
</pre>
</li>
<li>[https://stackoverflow.com/a/2545548 Convert a dataframe to a vector (by rows)] The solution is as.vector(t(mydf[i, ])) or c(mydf[i, ]). My example:
{{Pre}}
str(trainData)
# 'data.frame': 503 obs. of  500 variables:
#  $ bm001: num  0.429 1 -0.5 1.415 -1.899 ...
#  $ bm002: num  0.0568 1 0.5 0.3556 -1.16 ...
# ...
trainData[1:3, 1:3]
#        bm001      bm002    bm003
# 1  0.4289449 0.05676296 1.657966
# 2  1.0000000 1.00000000 1.000000
# 3 -0.5000000 0.50000000 0.500000
o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
# Warning message:
# In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
#  row names were found from a short variable and have been discarded
</pre>
</pre>


==== Zip file and url() function ====
'trees' data from the 'datasets' package
<pre>
<pre>
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
trees[1:3,]
source(con)
#  Girth Height Volume
close(con)
# 1  8.3    70  10.3
# 2  8.6    65  10.3
# 3  8.8    63  10.2
 
# Wrong ways:
data.frame(trees[1,] , trees[2,])
#  Girth Height Volume Girth.1 Height.1 Volume.1
# 1  8.3    70  10.3    8.6      65    10.3
data.frame(time=trees[1,] , status=trees[2,])
#  time.Girth time.Height time.Volume status.Girth status.Height status.Volume
# 1        8.3          70        10.3          8.6            65          10.3
data.frame(time=as.vector(trees[1,]) , status=as.vector(trees[2,]))
#  time.Girth time.Height time.Volume status.Girth status.Height status.Volume
# 1        8.3          70        10.3          8.6            65          10.3
data.frame(time=c(trees[1,]) , status=c(trees[2,]))
# time.Girth time.Height time.Volume status.Girth status.Height status.Volume
# 1        8.3          70        10.3          8.6            65          10.3
 
# Right ways:
# method 1: dropping row names
data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,])))
# OR
data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
#  time status
# 1  8.3    8.6
# 2 70.0  65.0
# 3 10.3  10.3
# method 2: keeping row names
data.frame(time=t(trees[1,]) , status=t(trees[2,]))
#          X1  X2
# Girth  8.3  8.6
# Height 70.0 65.0
# Volume 10.3 10.3
data.frame(time=unlist(trees[1,]) , status=unlist(trees[2,]))
#        time status
# Girth  8.3    8.6
# Height 70.0  65.0
# Volume 10.3  10.3
 
# Method 3: convert a data frame to a matrix
is.matrix(trees)
# [1] FALSE
trees2 <- as.matrix(trees)
data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
#        time status
# Girth  8.3    8.6
# Height 70.0  65.0
# Volume 10.3  10.3
 
dim(trees[1,])
# [1] 1 3
dim(trees2[1, ])
# NULL
trees[1, ]  # notice the row name '1' on the left hand side
#  Girth Height Volume
# 1  8.3    70  10.3
trees2[1, ]
#  Girth Height Volume
#    8.3  70.0  10.3
</pre>
</pre>
Here url() function is like file(),  gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.
</li>
</ul>


Another example of using url() is
=== Convert a list to data frame ===
[https://www.statology.org/convert-list-to-data-frame-r/ How to Convert a List to a Data Frame in R].
<pre>
<pre>
load(url("http:/www.example.com/example.RData"))
# method 1
data.frame(t(sapply(my_list,c)))
 
# method 2
library(dplyr)
bind_rows(my_list) # OR bind_cols(my_list)
 
# method 3
library(data.table)
rbindlist(my_list)
</pre>
</pre>


==== [http://cran.r-project.org/web/packages/downloader/index.html downloader] package ====
=== tibble and data.table ===
This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.
* [[R#tibble | tibble]]
* [[Tidyverse#data.table|data.table]]
 
=== Clean  a dataset ===
[https://finnstats.com/index.php/2021/04/04/how-to-clean-the-datasets-in-r/ How to clean the datasets in R]


==== Google drive file based on https using [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ====
== matrix ==
<pre>
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))
</pre>


==== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ====
=== Define and subset a matrix ===
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]
* [https://www.tutorialkart.com/r-tutorial/r-matrix/ Matrix in R]
** It is clear when a vector becomes a matrix the data is transformed column-wisely ('''byrow''' = FALSE, by default).
** When subsetting a matrix, it follows the format: '''X[rows, colums]''' or '''X[y-axis, x-axis]'''.


==== Github files https using RCurl package ====
* http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
* http://tonybreyal.wordpress.com/2011/11/24/source_https-sourcing-an-r-script-from-github/
<pre>
<pre>
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",  
data <- c(2, 4, 7, 5, 10, 1)
            ssl.verifypeer = FALSE)
A <- matrix(data, ncol = 3)
read.table(text=x)
print(A)
#      [,1] [,2] [,3]
# [1,]    2    7  10
# [2,]    4    5    1
 
A[1:1, 2:3, drop=F]
#      [,1] [,2]
# [1,]    7  10
</pre>
</pre>
* [http://cran.r-project.org/web/packages/gistr/index.html gistr] package


=== Create publication tables using '''tables''' package ===
=== Prevent automatic conversion of single column to vector ===
See p13 for example in http://www.ianwatson.com.au/stata/tabout_tutorial.pdf
use '''drop = FALSE''' such as mat[, 1, drop = FALSE].
 
=== complete.cases(): remove rows with missing in any column ===
It works on a sequence of vectors, matrices and data frames.


R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,
=== NROW vs nrow ===
<pre>
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/nrow ?nrow]. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.
> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
                                                 
                Sepal.Length      Sepal.Width   
Species    n  mean        sd  mean        sd 
setosa      50 5.01        0.35 3.43        0.38
versicolor  50 5.94        0.52 2.77        0.31
virginica  50 6.59        0.64 2.97        0.32
All        150 5.84        0.83 3.06        0.44
> str(iris)
'data.frame':  150 obs. of  5 variables:
$ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
$ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
$ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
$ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
$ Species    : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...
</pre>
and
<pre>
# This example shows some of the less common options       
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
> z <- rnorm(100)+5
> fmt <- function(x) {
  s <- format(x, digits=2)
  even <- ((1:length(s)) %% 2) == 0
  s[even] <- sprintf("(%s)", s[even])
  s
}
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
                  Status             
Sex    Statistic high  low    medium
Female mean      4.88  4.96  5.17
        sd        (1.20) (0.82) (1.35)
Male  mean      4.45  4.31  5.05
        sd        (1.01) (0.93) (0.75)
</pre>


See also a collection of R packages related to reproducible research in http://cran.r-project.org/web/views/ReproducibleResearch.html
=== matrix (column-major order) multiply a vector ===
* Matrices in R [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) are stored in a column-major order]. It means array slice A[,1] are contiguous.


=== Extracting tables from PDFs ===  
{{Pre}}
* [http://datascienceplus.com/extracting-tables-from-pdfs-in-r-using-the-tabulizer-package/ extracting Tables from PDFs in R] using Tabulizer. This needs the [https://cran.r-project.org/web/packages/rJava/index.html rJava] package. Some issue came out on my macOS 10.12 Sierra. '''Library not loaded: /Library/Java/JavaVirtualMachines/jdk-9.jdk/Contents/Home/lib/server/libjvm.dylib. Referenced from: /Users/XXXXXXX/Library/R/3.5/library/rJava/libs/rJava.so'''.
> matrix(1:6, 3,2)
* [https://cran.r-project.org/web/packages/pdftools/ pdftools] - Text Extraction, Rendering and Converting of PDF Documents. [https://ropensci.org/technotes/2018/12/14/pdftools-20/ pdf_text() and pdf_data()] functions. <syntaxhighlight lang='rsplus'>
    [,1] [,2]
library(pdftools)
[1,]    1    4
pdf_file <- "https://github.com/ropensci/tabulizer/raw/master/inst/examples/data.pdf"
[2,]    2    5
txt <- pdf_text(pdf_file) # length = number of pages
[3,]    3    6
# Suppose the table we are interested in is on page 1
> matrix(1:6, 3,2) * c(1,2,3) # c(1,2,3) will be recycled to form a matrix. Good quiz.
cat(txt[1]) # Good but not in a data frame format
    [,1] [,2]
[1,]    1    4
[2,]    4  10
[3,]    9  18
> matrix(1:6, 3,2) * c(1,2,3,4) # c(1,2,3,4) will be recycled
    [,1] [,2]
[1,]    1  16
[2,]    4    5
[3,]    9  12
</pre>
 
* [https://stackoverflow.com/a/20596490 How to divide each row of a matrix by elements of a vector in R]
 
=== add a vector to all rows of a matrix ===
[https://stackoverflow.com/a/39443126 add a vector to all rows of a matrix]. sweep() or rep() is the best.
 
=== sparse matrix ===
[https://stackoverflow.com/a/10555270 R convert matrix or data frame to sparseMatrix]
 
To subset a vector from some column of a sparseMatrix, we need to convert it to a regular vector, '''as.vector()'''.
 
== Attributes ==
* [https://statisticaloddsandends.wordpress.com/2020/10/19/attributes-in-r/ Attributes in R]
* [http://adv-r.had.co.nz/Data-structures.html#attributes Data structures] in "Advanced R"
 
== Names ==
[https://masalmon.eu/2023/11/06/functions-dealing-with-names/ Useful functions for dealing with object names]. (Un)Setting object names: stats::setNames(), unname() and rlang::set_names()
 
=== Print a vector by suppressing names ===
Use '''unname'''. sapply(, , USE.NAMES = FALSE).


pdf_data(pdf_file)[[1]] # data frame/tibble format
== format.pval/print p-values/format p values ==
</syntaxhighlight>
[https://rdrr.io/r/base/format.pval.html format.pval()]. By default it will show 5 significant digits (getOption("digits")-2).
{{Pre}}
> set.seed(1); format.pval(c(stats::runif(5), pi^-100, NA))
[1] "0.26551" "0.37212" "0.57285" "0.90821" "0.20168" "< 2e-16" "NA"
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04" "<2e-16"


=== Create flat tables in R console using ftable() ===
R> pvalue
<syntaxhighlight lang='rsplus'>
[1] 0.0004632104
> ftable(Titanic, row.vars = 1:3)
R> print(pvalue, digits =20)
                  Survived  No Yes
[1] 0.00046321036188223807528
Class Sex    Age                 
R> format.pval(pvalue)
1st  Male  Child            0   5
[1] "0.00046321"
            Adult          118  57
R> format.pval(pvalue * 1e-1)
      Female Child            0  1
[1] "4.6321e-05"
            Adult            4 140
R> format.pval(0.00004632)
2nd  Male  Child            0 11
[1] "4.632e-05"
            Adult          154  14
R> getOption("digits")
      Female Child            0  13
[1] 7
            Adult          13  80
</pre>
3rd  Male  Child          35  13
 
            Adult          387  75
=== Return type ===
      Female Child          17  14
The format.pval() function returns a string, so it’s not appropriate to use the returned object for operations like sorting.
            Adult          89  76
 
Crew  Male  Child            0  0
=== Wrong number of digits in format.pval() ===
            Adult          670 192
See [https://stackoverflow.com/questions/59779131/wrong-number-of-digits-in-format-pval here]. The solution is to apply round() and then format.pval().
      Female Child            0  0
<pre>
            Adult            3  20
x <- c(6.25433625041843e-05, NA, 0.220313341361346, NA, 0.154029880744594,
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
  0.0378437685448703, 0.023358329881356, NA, 0.0262561986351483,
            Survived  No Yes
  0.000251274794673796)
Class Sex                   
format.pval(x, digits=3)
1st  Male            118  62
# [1] "6.25e-05" "NA"       "0.220313" "NA"       "0.154030" "0.037844" "0.023358"
      Female            4 141
# [8] "NA"      "0.026256" "0.000251"
2nd  Male            154  25
 
      Female          13  93
round(x, 3) |> format.pval(digits=3, eps=.001)
3rd  Male            422  88
# [1] "<0.001" "NA"     "0.220"  "NA"    "0.154"  "0.038"  "0.023"  "NA"
      Female          106  90
# [9] "0.026" "<0.001"
Crew  Male            670 192
</pre>
      Female            3  20
 
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
=== dplr::mutate_if() ===
            Survived  No Yes
<pre>
Sex    Class               
library(dplyr)
Male  1st            118  62
df <- data.frame(
      2nd            154  25
  char_var = c("A", "B", "C"),
      3rd            422  88
  num_var1 = c(1.123456, 2.123456, 3.123456),
      Crew          670 192
  num_var2 = c(4.654321, 5.654321, 6.654321),
Female 1st              4 141
  stringsAsFactors = FALSE
      2nd            13  93
)
      3rd            106  90
 
      Crew            3  20
# Round numerical variables to 4 digits after the decimal point
> str(Titanic)
df_rounded <- df %>%
table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
  mutate_if(is.numeric, round, digits = 4)
- attr(*, "dimnames")=List of 4
</pre>
  ..$ Class  : chr [1:4] "1st" "2nd" "3rd" "Crew"
 
  ..$ Sex    : chr [1:2] "Male" "Female"
== Customize R: options() ==
  ..$ Age    : chr [1:2] "Child" "Adult"
 
  ..$ Survived: chr [1:2] "No" "Yes"
=== Change the default R repository, my .Rprofile ===
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
[[Rstudio#Change_repository|Change R repository]]
> x
 
          gear  3  4 5
Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local '''.Rprofile''' settings take precedence.
cyl vs am             
 
4   0  0        0  0  0
For example, I can specify the R mirror I like by creating a single line '''.Rprofile''' file under my home directory. Another good choice of repository is '''cloud.r-project.org'''.
      1        0  0  1
 
    1  0        1  2  0
Type '''file.edit("~/.Rprofile")'''
      1        0  6  1
{{Pre}}
6  0  0        0  0  0
local({
      1        0  2  1
  r = getOption("repos")
    1  0        2  2  0
   r["CRAN"] = "https://cran.rstudio.com/"
      1        0  0  0
   options(repos = r)
8  0  0      12  0  0
})
      1        0  0  2
options(continue = " ", editor = "nano")
    1  0        0  0  0
message("Hi MC, loading ~/.Rprofile")
      1        0  0  0
if (interactive()) {
> ftable(x, row.vars = c(2, 4))
  .Last <- function() try(savehistory("~/.Rhistory"))
        cyl  4    6    8    
}
        am   0  1  0  1  0  1
</pre>
vs gear                     
 
0 3        0  0  0  0 12  0
=== Change the default web browser for utils::browseURL() ===
  4        0  0  0  2  0  0
When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put
  5        0  1  0  1  0  2
<pre>
1  3        1  0  2  0  0  0
options(browser='seamonkey')
  4        2  6  2  0  0  0
</pre>
  5        0  1  0  0  0  0
in the '''.Rprofile''' of your home directory. If the browser is not in the global PATH, we need to put the full path above.
>
 
> ## Start with expressions, use table()'s "dnn" to change labels
For one-time only purpose, we can use the ''browser'' option in help.start() function:
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
{{Pre}}
        dnn = c("Cylinders", "V/S", "Transmission", "Gears"))
> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...
</pre>
 
We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See
* [https://stat.ethz.ch/pipermail/r-help/2003-August/037484.html Changing default browser in options()].
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/browseURL.html


          Cylinders    4    6    8 
=== Change the default editor ===
          Transmission  0  1  0  1  0  1
On my Linux and mac, the default editor is "vi". To change it to "nano",
V/S Gears                             
{{Pre}}
0  3                  0  0  0  0 12  0
options(editor = "nano")
    4                  0  0  0  2  0  0
</pre>
    5                  0  1  0  1  0  2
1  3                  1  0  2  0  0  0
    4                  2  6  2  0  0  0
    5                  0  1  0  0  0  0
</syntaxhighlight>


==== [https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/addmargins addmargins] ====
=== Change prompt and remove '+' sign ===
Puts Arbitrary Margins On Multidimensional Tables Or Arrays
See https://stackoverflow.com/a/1448823.
{{Pre}}
options(prompt="R> ", continue=" ")
</pre>


=== tracemem, data type, copy ===
=== digits ===
[http://stackoverflow.com/questions/18359940/r-programming-vector-a1-2-avoid-copying-the-whole-vector/18361181#18361181 How to avoid copying a long vector]
* [https://gist.github.com/arraytools/26a0b359541f4fc9fddc8f0a0c94489e Read and compute the sum of a numeric matrix file] using R vs Python vs C++. Note by default R does not show digits after the decimal point because the number is large.
 
* [https://stackoverflow.com/a/2288013 Controlling number of decimal digits in print output in R]
=== Tell if the current R is running in 32-bit or 64-bit mode ===
* [https://stackoverflow.com/a/10712012 ?print.default]
* [https://stackoverflow.com/a/12135122 Formatting Decimal places in R, round()]. [https://www.rdocumentation.org/packages/base/versions/3.5.3/topics/format format()] where '''nsmall''' controls the minimum number of digits to the right of the decimal point
* [https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17668 numerical error in round() causing round to even to fail] 2019-12-05
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Round signif()] rounds x to n significant digits.
<pre>
<pre>
8 * .Machine$sizeof.pointer
R> signif(pi, 3)
[1] 3.14
R> signif(pi, 5)
[1] 3.1416
</pre>
</li>
</ul>
* The default digits 7 may be too small. For example, '''if a number is very large, then we may not be able to see (enough) value after the decimal point'''. The acceptable range is 1-22. See the following examples
 
In R,
{{Pre}}
> options()$digits # Default
[1] 7
> print(.1+.2, digits=18)
[1] 0.300000000000000044
> 100000.07 + .04
[1] 100000.1
> options(digits = 16)
> 100000.07 + .04
[1] 100000.11
</pre>
</pre>
where '''sizeof.pointer''' returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.


=== 32- and 64-bit ===
In Python,
See [http://cran.r-project.org/doc/manuals/R-admin.html#Choosing-between-32_002d-and-64_002dbit-builds R-admin.html].
{{Pre}}
* For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
>>> 100000.07 + .04
* Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
100000.11
* Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).
</pre>


=== Handling length 2^31 and more in R 3.0.0 ===
=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing]: options(scipen) ===
[https://datasciencetut.com/how-to-turn-off-scientific-notation-in-r/ How to Turn Off Scientific Notation in R?]


From R News for 3.0.0 release:
This also helps with write.table() results. For example, 0.0003 won't become 3e-4 in the output file.
{{Pre}}
> numer = 29707; denom = 93874
> c(numer/denom, numer, denom)
[1] 3.164561e-01 2.970700e+04 9.387400e+04


''There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.  
# Method 1. Without changing the global option
''
> format(c(numer/denom, numer, denom), scientific=FALSE)
[1] "    0.3164561" "29707.0000000" "93874.0000000"


In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error
# Method 2. Change the global option
<pre>
> options(scipen=999)
> x <- seq(1, 2^31)
> numer/denom
Error in from:to : result would be too long a vector
[1] 0.3164561
> c(numer/denom, numer, denom)
[1]    0.3164561 29707.0000000 93874.0000000
> c(4/5, numer, denom)
[1]    0.8 29707.0 93874.0
</pre>
</pre>


However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):
=== Suppress warnings: options() and capture.output() ===
<pre>
Use [https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/options options()]. If ''warn'' is negative all warnings are ignored. If ''warn'' is zero (the default) warnings are stored until the top--level function returns.  
> system.time(x <- seq(1,2^31))
{{Pre}}
  user  system elapsed
op <- options("warn")
  8.604  11.060 120.815
options(warn = -1)
> length(x)
....
[1] 2147483648
options(op)
> length(x)/2^20
 
[1] 2048
# OR
> gc()
warnLevel <- options()$warn
            used    (Mb) gc trigger    (Mb)  max used    (Mb)
options(warn = -1)
Ncells    183823    9.9    407500    21.8    350000    18.7
...
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
options(warn = warnLevel)
>
</pre>
</pre>
Note:
# 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
# On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
# My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
# For large dataset, it makes sense to use database or specially crafted packages like [http://cran.r-project.org/web/packages/bigmemory/ bigmemory] or [http://cran.r-project.org/web/packages/ff/ ff] or [https://privefl.github.io/bigstatsr/ bigstatsr].


=== NA in index ===
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/warning suppressWarnings()]
* Question: what is seq(1, 3)[c(1, 2, NA)]?
<pre>
suppressWarnings( foo() )


Answer: It will reserve the element with NA in indexing and return the value NA for it.
foo <- capture.output(
bar <- suppressWarnings(
{print( "hello, world" );
  warning("unwanted" )} ) )
</pre>


* Question: What is TRUE & NA?
[https://www.rdocumentation.org/packages/utils/versions/3.6.2/topics/capture.output capture.output()]
Answer: NA
<pre>
str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")
</pre>


* Question: What is FALSE & NA?
=== Converts warnings into errors ===
Answer: FALSE
options(warn=2)


* Question: c("A", "B", NA) != "" ?
=== demo() function ===
Answer: TRUE TRUE NA
<ul>
<li>[https://stackoverflow.com/a/18746519 How to wait for a keypress in R?] PS [https://stat.ethz.ch/R-manual/R-devel/library/base/html/readline.html readline()] is different from readLines().
<pre>
for(i in 1:2) { print(i); readline("Press [enter] to continue")}
</pre>
<li>Hit 'ESC' or Ctrl+c to skip the prompt "Hit <Return> to see next plot:" </li>
<li>demo() uses [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/options options()] to ask users to hit Enter on each plot
<pre>
op <- options(device.ask.default = ask)  # ask = TRUE
on.exit(options(op), add = TRUE)
</pre>
</li>
</ul>


* Question: which(c("A", "B", NA) != "") ?
== sprintf ==
Answer: 1 2
=== paste, paste0, sprintf ===
[https://www.r-bloggers.com/paste-paste0-and-sprintf/ this post], [https://www.r-bloggers.com/2023/09/3-r-functions-that-i-enjoy/ 3 R functions that I enjoy]


* Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?
=== sep vs collapse in paste() ===
Answer: TRUE TRUE FALSE
* sep is used if we supply '''multiple separate objects''' to paste(). A more powerful function is [https://tidyr.tidyverse.org/reference/unite.html tidyr::unite()] function.
* collapse is used to make the output of length 1. It is commonly used if we have only 1 input object
<pre>
R> paste("a", "A", sep=",") # multi-vec -> multi-vec
[1] "a,A"
R> paste(c("Elon", "Taylor"), c("Mask", "Swift"))
[1] "Elon Mask"    "Taylor Swift"
# OR
R> sprintf("%s, %s", c("Elon", "Taylor"), c("Mask", "Swift"))


* Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?
R> paste(c("a", "A"), collapse="-") # one-vec/multi-vec  -> one-scale
Answer: TRUE TRUE FALSE
[1] "a-A"


'''Conclusion''': In order to exclude empty or NA for numerical or character data type, we can use '''which()''' or a convenience function '''keep.complete(x) <- function(x) x != "" & !is.na(x)'''. This will guarantee return logical values and not contain NAs.
# When use together, sep first and collapse second
R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
[1] "a,A - b,B - c,C"
R> paste(letters[1:3], LETTERS[1:3], sep=",")
[1] "a,A" "b,B" "c,C"
R> paste(letters[1:3], LETTERS[1:3], sep=",") |> paste(collapse=" - ")
[1] "a,A - b,B - c,C"
</pre>


Don't just use x != "" OR !is.na(x).
=== Format number as fixed width, with leading zeros ===
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1


=== Constant ===
{{Pre}}
Add 'L' after a constant. For example,
# sprintf()
<syntaxhighlight lang='rsplus'>
a <- seq(1,101,25)
for(i in 1L:n) { }
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"


if (max.lines > 0L) { }
# formatC()
paste("name", formatC(a, width=3, flag="0"), sep="_")
[1] "name_001" "name_026" "name_051" "name_076" "name_101"


label <- paste0(n-i+1L, ": ")
# gsub()
paste0("bm", gsub(" ", "0", format(5:15)))
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"
</pre>


n <- length(x); if(n == 0L) { }
=== formatC and prettyNum (prettifying numbers) ===
</syntaxhighlight>
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/formatC formatC() & prettyNum()]
* [[R#format.pval|format.pval()]]
<pre>
R> (x <- 1.2345 * 10 ^ (-8:4))
[1] 1.2345e-08 1.2345e-07 1.2345e-06 1.2345e-05 1.2345e-04 1.2345e-03
[7] 1.2345e-02 1.2345e-01 1.2345e+00 1.2345e+01 1.2345e+02 1.2345e+03
[13] 1.2345e+04
R> formatC(x)
[1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "0.0001234" "0.001234"
  [7] "0.01235"  "0.1235"    "1.234"    "12.34"    "123.4"    "1234"
[13] "1.234e+04"
R> formatC(x, digits=3)
[1] "1.23e-08" "1.23e-07" "1.23e-06" "1.23e-05" "0.000123" "0.00123"
[7] "0.0123"  "0.123"    "1.23"    "12.3"    " 123"    "1.23e+03"
[13] "1.23e+04"
R> formatC(x, digits=3, format="e")
[1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "1.234e-04" "1.234e-03"
[7] "1.235e-02" "1.235e-01" "1.234e+00" "1.234e+01" "1.234e+02" "1.234e+03"
[13] "1.234e+04"


=== Data frame ===
R> x <- .000012345
* http://blog.datacamp.com/15-easy-solutions-data-frame-problems-r/
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"
</pre>


==== stringsAsFactors = FALSE ====
=== format(x, scientific = TRUE) vs round() vs format.pval() ===
http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/
Print numeric data in exponential format, so .0001 prints as 1e-4
<syntaxhighlight lang='r'>
format(c(0.00001156, 0.84134, 2.1669), scientific = T, digits=4)
# [1] "1.156e-05" "8.413e-01" "2.167e+00"
round(c(0.00001156, 0.84134, 2.1669), digits=4)
# [1] 0.0000 0.8413 2.1669


==== Convert data frame factor columns to characters ====
format.pval(c(0.00001156, 0.84134, 2.1669)) # output is char vector
[https://stackoverflow.com/questions/2851015/convert-data-frame-columns-from-factors-to-characters Convert data.frame columns from factors to characters]
# [1] "1.156e-05" "0.84134"  "2.16690"
<syntaxhighlight lang='rsplus'>
format.pval(c(0.00001156, 0.84134, 2.1669), digits=4)
# Method 1:
# [1] "1.156e-05" "0.8413"    "2.1669"
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)
 
# Method 2:
bob[] <- lapply(bob, as.character)
</syntaxhighlight>
</syntaxhighlight>


==== data.frame to vector ====
== Creating publication quality graphs in R ==
<syntaxhighlight lang='rsplus'>
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
> a= matrix(1:6, 2,3)
> rownames(a) <- c("a", "b")
> colnames(a) <- c("x", "y", "z")
> a
  x y z
a 1 3 5
b 2 4 6
> unlist(data.frame(a))
x1 x2 y1 y2 z1 z2
1  2  3  4  5  6
</syntaxhighlight>


==== merge ====
== HDF5 : Hierarchical Data Format==
[https://jozefhajnala.gitlab.io/r/r006-merge/ How to perform merges (joins) on two or more data frames with base R, tidyverse and data.table]
HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.


==== matrix vs data.frame ====
* https://en.wikipedia.org/wiki/Hierarchical_Data_Format
<syntaxhighlight lang='rsplus'>
* [https://support.hdfgroup.org/HDF5/ HDF5 tutorial] and others
ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
unique(ip1$Priority)
* rhdf5 is used by [http://amp.pharm.mssm.edu/archs4/data.html ARCHS4] where you can download R program that will download hdf5 file storing expression and metadata such as gene ID, sample/GSM ID, tissues, et al.
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])  # OK


ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
== Formats for writing/saving and sharing data ==
unique(ip2$Priority)    # OK
[http://www.econometricsbysimulation.com/2016/12/efficiently-saving-and-sharing-data-in-r_46.html Efficiently Saving and Sharing Data in R]
</syntaxhighlight>


=== matrix multiply a vector ===
== Write unix format files on Windows and vice versa ==
* [https://en.wikipedia.org/wiki/Row-_and_column-major_order#Programming_languages_and_libraries R (like Fortran) is following the column-major order]
https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html


<syntaxhighlight lang='rsplus'>
== with() and within() functions ==
> matrix(1:6, 3,2)
* [https://www.r-bloggers.com/2023/07/simplify-your-code-with-rs-powerful-functions-with-and-within/ Simplify Your Code with R’s Powerful Functions: with() and within()]
    [,1] [,2]
* within() is similar to with() except it is used to create new columns and merge them with the original data sets. But if we just want to create a new column, we can just use df$newVar = . The following example is from [http://www.youtube.com/watch?v=pZ6Bnxg9E8w&list=PLOU2XLYxmsIK9qQfztXeybpHvru-TrqAP youtube video].
[1,]    1    4
<pre>
[2,]    2    5
closePr <- with(mariokart, totalPr - shipPr)
[3,]    3    6
head(closePr, 20)
> matrix(1:6, 3,2) * c(1,2,3)
    [,1] [,2]
[1,]    1    4
[2,]    4  10
[3,]    9  18
> matrix(1:6, 3,2) * c(1,2,3,4)
    [,1] [,2]
[1,]    1  16
[2,]    4    5
[3,]    9  12
</syntaxhighlight>


=== Print a vector by suppressing names ===
mk <- within(mariokart, {
Use '''unname'''.
            closePr <- totalPr - shipPr
    })
head(mk) # new column closePr


=== format.pval ===
mk <- mariokart
<syntaxhighlight lang='rsplus'>
aggregate(. ~ wheels + cond, mk, mean)
> args(format.pval)
# create mean according to each level of (wheels, cond)
function (pv, digits = max(1L, getOption("digits") - 2L), eps = .Machine$double.eps,
    na.form = "NA", ...)  


> format.pval(c(stats::runif(5), pi^-100, NA))
aggregate(totalPr ~ wheels + cond, mk, mean)
[1] "0.19571" "0.46793" "0.71696" "0.93200" "0.74485" "< 2e-16" "NA"   
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04"  "<2e-16"
</syntaxhighlight>


=== [https://stackoverflow.com/questions/5352099/how-to-disable-scientific-notation Disable scientific notation in printing] ===
tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)
<syntaxhighlight lang='rsplus'>
</pre>
> numer = 29707; denom = 93874
> c(numer/denom, numer, denom)  
[1] 3.164561e-01 2.970700e+04 9.387400e+04


# Method 1. Without changing the global option
== stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals ==
> format(c(numer/denom, numer, denom), scientific=FALSE)
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
[1] "    0.3164561" "29707.0000000" "93874.0000000"
* [https://www.dataanalytics.org.uk/tally-plots-in-r/ Tally plots in R]
* https://stackoverflow.com/questions/14736556/ascii-plotting-functions-for-r
* [https://cran.r-project.org/web/packages/txtplot/index.html txtplot] package


# Method 2. Change the global option
== Plot histograms as lines ==
> options(scipen=999)
https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.  
> numer/denom
<pre>
[1] 0.3164561
x2=invisible(hist(out2$EB))
> c(numer/denom, numer, denom)
y2=invisible(hist(out2$Bench))
[1]    0.3164561 29707.0000000 93874.0000000
z2=invisible(hist(out2$EB0.001))
> c(4/5, numer, denom)
[1]    0.8 29707.0 93874.0
</syntaxhighlight>


=== sprintf ===
plot(x=x2$mids, y=x2$density, type="l")
==== Format number as fixed width, with leading zeros ====
lines(y2$mids, y2$density, lty=2, pwd=2)
* https://stackoverflow.com/questions/8266915/format-number-as-fixed-width-with-leading-zeros
lines(z2$mids, z2$density, lty=3, pwd=2)
* https://stackoverflow.com/questions/14409084/pad-with-leading-zeros-to-common-width?rq=1
</pre>


<syntaxhighlight lang='rsplus'>
== Histogram with density line ==
# sprintf()
<pre>
a <- seq(1,101,25)
hist(x, prob = TRUE)
sprintf("name_%03d", a)
lines(density(x), col = 4, lwd = 2)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
</pre>
The overlayed density may looks strange in cases for example counts from single-cell RNASeq or p-values from RNASeq (there is a peak around x=0).


# formatC()
== Graphical Parameters, Axes and Text, Combining Plots ==
paste("name", formatC(a, width=3, flag="0"), sep="_")
[http://www.statmethods.net/advgraphs/axes.html statmethods.net]
[1] "name_001" "name_026" "name_051" "name_076" "name_101"
</syntaxhighlight>


==== sprintf does not print ====
== 15 Questions All R Users Have About Plots ==
Use cat() or print() outside sprintf(). sprintf() do not print in a non interactive mode.
See [https://www.datacamp.com/tutorial/15-questions-about-r-plots 15 Questions All R Users Have About Plots]. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.
<syntaxhighlight lang='rsplus'>
cat(sprintf('%5.2f\t%i\n',1.234, l234))
</syntaxhighlight>


=== Creating publication quality graphs in R ===
# How To Draw An Empty R Plot? plot.new()
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
# How To Set The Axis Labels And Title Of The R Plots?
 
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()  
=== HDF5 : Hierarchical Data Format===
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext(). [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par ?par].
HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.
# How To Add Or Change The R Plot’s Legend? legend()
 
# How To Draw A Grid In Your R Plot? [https://r-charts.com/base-r/grid/ grid()]
* https://en.wikipedia.org/wiki/Hierarchical_Data_Format
* [https://support.hdfgroup.org/HDF5/ HDF5 tutorial] and others
* [http://www.bioconductor.org/packages/release/bioc/html/rhdf5.html rhdf5] package
* rhdf5 is used by [http://amp.pharm.mssm.edu/archs4/data.html ARCHS4] where you can download R program that will download hdf5 file storing expression and metadata such as gene ID, sample/GSM ID, tissues, et al.
 
<syntaxhighlight lang='rsplus'>
> h5ls(destination_file)
  group                          name      otype  dclass          dim
0      /                          data  H5I_GROUP                     
1  /data                    expression H5I_DATASET INTEGER 35238 x 65429
2      /                          info  H5I_GROUP                     
3  /info                        author H5I_DATASET  STRING            1
4  /info                        contact H5I_DATASET  STRING            1
5  /info                  creation-date H5I_DATASET  STRING            1
6  /info                            lab H5I_DATASET  STRING            1
7  /info                        version H5I_DATASET  STRING            1
8      /                          meta  H5I_GROUP                     
9  /meta          Sample_channel_count H5I_DATASET  STRING        65429
10 /meta    Sample_characteristics_ch1 H5I_DATASET  STRING        65429
11 /meta        Sample_contact_address H5I_DATASET  STRING        65429
12 /meta            Sample_contact_city H5I_DATASET  STRING        65429
13 /meta        Sample_contact_country H5I_DATASET  STRING        65429
14 /meta      Sample_contact_department H5I_DATASET  STRING        65429
15 /meta          Sample_contact_email H5I_DATASET  STRING        65429
16 /meta      Sample_contact_institute H5I_DATASET  STRING        65429
17 /meta      Sample_contact_laboratory H5I_DATASET  STRING        65429
18 /meta            Sample_contact_name H5I_DATASET  STRING        65429
19 /meta          Sample_contact_phone H5I_DATASET  STRING        65429
20 /meta Sample_contact_zip-postal_code H5I_DATASET  STRING        65429
21 /meta        Sample_data_processing H5I_DATASET  STRING        65429
22 /meta          Sample_data_row_count H5I_DATASET  STRING        65429
23 /meta            Sample_description H5I_DATASET  STRING        65429
24 /meta    Sample_extract_protocol_ch1 H5I_DATASET  STRING        65429
25 /meta          Sample_geo_accession H5I_DATASET  STRING        65429
26 /meta        Sample_instrument_model H5I_DATASET  STRING        65429
27 /meta        Sample_last_update_date H5I_DATASET  STRING        65429
28 /meta      Sample_library_selection H5I_DATASET  STRING        65429
29 /meta          Sample_library_source H5I_DATASET  STRING        65429
30 /meta        Sample_library_strategy H5I_DATASET  STRING        65429
31 /meta            Sample_molecule_ch1 H5I_DATASET  STRING        65429
32 /meta            Sample_organism_ch1 H5I_DATASET  STRING        65429
33 /meta            Sample_platform_id H5I_DATASET  STRING        65429
34 /meta                Sample_relation H5I_DATASET  STRING        65429
35 /meta              Sample_series_id H5I_DATASET  STRING        65429
36 /meta        Sample_source_name_ch1 H5I_DATASET  STRING        65429
37 /meta                  Sample_status H5I_DATASET  STRING        65429
38 /meta        Sample_submission_date H5I_DATASET  STRING        65429
39 /meta    Sample_supplementary_file_1 H5I_DATASET  STRING        65429
40 /meta    Sample_supplementary_file_2 H5I_DATASET  STRING        65429
41 /meta              Sample_taxid_ch1 H5I_DATASET  STRING        65429
42 /meta                  Sample_title H5I_DATASET  STRING        65429
43 /meta                    Sample_type H5I_DATASET  STRING        65429
44 /meta                          genes H5I_DATASET  STRING        35238
</syntaxhighlight>
 
=== Formats for writing/saving and sharing data ===
[http://www.econometricsbysimulation.com/2016/12/efficiently-saving-and-sharing-data-in-r_46.html Efficiently Saving and Sharing Data in R]
 
=== Write unix format files on Windows and vice versa ===
https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html
 
=== with() and within() functions ===
within() is similar to with() except it is used to create new columns and merge them with the original data sets. See [http://www.youtube.com/watch?v=pZ6Bnxg9E8w&list=PLOU2XLYxmsIK9qQfztXeybpHvru-TrqAP youtube video].
<pre>
closePr <- with(mariokart, totalPr - shipPr)
head(closePr, 20)
 
mk <- within(mariokart, {
            closePr <- totalPr - shipPr
    })
head(mk) # new column closePr
 
mk <- mariokart
aggregate(. ~ wheels + cond, mk, mean)
# create mean according to each level of (wheels, cond)
 
aggregate(totalPr ~ wheels + cond, mk, mean)
 
tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)
</pre>
 
=== stem(): stem-and-leaf plot, bar chart on terminals ===
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
* https://stackoverflow.com/questions/14736556/ascii-plotting-functions-for-r
* [https://cran.r-project.org/web/packages/txtplot/index.html txtplot] package
 
=== Graphical Parameters, Axes and Text, Combining Plots ===
[http://www.statmethods.net/advgraphs/axes.html statmethods.net]
 
=== 15 Questions All R Users Have About Plots ===
See http://blog.datacamp.com/15-questions-about-r-plots/. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.
 
# How To Draw An Empty R Plot? plot.new()
# How To Set The Axis Labels And Title Of The R Plots?
# How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()  
# How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext()
# How To Add Or Change The R Plot’s Legend? legend()
# How To Draw A Grid In Your R Plot? grid()  
# How To Draw A Plot With A PNG As Background? rasterImage() from the '''png''' package
# How To Draw A Plot With A PNG As Background? rasterImage() from the '''png''' package
# How To Adjust The Size Of Points In An R Plot? cex argument
# How To Adjust The Size Of Points In An R Plot? cex argument
Line 5,436: Line 5,603:
# How To Add Error Bars In An R Plot? arrows()
# How To Add Error Bars In An R Plot? arrows()
# How To Save A Plot As An Image On Disc  
# How To Save A Plot As An Image On Disc  
# How To Plot Two R Plots Next To Each Other? par(mfrow), '''gridBase''' package, '''lattice''' package
# How To Plot Two R Plots Next To Each Other? '''par(mfrow)'''[which means Multiple Figures (use ROW-wise)], '''gridBase''' package, '''lattice''' package
# How To Plot Multiple Lines Or Points? plot(), lines()
# How To Plot Multiple Lines Or Points? plot(), lines()
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
# How To Fix The Aspect Ratio For Your R Plots? asp parameter
# What Is The Function Of hjust And vjust In ggplot2?
# What Is The Function Of hjust And vjust In ggplot2?


=== jitter function ===
== jitter function ==
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
* [https://statistical-programming.com/jitter-r-function-example/ The jitter R Function] 3 Example Codes (Basic Application & Boxplot Visualization)
** jitter(, amount) function adds a random variation between -amount/2 and amount/2 to each element in x
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
* [https://stackoverflow.com/a/17552046 What does the “jitter” function do in R?]
* [https://www.r-bloggers.com/2023/09/when-to-use-jitter/ When to use Jitter]
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]
* [https://stats.stackexchange.com/a/146174 How to calculate Area Under the Curve (AUC), or the c-statistic, by hand]


=== Scatterplot with the "rug" function ===
:[[File:Jitterbox.png|200px]]
 
== Scatterplot with the "rug" function ==
<pre>
<pre>
require(stats)  # both 'density' and its default method
require(stats)  # both 'density' and its default method
Line 5,456: Line 5,626:
})
})
</pre>
</pre>
[[File:RugFunction.png|200px]]
[[:File:RugFunction.png]]


See also the [https://stat.ethz.ch/R-manual/R-devel/library/graphics/html/stripchart.html stripchart()] function which produces one dimensional scatter plots (or dot plots) of the given data.
See also the [https://stat.ethz.ch/R-manual/R-devel/library/graphics/html/stripchart.html stripchart()] function which produces one dimensional scatter plots (or dot plots) of the given data.


=== Identify/Locate Points in a Scatter Plot ===
== Identify/Locate Points in a Scatter Plot ==
[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]
<ul>
<li>[https://www.rdocumentation.org/packages/graphics/versions/3.5.1/topics/identify ?identify]
<li>[https://stackoverflow.com/a/23234142 Using the identify function in R]
<pre>
plot(x, y)
identify(x, y, labels = names, plot = TRUE)
# Use left clicks to select points we want to identify and "esc" to stop the process
# This will put the labels on the plot and also return the indices of points
# [1] 143
names[143]
</pre>
</ul>


=== Draw a single plot with two different y-axes ===
== Draw a single plot with two different y-axes ==
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html


=== Draw Color Palette ===
== Draw Color Palette ==
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/


=== SVG ===
=== Default palette before R 4.0 ===
==== Embed svg in html ====
palette() # black, red, green3, blue, cyan, magenta, yellow, gray
 
<pre>
# Example from Coursera "Statistics for Genomic Data Science" by Jeff Leek
tropical = c('darkorange', 'dodgerblue', 'hotpink', 'limegreen', 'yellow')
palette(tropical)
plot(1:5, 1:5, col=1:5, pch=16, cex=5)
</pre>
 
=== New palette in R 4.0.0 ===
[https://youtu.be/I4k0LkTOKvU?t=464 R 4.0: 3 new features], [https://blog.revolutionanalytics.com/2020/04/r-400-is-released.html R 4.0.0 now available, and a look back at R's history]. For example, we can select "ggplot2" palette to make the base graphics charts that match the color scheme of ggplot2.
<pre>
R> palette()
[1] "black"  "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "gray62"
R> palette.pals()
[1] "R3"              "R4"              "ggplot2"       
[4] "Okabe-Ito"      "Accent"          "Dark 2"       
[7] "Paired"          "Pastel 1"        "Pastel 2"     
[10] "Set 1"          "Set 2"          "Set 3"         
[13] "Tableau 10"      "Classic Tableau" "Polychrome 36" 
[16] "Alphabet"
R> palette.colors(palette='R4') # same as palette()
[1] "#000000" "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "#9E9E9E"
R> palette("R3")  # nothing return on screen but palette has changed
R> palette()
[1] "black"  "red"    "green3"  "blue"    "cyan"    "magenta" "yellow"
[8] "gray" 
R> palette("R4") # reset to the default color palette; OR palette("default")
 
R> scales::show_col(palette.colors(palette = "Okabe-Ito"))
R> for(id in palette.pals()) {
    scales::show_col(palette.colors(palette = id))
    title(id)
    readline("Press [enter] to continue")
  }
</pre>
The '''palette''' function can also be used to change the color palette. See [https://data.library.virginia.edu/setting-up-color-palettes-in-r/ Setting up Color Palettes in R]
<pre>
palette("ggplot2")
palette(palette()[-1]) # Remove 'black'
  # OR palette(palette.colors(palette = "ggplot2")[-1] )
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))
 
cc <- palette()
palette(c(cc,"purple","brown")) # Add two colors
</pre>
<pre>
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502
</pre>
 
=== evoPalette ===
[http://gradientdescending.com/evolve-new-colour-palettes-in-r-with-evopalette/ Evolve new colour palettes in R with evoPalette]
 
=== rtist ===
[https://github.com/tomasokal/rtist?s=09 rtist]: Use the palettes of famous artists in your own visualizations.
 
== SVG ==
=== Embed svg in html ===
* http://www.magesblog.com/2016/02/using-svg-graphics-in-blog-posts.html
* http://www.magesblog.com/2016/02/using-svg-graphics-in-blog-posts.html


==== svglite ====
=== svglite ===
https://blog.rstudio.org/2016/11/14/svglite-1-2-0/
svglite is better R's svg(). It was used by ggsave().
[https://www.rstudio.com/blog/svglite-1-2-0/ svglite 1.2.0], [https://r-graphics.org/recipe-output-vector-svg R Graphics Cookbook].


==== pdf -> svg ====
=== pdf -> svg ===
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].


=== read.table ===
=== svg -> png ===
==== clipboard ====
[https://laustep.github.io/stlahblog/posts/SVG2PNG.html SVG to PNG] using the [https://cran.rstudio.com/web/packages/gyro/index.html gyro] package
<syntaxhighlight lang="rsplus">
 
== read.table ==
=== clipboard ===
{{Pre}}
source("clipboard")
source("clipboard")
read.table("clipboard")
read.table("clipboard")
</syntaxhighlight>
</pre>


==== inline text ====
=== inline text ===
<syntaxhighlight lang="rsplus">
{{Pre}}
mydf <- read.table(header=T, text='
mydf <- read.table(header=T, text='
  cond yval
  cond yval
Line 5,494: Line 5,739:
     C 1.6
     C 1.6
')
')
</syntaxhighlight>
</pre>


==== http(s) connection ====
=== http(s) connection ===
<syntaxhighlight lang="rsplus">
{{Pre}}
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",  
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",  
                           ssl.verifypeer = FALSE)
                           ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)
ip <- read.table(textConnection(temp), as.is=TRUE)
</syntaxhighlight>
</pre>


==== read only specific columns ====
=== read only specific columns ===
Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.
Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.
<syntaxhighlight lang="rsplus">
{{Pre}}
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),  
x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)),  
                 skip=62, header=T, stringsAsFactors = FALSE)[, 1]
                 skip=62, header=T, stringsAsFactors = FALSE)[, 1]
Line 5,511: Line 5,756:
system.time(x <- read.delim("Methylation450k.txt",  
system.time(x <- read.delim("Methylation450k.txt",  
                 colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
                 colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))
</syntaxhighlight>
</pre>


To know the number of columns, we might want to read the first row first.
To know the number of columns, we might want to read the first row first.
<syntaxhighlight lang="rsplus">
{{Pre}}
library(magrittr)
library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
</syntaxhighlight>
</pre>


Another method is to use '''pipe()''', '''cut''' or '''awk'''. See [https://stackoverflow.com/questions/2193742/ways-to-read-only-select-columns-from-a-file-into-r-a-happy-medium-between-re ways to read only selected columns from a file into R]
Another method is to use '''pipe()''', '''cut''' or '''awk'''. See [https://stackoverflow.com/questions/2193742/ways-to-read-only-select-columns-from-a-file-into-r-a-happy-medium-between-re ways to read only selected columns from a file into R]


=== Serialization ===
=== check.names = FALSE in read.table() ===
<pre>
gx <- read.table(file, header = T, row.names =1)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH."        "IGHV1.69"    "IGKV1.5"    "IGKV2.24"    "KRTAP13.2" 
# [7] "KRTAP19.1"  "KRTAP2.4"    "KRTAP5.9"    "KRTAP6.3"    "Kua.UEV" 
 
gx <- read.table(file, header = T, row.names =1, check.names = FALSE)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH@"        "IGHV1-69"    "IGKV1-5"    "IGKV2-24"    "KRTAP13-2" 
# [7] "KRTAP19-1"  "KRTAP2-4"    "KRTAP5-9"    "KRTAP6-3"    "Kua-UEV" 
</pre>
 
=== setNames() ===
Change the colnames. See an example from [https://www.tidymodels.org/start/models/ tidymodels]
 
=== Testing for valid variable names ===
[https://www.r-bloggers.com/testing-for-valid-variable-names/ Testing for valid variable names]
 
=== make.names(): Make syntactically valid names out of character vectors ===
* [https://stat.ethz.ch/R-manual/R-devel/library/base/html/make.names.html make.names()]
* A valid variable name consists of letters, numbers and the '''dot''' or '''underline''' characters. The variable name starts with a letter or the dot not followed by a number. See [https://www.tutorialspoint.com/r/r_variables.htm R variables].
<pre>
make.names("abc-d") # [1] "abc.d"
</pre>
 
== Serialization ==
If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the  
If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the  
[https://stat.ethz.ch/pipermail/r-devel/attachments/20130628/56473803/attachment.pl post] on R mailing list.
[https://stat.ethz.ch/pipermail/r-devel/attachments/20130628/56473803/attachment.pl post] on R mailing list.
Line 5,536: Line 5,807:
then read <length> bytes from the connection.
then read <length> bytes from the connection.


=== socketConnection ===
== socketConnection ==
See ?socketconnection.  
See ?socketconnection.  


==== Simple example ====
=== Simple example ===
from the socketConnection's manual.
from the socketConnection's manual.


Line 5,562: Line 5,833:
</pre>
</pre>


==== Use nc in client ====
=== Use nc in client ===


The client does not have to be the R. We can use telnet, nc, etc. See the post [https://stat.ethz.ch/pipermail/r-sig-hpc/2009-April/000144.html here]. For example, on the client machine, we can issue
The client does not have to be the R. We can use telnet, nc, etc. See the post [https://stat.ethz.ch/pipermail/r-sig-hpc/2009-April/000144.html here]. For example, on the client machine, we can issue
Line 5,578: Line 5,849:
Some post I don't have a chance to read. http://digitheadslabnotebook.blogspot.com/2010/09/how-to-send-http-put-request-from-r.html
Some post I don't have a chance to read. http://digitheadslabnotebook.blogspot.com/2010/09/how-to-send-http-put-request-from-r.html


==== Use curl command in client ====
=== Use curl command in client ===
On the server,
On the server,
<pre>
<pre>
Line 5,596: Line 5,867:
</pre>
</pre>


==== Use telnet command in client ====
=== Use telnet command in client ===
On the server,
On the server,
<pre>
<pre>
Line 5,621: Line 5,892:
Some [http://blog.gahooa.com/2009/01/23/basics-of-telnet-and-http/ tutorial] about using telnet on http request. And [http://unixhelp.ed.ac.uk/tables/telnet_commands.html this] is a summary of using telnet.
Some [http://blog.gahooa.com/2009/01/23/basics-of-telnet-and-http/ tutorial] about using telnet on http request. And [http://unixhelp.ed.ac.uk/tables/telnet_commands.html this] is a summary of using telnet.


=== Subsetting ===
== Subsetting ==
[http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Subset-assignment Subset assignment of R Language Definition] and [http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Manipulation-of-functions Manipulation of functions].
[http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Subset-assignment Subset assignment of R Language Definition] and [http://lib.stat.cmu.edu/R/CRAN/doc/manuals/R-lang.html#Manipulation-of-functions Manipulation of functions].


Line 5,631: Line 5,902:
</pre>
</pre>


==== Avoid Coercing Indices To Doubles ====
=== Avoid Coercing Indices To Doubles ===
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]


=== as.formula() ===
=== Careful on NA value ===
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
See the example below. base::subset() or dplyr::filter() can remove NA subsets.
<syntaxhighlight lang='rsplus'>
<pre>
R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3  3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6
</pre>
 
=== Implicit looping ===
<pre>
set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]    # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7
</pre>
 
== modelling ==
=== update() ===
* [https://www.rdocumentation.org/packages/stats/versions/3.6.1/topics/update ?update]
* [https://stackoverflow.com/a/5118337 Reusing a Model Built in R]
 
=== Extract all variable names in lm(), glm(), ... ===
all.vars(formula(Model)[-2])
 
=== as.formula(): use a string in formula in lm(), glm(), ... ===
* [https://www.r-bloggers.com/2019/08/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula]
* [https://stackoverflow.com/questions/5251507/how-to-succinctly-write-a-formula-with-many-variables-from-a-data-frame How to succinctly write a formula with many variables from a data frame?]
{{Pre}}
? as.formula
? as.formula
xnam <- paste("x", 1:25, sep="")
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</syntaxhighlight>
</pre>
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
* [http://www.win-vector.com/blog/2018/09/r-tip-how-to-pass-a-formula-to-lm/ How to Pass A formula to lm], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/bquote ?bquote], [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/eval ?eval]
<syntaxhighlight lang='rsplus'>
{{Pre}}
outcome <- "mpg"
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")
variables <- c("cyl", "disp", "hp", "carb")
Line 5,690: Line 6,001:
(Intercept)          cyl        disp          hp        carb   
(Intercept)          cyl        disp          hp        carb   
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
   34.021595    -1.048523    -0.026906    0.009349    -0.926863  
</syntaxhighlight>
</pre>
* [https://statisticaloddsandends.wordpress.com/2019/08/24/changing-the-variable-inside-an-r-formula/ Changing the variable inside an R formula] 1. as.formula() 2. subset by [[i]] 3. get() 4. eval(parse()).
 
=== reformulate ===
[https://www.r-bloggers.com/2023/06/simplifying-model-formulas-with-the-r-function-reformulate/ Simplifying Model Formulas with the R Function ‘reformulate()’]
 
=== I() function ===
I() means isolates. See [https://stackoverflow.com/a/24192745 What does the capital letter "I" in R linear regression formula mean?],  [https://stackoverflow.com/a/8055683 In R formulas, why do I have to use the I() function on power terms, like y ~ I(x^3)]
 
=== Aggregating results from linear model ===
https://stats.stackexchange.com/a/6862
 
== Replacement function "fun(x) <- a" ==
[https://stackoverflow.com/questions/11563154/what-are-replacement-functions-in-r What are Replacement Functions in R?]
<pre>
R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
    x[x > value] <- Inf
    x
}
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]  1  3 Inf Inf
 
R> "cutoff<-"(x = xx, value = 65)
[1]  1  3 Inf Inf
</pre>
The statement '''fun(x) <- a''' and R will read '''x <- "fun<-"(x,a) '''


=== S3 and S4 methods ===
== S3 and S4 methods and signature ==
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* How S4 works in R https://www.rdocumentation.org/packages/methods/versions/3.5.1/topics/Methods_Details
* Software for Data Analysis: Programming with R by John Chambers
* Software for Data Analysis: Programming with R by John Chambers
* Programming with Data: A Guide to the S Language  by John Chambers
* Programming with Data: A Guide to the S Language  by John Chambers
* [https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers, 2016
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
* https://www.rmetrics.org/files/Meielisalp2009/Presentations/Chalabi1.pdf
* [https://njtierney.github.io/r/missing%20data/rbloggers/2016/11/06/simple-s3-methods/ A Simple Guide to S3 Methods]
* [https://rstudio-education.github.io/hopr/s3.html Hands-On Programming with R] by Garrett Grolemund
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
* https://www.stat.auckland.ac.nz/S-Workshop/Gentleman/S4Objects.pdf
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package]
* [http://cran.r-project.org/web/packages/packS4/index.html packS4: Toy Example of S4 Package], * [https://cran.r-project.org/doc/contrib/Genolini-S4tutorialV0-5en.pdf A (Not So) Short Introduction to S4]
* http://www.cyclismo.org/tutorial/R/s4Classes.html
* http://www.cyclismo.org/tutorial/R/s4Classes.html
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* [https://cran.r-project.org/doc/contrib/Genolini-S4tutorialV0-5en.pdf A (Not So) Short Introduction to S4]
* http://adv-r.had.co.nz/S4.html, http://adv-r.had.co.nz/OO-essentials.html
* http://adv-r.had.co.nz/S4.html
* [https://appsilon.com/object-oriented-programming-in-r-part-1/ Object-Oriented Programming in R (Part 1): An Introduction], [https://appsilon.com/object-oriented-programming-in-r-part-2/ Part 2: S3 Simplified]


To get the source code of S4 methods, we can use showMethod(), getMethod() and showMethod(). For example
=== Debug an S4 function ===
<syntaxhighlight lang='rsplus'>
* '''showMethods('FUNCTION')'''
library(qrqc)
* '''getMethod('FUNCTION', 'SIGNATURE') ''' 
showMethods("gcPlot")
* '''debug(, signature)'''
getMethod("gcPlot", "FASTQSummary") # get an error
{{Pre}}
showMethods("gcPlot", "FASTQSummary") # good.
> args(debug)
</syntaxhighlight>
function (fun, text = "", condition = NULL, signature = NULL)  


* '''Debug a S4 function'''
<syntaxhighlight lang='rsplus'>
> library(genefilter) # Bioconductor
> library(genefilter) # Bioconductor
> showMethods("nsFilter")
> showMethods("nsFilter")
Line 5,720: Line 6,059:
eset="ExpressionSet"
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")
> debug(nsFilter, signature="ExpressionSet")
</syntaxhighlight>
 
library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code
</pre>
See the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L367 source code] of '''normalizationFactors<-''' (setReplaceMethod() is used) and the [https://github.com/mikelove/DESeq2/blob/445ae6c61d06de69d465b57f23e1c743d9b4537d/R/methods.R#L385 source code] of '''estimateSizeFactors()'''. We can see how ''avgTxLength'' was used in estimateNormFactors().
 
Another example
<pre>
library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)
 
showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"
 
debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz,
#      normalization = ssgsea.norm, verbose = verbose,
#      BPPARAM = BPPARAM))
 
isdebugged("gsva")
# [1] TRUE
undebug(gsva)
</pre>


* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
<syntaxhighlight lang='rsplus'>
{{Pre}}
library(IRanges)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir <- IRanges(start=c(10, 20, 30), width=5)
Line 5,754: Line 6,127:
##  
##  
## Known Subclasses: "NormalIRanges"
## Known Subclasses: "NormalIRanges"
</syntaxhighlight>
</pre>


==== See what methods work on an object ====
=== Check if a function is an S4 method ===
see what methods work on an object, e.g. a GRanges object:
'''isS4(foo)'''
<syntaxhighlight lang='rsplus'>methods(class="GRanges")</syntaxhighlight> Or if you have an object, x: <syntaxhighlight lang='rsplus'>methods(class=class(x))</syntaxhighlight>


==== View S3 function definition: double colon '::' and triple colon ':::' operators ====
=== How to access the slots of an S4 object ===
?":::"
* @ will let you access the slots of an S4 object.
* Note that often the best way to do this is to not access the slot directly but rather through an accessor function (e.g. coefs() rather than digging out the coefficients with $ or @). However, often such functions do not exist so you have to access the slots directly. This will mean that your code breaks if the internal implementation changes, however.
* [https://kasperdanielhansen.github.io/genbioconductor/html/R_S4.html#slots-and-accessor-functions R - S4 Classes and Methods] Hansen. '''getClass()''' or '''getClassDef()'''.


* pkg::name returns the value of the exported variable name in namespace pkg
=== setReplaceMethod() ===
* pkg:::name returns the value of the internal variable name
* [https://stackoverflow.com/a/24253311 What's the difference between setMethod(“$<-”) and set setReplaceMethod(“$”)?]
* [https://stackoverflow.com/a/49267668 What is setReplaceMethod() and how does it work?]


<syntaxhighlight lang='rsplus'>
=== See what methods work on an object ===
see what methods work on an object, e.g. a GRanges object:
<pre>
methods(class="GRanges")
</pre>
Or if you have an object, x:
<pre>
methods(class=class(x))
</pre>
 
=== View S3 function definition: double colon '::' and triple colon ':::' operators and getAnywhere() ===
?":::"
 
* pkg::name returns the value of the exported variable name in namespace pkg
* pkg:::name returns the value of the internal variable name
 
<pre>
base::"+"
base::"+"
stats:::coef.default
stats:::coef.default
</syntaxhighlight>


==== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ====
predict.ppr
# Error: object 'predict.ppr' not found
stats::predict.ppr
# Error: 'predict.ppr' is not an exported object from 'namespace:stats'
stats:::predict.ppr  # OR 
getS3method("predict", "ppr")
 
getS3method("t", "test")
</pre>
 
[https://stackoverflow.com/a/19226817 methods() + getAnywhere() functions]
 
=== Read the source code (include Fortran/C, S3 and S4 methods) ===
* [https://github.com/jimhester/lookup#readme lookup] package
* [https://blog.r-hub.io/2019/05/14/read-the-source/ Read the source]
* Find the source code in [https://stackoverflow.com/a/19226817 UseMethod("XXX")] for S3 methods.
 
=== S3 method is overwritten ===
For example, the select() method from dplyr is overwritten by [https://github.com/cran/grpreg/blob/master/NAMESPACE grpreg] package.
 
An easy solution is to load grpreg before loading dplyr.
 
* https://stackoverflow.com/a/14407095
* [https://njtierney.github.io/r/missing%20data/rbloggers/2016/11/06/simple-s3-methods/ A Simple Guide to S3 Methods] and [https://github.com/njtierney/A-Simple-Guide-to-S3-Methods/blob/master/SimpleS3.Rmd its source]
* [https://developer.r-project.org/Blog/public/2019/08/19/s3-method-lookup/index.html S3 Method Lookup]
 
=== mcols() and DataFrame() from Bioc [http://bioconductor.org/packages/release/bioc/html/S4Vectors.html S4Vectors] package ===
* mcols: Get or set the metadata columns.
* mcols: Get or set the metadata columns.
* colData: SummarizedExperiment instances from GenomicRanges
* colData: SummarizedExperiment instances from GenomicRanges
Line 5,777: Line 6,193:


For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
For example, in [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package  
<syntaxhighlight lang='rsplus'>
{{Pre}}
> mcols(ddsNoPrior[genes, ])
> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
DataFrame with 2 rows and 21 columns
Line 5,796: Line 6,212:
1      TRUE        3  210.4045 0.2648753
1      TRUE        3  210.4045 0.2648753
2      TRUE        9  243.7455 0.3248949
2      TRUE        9  243.7455 0.3248949
</syntaxhighlight>
</pre>
 
== Pipe ==
<ul>
<li>[https://www.tidyverse.org/blog/2023/04/base-vs-magrittr-pipe/ Differences between the base R and magrittr pipes] 4/21/2023
<li>[https://win-vector.com/2020/12/05/r-is-getting-an-official-pipe-operator/ R is Getting an Official Pipe Operator], [https://win-vector.com/2020/12/07/my-opinion-on-rs-upcoming-pipe/ My Opinion on R’s Upcoming Pipe]
<li> a(b(x)) vs '''x |> b() |> a()'''. See [https://twitter.com/henrikbengtsson/status/1335328090390597632 this tweet] in R-dev 2020-12-04.
<pre>
e0 <- quote(a(b(x)))
e1 <- quote(x |> b() |> a())
identical(e0, e1)
</pre>
</li>
<li>
[https://selbydavid.com/2021/05/18/pipes/ There are now 3 different R pipes]
</li>
<li>[https://stackoverflow.com/a/67629310 Error: The pipe operator requires a function call as RHS].
<pre>
# native pipe
foo |> bar()
# magrittr pipe
foo %>% bar
</pre>
</li>
<li>[https://www.infoworld.com/article/3621369/use-the-new-r-pipe-built-into-r-41.html Use the new R pipe built into R 4.1] </li>
<li>[https://towardsdatascience.com/the-new-native-pipe-operator-in-r-cbc5fa8a37bd The New Native Pipe Operator in R] </li>
<li>[https://ivelasq.rbind.io/blog/understanding-the-r-pipe/ Understanding the native R pipe |> ] </li>
<li>[https://medium.com/number-around-us/navigating-the-data-pipes-an-r-programming-journey-with-mario-bros-1aa621af1926 Navigating the Data Pipes: An R Programming Journey with Mario Bros]
</ul>
 
Packages take advantage of pipes
<ul>
<li>[https://cran.r-project.org/web/packages/rstatix/index.html rstatix]: Pipe-Friendly Framework for Basic Statistical Tests
</ul>


=== findInterval() ===
== findInterval() ==
Related functions are cuts() and split(). See also
Related functions are cuts() and split(). See also
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://books.google.com/books?id=oKY5QeSWb4cC&pg=PT310&lpg=PT310&dq=r+findinterval3&source=bl&ots=YjNMkHrTMw&sig=y_wIA1um420xVCI5IoGivABge-s&hl=en&sa=X&ei=gm_yUrSqLKXesAS2_IGoBQ&ved=0CFIQ6AEwBTgo#v=onepage&q=r%20findinterval3&f=false R Graphs Cookbook]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
* [http://adv-r.had.co.nz/Rcpp.html Hadley Wickham]


=== do.call, rbind, lapply ===
== Assign operator ==
Lots of examples. See for example [https://stat.ethz.ch/pipermail/r-help/attachments/20140423/62d8d103/attachment.pl this one] for creating a data frame from a vector.
* Earlier versions of R used underscore (_) as an assignment operator.
<syntaxhighlight lang='rsplus'>
* [https://developer.r-project.org/equalAssign.html Assignments with the = Operator]
x <- readLines(textConnection("---CLUSTER 1 ---
* In R 1.8.0 (2003), the assign operator has been removed. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
3
* In R 1.9.0 (2004), "_" is allowed in valid names. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
4
 
5
: [[File:R162.png|200px]]
6
---CLUSTER 2 ---
9
10
8
11"))


# create a list of where the 'clusters' are
== Operator precedence ==
clust <- c(grep("CLUSTER", x), length(x) + 1L)
The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.


  # get size of each cluster
== order(), rank() and sort() ==
  clustSize <- diff(clust) - 1L
If we want to find the indices of the first 25 genes with the smallest p-values, we can use '''order(pval)[1:25]'''.
<pre>
> x = sample(10)
> x
  [1]  4  3 10  7  5  8  6  1  9  2
> order(x)
[1]  8 10  2  1  5  7  4  6  9  3
> rank(x)
  [1]  4  3 10  7  5  8  6  1  9  2
> rank(10*x)
[1]  4  3 10  7  5  8  6  1  9  2


# get cluster number
> x[order(x)]
  clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])
  [1]  1  2  3  4  5  6  7  8  9 10
> sort(x)
[1] 1  2  3  4  5  6  7  8  9 10
</pre>


result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
=== relate order() and rank() ===
    cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
<ul>
        , Cluster = .cl
<li>Order to rank: rank() = order(order())
        )
<syntaxhighlight lang='r'>
    }))
set.seed(1)
x <- rnorm(5)
order(x)
# [1] 3 1 2 5 4
rank(x)
# [1] 2 3 1 5 4
order(order(x))
# [1] 2 3 1 5 4
all(rank(x) == order(order(x)))
# TRUE
</syntaxhighlight>


result
<li>Order to Rank method 2: rank(order()) = 1:n
<syntaxhighlight lang='r'>
ord <- order(x)
ranks <- integer(length(x))
ranks[ord] <- seq_along(x)
ranks
# [1] 2 3 1 5 4
</syntaxhighlight>


    Object Cluster
<li>Rank to Order:
[1,] "3"    "1"
<syntaxhighlight lang='r'>
[2,] "4"    "1"
ranks <- rank(x)
[3,] "5"    "1"
ord <- order(ranks)
[4,] "6"    "1"
ord
[5,] "9"    "2"
# [1] 3 1 2 5 4
[6,] "10"  "2"
[7,] "8"    "2"
[8,] "11"  "2"
</syntaxhighlight>
</syntaxhighlight>
</ul>


A 2nd example is to [http://datascienceplus.com/working-with-data-frame-in-r/ sort a data frame] by using do.call(order, list()).
=== OS-dependent results on sorting string vector ===
Gene symbol case.
<pre>
# mac:
order(c("DC-UbP", "DC2")) # c(1,2)


=== How to get examples from help file ===
# linux:
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
order(c("DC-UbP", "DC2")) # c(2,1)
Method 1:
<pre>
example(acf, give.lines=TRUE)
</pre>
</pre>
Method 2:
 
Affymetric id case.
<pre>
<pre>
Rd <- utils:::.getHelpFile(?acf)
# mac:
tools::Rd2ex(Rd)
order(c("202800_at", "2028_s_at")) # [1] 2 1
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"
 
# linux
order(c("202800_at", "2028_s_at")) # [1] 1 2
sort(c("202800_at", "2028_s_at")) # [1] "202800_at" "2028_s_at"
</pre>
</pre>
It does not matter if we include factor() on the character vector.


=== "[" and "[[" with the sapply() function ===
The difference is related to locale. See
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
 
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/locales ?locales] in R
* On OS, type '''locale'''
* [https://stackoverflow.com/questions/39171613/sort-produces-different-results-in-ubuntu-and-windows sort() produces different results in Ubuntu and Windows]
* To fix the inconsistency problem, we can set the locale in R code to "C" or use the stringr package (the locale is part of [https://www.rdocumentation.org/packages/stringr/versions/1.4.0/topics/str_order str_order()]'s arguments).
<pre>
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
# both mac and linux
stringr::str_order(c("202800_at", "2028_s_at")) # [1] 2 1
stringr::str_order(c("DC-UbP", "DC2")) # [1] 1 2
 
# Or setting the locale to "C"
Sys.setlocale("LC_ALL", "C"); sort(c("DC-UbP", "DC2"))
# Or
Sys.setlocale("LC_COLLATE", "C"); sort(c("DC-UbP", "DC2"))
# But not
Sys.setlocale("LC_ALL", "en_US.UTF-8"); sort(c("DC-UbP", "DC2"))
</pre>
</pre>
is the same as
 
=== unique() ===
It seems it does not sort. [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/unique ?unique].
<pre>
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
# mac & linux
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"
</pre>
</pre>


=== Dealing with date ===
== do.call ==
'''do.call''' constructs and executes a function call from a name or a function and a list of arguments to be passed to it.
 
[https://www.r-bloggers.com/2023/05/the-do-call-function-in-r-unlocking-efficiency-and-flexibility/ The do.call() function in R: Unlocking Efficiency and Flexibility]
 
Below are some examples from the [https://stat.ethz.ch/R-manual/R-devel/library/base/html/do.call.html help].
 
* Usage
{{Pre}}
do.call(what, args, quote = FALSE, envir = parent.frame())
# what: either a function or a non-empty character string naming the function to be called.
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
# quote: a logical value indicating whether to quote the arguments.
# envir: an environment within which to evaluate the call. This will be most useful
#        if what is a character string and the arguments are symbols or quoted expressions.
</pre>
* do.call() is similar to [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/lapply lapply()] but not the same. It seems do.call() can make a simple function vectorized.
{{Pre}}
> do.call("complex", list(imag = 1:3))
[1] 0+1i 0+2i 0+3i
> lapply(list(imag = 1:3), complex)
$imag
[1] 0+0i
> complex(imag=1:3)
[1] 0+1i 0+2i 0+3i
> do.call(function(x) x+1, list(1:3))
[1] 2 3 4
</pre>
* Applying do.call with Multiple Arguments
<pre>
<pre>
d1 = date()
> do.call("sum", list(c(1,2,3,NA), na.rm = TRUE))
class(d1) # "character"
[1] 6
d2 = Sys.Date()
> do.call("sum", list(c(1,2,3,NA) ))
class(d2) # "Date"
[1] NA
</pre>
* [https://www.stat.berkeley.edu/~s133/Docall.html do.call() allows you to call any R function, but instead of writing out the arguments one by one, you can use a list to hold the arguments of the function.]
{{Pre}}
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
> length(tmp)
[1] 3
> tmp[1:4,]
  Var1 Var2 Var3
1    a    1    +
2    b    1    +
3    a    2    +
4    b    2    +
> c(tmp, sep = "")
$Var1
[1] a b a b a b a b a b a b
Levels: a b
 
$Var2
[1] 1 1 2 2 3 3 1 1 2 2 3 3


format(d2, "%a %b %d")
$Var3
[1] + + + + + + - - - - - -
Levels: + -


library(lubridate); ymd("20140108") # "2014-01-08 UTC"
$sep
mdy("08/04/2013") # "2013-08-04 UTC"
[1] ""
dmy("03-04-2013") # "2013-04-03 UTC"
> do.call("paste", c(tmp, sep = ""))
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
[1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland")
[12] "b3-"
# "2011-08-03 10:15:03 NZST"
?Sys.timezone
x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues
</pre>
</pre>
* http://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/
* ''environment'' and ''quote'' arguments.
* http://cran.r-project.org/web/packages/lubridate/vignettes/lubridate.html
{{Pre}}
* http://rpubs.com/seandavi/GEOMetadbSurvey2014
> A <- 2
* We want our dates and times as class "Date" or the class "POSIXct", "POSIXlt". For more information type ?POSIXlt.
> f <- function(x) print(x^2)
> env <- new.env()
> assign("A", 10, envir = env)
> assign("f", f, envir = env)
> f <- function(x) print(x)
> f(A) 
[1] 2
> do.call("f", list(A))
[1] 2
> do.call("f", list(A), envir = env) 
[1] 4
> do.call(f, list(A), envir = env) 
[1] 2                      # Why?


=== [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] and deparse/substitute ===
> eval(call("f", A))                     
* [https://cran.r-project.org/web/packages/lazyeval/vignettes/lazyeval.html Vignette] from the [https://cran.r-project.org/web/packages/lazyeval/index.html lazyeval] package. It is needed in three cases
[1] 2
** Labelling: turn an argument into a label
> eval(call("f", quote(A)))             
** Formulas
[1] 2
** Dot-dot-dot
> eval(call("f", A), envir = env)       
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/substitute substitute(expr, env)] - capture expression.
[1] 4
** substitute() is often paired with deparse() to create informative labels for data sets and plots.
> eval(call("f", quote(A)), envir = env) 
** Use 'substitute' to include the variable's name in a plot title, e.g.: '''var <- "abc"; hist(var,main=substitute(paste("Dist of ", var))) ''' will show the title "Dist of var" instead of "Dist of abc" in the title.
[1] 100
* quote(expr) - similar to substitute() but do nothing?? [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/noquote noquote] - print character strings without quotes
</pre>
* eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
* deparse(expr) - turns unevaluated expressions into character strings. For example,
{{Pre}}
> foo <- function(a=1, b=2, ...) {
        list(arg=do.call(c, as.list(match.call())[-1]))
  }
> foo()
$arg
NULL
> foo(a=1)
$arg
a  
1
> foo(a=1, b=2, c=3)
$arg
a b c
1 2 3
</pre>
* do.call() + switch(). See [https://github.com/satijalab/seurat/blob/13b615c27eeeac85e5c928aa752197ac224339b9/R/preprocessing.R#L2450 an example] from Seurat::NormalizeData.
<pre>
<pre>
> deparse(args(lm))
do.call(
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "
  what = switch(
[2] "   model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
    EXPR = margin,
[3] "   contrasts = NULL, offset, ...) "                                   
    '1' = 'rbind',
[4] "NULL"   
    '2' = 'cbind',
    stop("'margin' must be 1 or 2")
  ),
   args = normalized.data
)
switch('a', 'a' = rnorm(3), 'b'=rnorm(4)) # switch returns a value
do.call(switch('a', 'a' = 'rnorm', 'b'='rexp'), args=list(n=4)) # switch returns a function
</pre>
* The function we want to call is a string that may change: [https://github.com/cran/glmnet/blob/master/R/cv.glmnet.raw.R#L66 glmnet]
<pre>
# Suppose we want to call cv.glmnet or cv.coxnet or cv.lognet or cv.elnet .... depending on the case
fun = paste("cv", subclass, sep = ".")
cvstuff = do.call(fun, list(predmat,y,type.measure,weights,foldid,grouped))
</pre>
 
=== expand.grid, mapply, vapply ===
[https://shikokuchuo.net/posts/10-combinations/ A faster way to generate combinations for mapply and vapply]


> deparse(args(lm), width=20)
=== do.call vs mapply ===
[1] "function (formula, data, "       "   subset, weights, "          
* do.call() is doing what [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/mapply mapply()] does but do.call() uses a list instead of multiple arguments. So do.call() more close to [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/funprog base::Map()] function.
[3] "   na.action, method = \"qr\", " "   model = TRUE, x = FALSE, "  
{{Pre}}
[5] "   y = FALSE, qr = TRUE, "       "   singular.ok = TRUE, "      
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
[7] "   contrasts = NULL, "           "   offset, ...) "             
      Var1
[9] "NULL"
[1,] "a1+"
[2,] "b1+"
[3,] "a2+"
[4,] "b2+"
[5,] "a3+"
[6,] "b3+"
[7,] "a1-"
[8,] "b1-"
[9,] "a2-"
[10,] "b2-"
[11,] "a3-"
[12,] "b3-"
# It does not work if we do not explicitly specify the arguments in mapply()
> mapply(paste, tmp, sep = "")
      Var1 Var2 Var3
[1,] "a"  "1"  "+"
[2,] "b"  "1"  "+"  
[3,] "a"  "2"  "+"
[4,] "b"  "2"  "+"  
[5,] "a"  "3"  "+"
[6,] "b"  "3"  "+"
[7,] "a"  "1"  "-"
[8,] "b"  "1"  "-"  
[9,] "a"  "2"  "-"  
[10,] "b"  "2"  "-"
[11,] "a"  "3" "-"
[12,] "b"  "3"  "-"
</pre>
* mapply is useful in generating variables with a vector of parameters. For example suppose we want to generate variables from exponential/weibull distribution and a vector of scale parameters (depending on some covariates). In this case we can use ([https://stackoverflow.com/a/17031993 Simulating Weibull distributions from vectors of parameters in R])
{{Pre}}
set.seed(1)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
# [1] 1.326108 9.885284
set.seed(1)
x <- replicate(1000, mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1)))
dim(x) # [1]  2 1000
rowMeans(x)
# [1]  1.032209 10.104131
</pre>
{{Pre}}
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
# [1] 1.326108 9.885284
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))
</pre>
</pre>
* parse(text) - returns the parsed but unevaluated expressions in a list. See [[R#Create_a_Simple_Socket_Server_in_R|Create a Simple Socket Server in R]] for the application of '''eval(parse(text))'''. Be cautious!
** [http://r.789695.n4.nabble.com/using-eval-parse-paste-in-a-loop-td849207.html eval(parse...)) should generally be avoided]
** [https://stackoverflow.com/questions/13649979/what-specifically-are-the-dangers-of-evalparse What specifically are the dangers of eval(parse(…))?]


Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).  
=== do.call vs lapply ===
<syntaxhighlight lang='rsplus'>
[https://stackoverflow.com/a/10801883 What's the difference between lapply and do.call?] It seems to me the best usage is combining both functions: '''do.call(..., lapply())'''
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3
 
* lapply returns a list of the same length as X, each element of which is the result of applying FUN to the corresponding element of X.
* do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it. '''It is widely used, for example, to assemble lists into simpler structures (often with rbind or cbind).'''
* Map applies a function to the corresponding elements of given vectors... Map is a simple wrapper to mapply which does not attempt to simplify the result, similar to Common Lisp's mapcar (with arguments being recycled, however). Future versions may allow some control of the result type.


f1(1:3)
{{Pre}}
f2(1:3)
> lapply(iris, class) # same as Map(class, iris)
f3(1:3)
$Sepal.Length
[1] "numeric"


# Or
$Sepal.Width
myfun <- function(f, a) {
[1] "numeric"
    eval(parse(text = f))(a)
}
myfun("f1", 1:3)
myfun("f2", 1:3)
myfun("f3", 1:3)


# Or with lapply
$Petal.Length
method <- c("f1", "f2", "f3")
[1] "numeric"
res <- lapply(method, function(M) {
                    Mres <- eval(parse(text = M))(1:3)
                    return(Mres)
})
names(res) <- method
</syntaxhighlight>


=== The ‘…’ argument ===
$Petal.Width
See [http://cran.r-project.org/doc/manuals/R-intro.html#The-three-dots-argument Section 10.4 of An Introduction to R]. Especially, the expression '''list(...)''' evaluates all such arguments and returns them in a named list
[1] "numeric"


=== Lazy evaluation in R functions arguments ===
$Species
* http://adv-r.had.co.nz/Functions.html
[1] "factor"
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html


'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  
> x <- lapply(iris, class)
> do.call(c, x)
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species
  "numeric"    "numeric"    "numeric"    "numeric"    "factor"
</pre>


* Example 1. By default, R function arguments are lazy.
https://stackoverflow.com/a/10801902
<pre>
* '''lapply''' applies a function '''over a list'''. So there will be several function calls.
f <- function(x) {
* '''do.call''' calls a function with '''a list of arguments''' (... argument) such as [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/c c()] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/cbind rbind()/cbind()] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/sum sum] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/order order] or [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/Extract "["] or paste. So there is only one function call.
  999
{{Pre}}
}
> X <- list(1:3,4:6,7:9)
f(stop("This is an error!"))
> lapply(X,mean)
#> [1] 999
[[1]]
</pre>
[1] 2


* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
[[2]]
<pre>
[1] 5
add <- function(x) {
  force(x)
  function(y) x + y
}
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
</pre>


* Example 3. Default arguments are evaluated inside the function.
[[3]]
<pre>
[1] 8
f <- function(x = ls()) {
> do.call(sum, X)
  a <- 1
[1] 45
  x
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
}
[1] 45
> do.call(mean, X) # Error
> do.call(rbind,X)
    [,1] [,2] [,3]
[1,]    1    2    3
[2,]    4    5    6
[3,]    7    8    9
> lapply(X,rbind)
[[1]]
    [,1] [,2] [,3]
[1,]    1    2    3


# ls() evaluated inside f:
[[2]]
f()
    [,1] [,2] [,3]
# [1] "a" "x"
[1,]   4    5    6


# ls() evaluated in global environment:
[[3]]
f(ls())
    [,1] [,2] [,3]
# [1] "add"    "adders" "f"
[1,]    7    8    9
> mapply(mean, X, trim=c(0,0.5,0.1))
[1] 2 5 8
> mapply(mean, X)  
[1] 2 5 8
</pre>
</pre>
Below is a good example to show the difference of lapply() and do.call() - [https://stackoverflow.com/a/42734863 Generating Random Strings].
{{Pre}}
> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x
[[1]]
[1] "Y" "D" "G" "A"
[[2]]
[1] "B" "W" "K" "N"


* Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
> lapply(x, paste0)
<pre>
[[1]]
x <- NULL
[1] "Y" "D" "G" "A"
if (!is.null(x) && x > 0) {


}
[[2]]
</pre>
[1] "B" "W" "K" "N"


=== Backtick sign, infix/prefix/postfix operators ===  
> lapply(x, paste0, collapse= "")
The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in [http://adv-r.had.co.nz/Functions.html this note].
[[1]]
[1] "YDGA"


'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
[[2]]
<pre>
[1] "BWKN"
1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix
</pre>


=== List data type ===
> do.call(paste0, x)
==== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ====
[1] "YB" "DW" "GK" "AN"
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5
</pre>
</pre>


=== Error handling and exceptions, tryCatch(), stop(), warning() and message() ===
=== do.call + rbind + lapply ===
* http://adv-r.had.co.nz/Exceptions-Debugging.html
Lots of examples. See for example [https://stat.ethz.ch/pipermail/r-help/attachments/20140423/62d8d103/attachment.pl this one] for creating a data frame from a vector.
* try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
{{Pre}}
<pre>
x <- readLines(textConnection("---CLUSTER 1 ---
out <- try({
3
  a <- 1
4
  b <- "x"
5
  a + b
6
})
---CLUSTER 2 ---
9
10
8
11"))


elements <- list(1:10, c(-1, 10), c(T, F), letters)
# create a list of where the 'clusters' are
results <- lapply(elements, log)
clust <- c(grep("CLUSTER", x), length(x) + 1L)
is.error <- function(x) inherits(x, "try-error")
succeeded <- !sapply(results, is.error)
</pre>
* tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
<pre>
tryCatch(expr, ..., finally)


show_condition <- function(code) {
# get size of each cluster
  tryCatch(code,
clustSize <- diff(clust) - 1L
    error = function(c) "error",
    warning = function(c) "warning",
    message = function(c) "message"
  )
}
show_condition(stop("!"))
#> [1] "error"
show_condition(warning("?!"))
#> [1] "warning"
show_condition(message("?"))
#> [1] "message"
show_condition(10)
#> [1] 10
</pre>
Below is another snippet from available.packages() function,
<pre>
z <- tryCatch(download.file(....), error = identity)
if (!inherits(z, "error")) STATEMENTS
</pre>


=== Using list type ===
# get cluster number
==== Avoid if-else or switch ====
  clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])
?plot.stepfun.
<pre>
y0 <- c(1,2,4,3)
sfun0 <- stepfun(1:3, y0, f = 0)
sfun.2 <- stepfun(1:3, y0, f = .2)
sfun1  <- stepfun(1:3, y0, right = TRUE)


tt <- seq(0, 3, by = 0.1)
result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
op <- par(mfrow = c(2,2))
    cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
        , Cluster = .cl
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
        )
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
     }))
##-- This is  revealing :
plot(sfun0, verticals = FALSE,
     main = "stepfun(x, y0, f=f) for f = 0, .2, 1")


for(i in 1:3)
result
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)


par(op)
    Object Cluster
[1,] "3"    "1"
[2,] "4"    "1"
[3,] "5"    "1"
[4,] "6"    "1"
[5,] "9"    "2"
[6,] "10"  "2"
[7,] "8"    "2"
[8,] "11"  "2"
</pre>
</pre>
[[File:StepfunExample.svg|400px]]


=== Open a new Window device ===
A 2nd example is to [http://datascienceplus.com/working-with-data-frame-in-r/ sort a data frame] by using do.call(order, list()).
X11() or dev.new()


=== par() ===
Another example is to reproduce aggregate(). aggregate() = do.call() + by().
?par
{{Pre}}
attach(mtcars)
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
# the above approach give the same result as the following
# except it does not have an extra Group.x columns
aggregate(mtcars, list(cyl, vs), FUN=mean)
</pre>


==== text size and font on main, lab & axis ====
== Run examples ==
* [https://www.statmethods.net/advgraphs/parameters.html Graphical Parameters] from statmethods.net.
When we call help(FUN), it shows the document in the browser. The browser will show
* [https://designdatadecisions.wordpress.com/2015/06/09/graphs-in-r-overlaying-data-summaries-in-dotplots/ Overlaying Data Summaries in Dotplots]
<pre>
example(FUN, package = "XXX") was run in the console
To view output in the browser, the knitr package must be installed
</pre>


Examples:
== How to get examples from help file, example() ==
* cex.main=0.9
[https://blog.r-hub.io/2020/01/27/examples/ Code examples in the R package manuals]:
* cex.lab=0.8
<pre>
* font.lab=2
# How to run all examples from a man page
* cex.axis=0.8
example(within)
* font.axis=2
* col.axis="grey50"


==== layout ====
# How to check your examples?
http://datascienceplus.com/adding-text-to-r-plot/
devtools::run_examples()
testthat::test_examples()
</pre>


==== reset the settings ====
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
<syntaxhighlight lang='rsplus'>
Method 1:
op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1)
<pre>
....
example(acf, give.lines=TRUE)
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1
</pre>
</syntaxhighlight>
Method 2:
<pre>
Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)
</pre>


==== mtext (margin text) vs title ====
== "[" and "[[" with the sapply() function ==
* https://datascienceplus.com/adding-text-to-r-plot/
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
* https://datascienceplus.com/mastering-r-plot-part-2-axis/
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
</pre>
is the same as
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
</pre>


==== mgp (axis label locations) ====
== Dealing with dates ==
# The margin line (in ‘mex’ units) for the axis title, axis labels and axis line.  Note that ‘mgp[1]’ affects ‘title’ whereas ‘mgp[2:3]’ affect ‘axis’.  The default is ‘c(3, 1, 0). If we like to make the axis labels closer to an axis, we can use mgp=c(2.3, 1, 0) for example.
<ul>
# http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the labels (i.e. xlab and ylab in plot), the second the tick-mark labels, and third the tick marks. The default is c(3, 1, 0).
<li>Simple examples
<syntaxhighlight lang='rsplus'>
dates <- c("January 15, 2023", "December 31, 1999")
date_objects <- as.Date(dates, format = "%B %d, %Y") # format is for the input
# [1] "2023-01-15" "1999-12-31"
</syntaxhighlight>


==== pch ====
<li>Find difference
[[File:R pch.png|250px]]
<syntaxhighlight lang='rsplus'>
# Convert the dates to Date objects
date1 <- as.Date("6/29/21", format="%m/%d/%y")
date2 <- as.Date("11/9/21", format="%m/%d/%y")


([https://www.statmethods.net/advgraphs/parameters.html figure source])
# Calculate the difference in days
diff_days <- as.numeric(difftime(date2, date1, units="days")) # 133
# In months
diff_days / (365.25/12)  # 4.36961 


* Full circle: pch=16
# OR using the lubridate package
library(lubridate)
# Convert the dates to Date objects
date1 <- mdy("6/29/21")
date2 <- mdy("11/9/21")
interval(date1, date2) %/% months(1)
</syntaxhighlight>


==== lty (line type) ====
<li>http://cran.r-project.org/web/packages/lubridate/vignettes/lubridate.html
[[File:R lty.png|250px]]
<syntaxhighlight lang='rsplus'>
d1 = date()
class(d1) # "character"
d2 = Sys.Date()
class(d2) # "Date"


([http://www.sthda.com/english/wiki/line-types-in-r-lty figure source])
format(d2, "%a %b %d")


==== las (label style) ====
library(lubridate); ymd("20140108") # "2014-01-08 UTC"
0: The default, parallel to the axis
mdy("08/04/2013") # "2013-08-04 UTC"
dmy("03-04-2013") # "2013-04-03 UTC"
ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland")
# "2011-08-03 10:15:03 NZST"
?Sys.timezone
x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
wday(x[1]) # 3
wday(x[1], label=TRUE) # Tues
</syntaxhighlight>


1: Always horizontal
<li>http://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/
<li>http://rpubs.com/seandavi/GEOMetadbSurvey2014
<li>We want our dates and times as class "Date" or the class "POSIXct", "POSIXlt". For more information type ?POSIXlt.
<li>[https://cran.r-project.org/web/packages/anytime/index.html anytime] package
<li>weeks to Christmas difftime(as.Date(“2019-12-25”), Sys.Date(), units =“weeks”)
<li>[https://blog.rsquaredacademy.com/handling-date-and-time-in-r/ A Comprehensive Introduction to Handling Date & Time in R] 2020
<li>[https://www.spsanderson.com/steveondata/posts/rtip-2023-05-12/index.html Working with Dates and Times Pt 1]
* Three major functions: as.Date(), as.POSIXct(), and as.POSIXlt().
* '''POSIXct''' is a class in R that represents date-time data. The ct stands for “calendar time” and it represents the (signed) number of seconds since the beginning of 1970 as a numeric vector1.  '''It stores date time as integer.'''
* '''POSIXlt''' is a class in R that represents date-time data. It stands for “local time” and is a list with components as integer vectors, which can represent a vector of broken-down times. '''It stores date time as list:sec, min, hour, mday, mon, year, wday, yday, isdst, zone, gmtoff'''.


2: Perpendicular to the axis
<li>[https://www.r-bloggers.com/2023/11/r-lubridate-how-to-efficiently-work-with-dates-and-times-in-r/ R lubridate: How To Efficiently Work With Dates and Times in R] 2023
</ul>


3: Always vertical
== Nonstandard/non-standard evaluation, deparse/substitute and scoping ==
 
* [https://www.brodieg.com/2020/05/05/on-nse/ Standard and Non-Standard Evaluation in R]
==== oma (outer margin), common title for two plots ====
* [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] from Advanced R book.
The following trick is useful when we want to draw multiple plots with a common title.
* [https://edwinth.github.io/blog/nse/ Non-standard evaluation, how tidy eval builds on base R]
 
* [https://cran.r-project.org/web/packages/lazyeval/vignettes/lazyeval.html Vignette] from the [https://cran.r-project.org/web/packages/lazyeval/index.html lazyeval] package. It is needed in three cases
<syntaxhighlight lang='rsplus'>
** Labelling: turn an argument into a label
par(mfrow=c(1,2),oma = c(0, 0, 2, 0)) # oma=c(0, 0, 0, 0) by default
** Formulas
plot(1:10,  main="Plot 1")
** Dot-dot-dot
plot(1:100,  main="Plot 2")
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/substitute substitute(expr, env)] - capture expression. The return mode is a '''call'''.
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
** substitute() is often paired with '''deparse'''() to create informative labels for data sets and plots. The return mode of deparse() is '''character strings'''.
** Use 'substitute' to include the variable's name in a plot title, e.g.: '''var <- "abc"; hist(var,main=substitute(paste("Dist of ", var))) ''' will show the title "Dist of var" instead of "Dist of abc" in the title.
** [https://stackoverflow.com/a/34079727 Passing a variable name to a function in R]
** Example:
::<syntaxhighlight lang='rsplus'>
f <- function(x) {
  substitute(x)
}
f(1:10)
# 1:10
class(f(1:10)) # or mode()
# [1] "call"
g <- function(x) deparse(substitute(x))
g(1:10)
# [1] "1:10"
class(g(1:10)) # or mode()
# [1] "character"
</syntaxhighlight>
* quote(expr) - similar to substitute() but do nothing?? [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/noquote noquote] - print character strings without quotes
:<syntaxhighlight lang='rsplus'>
mode(quote(1:10))
# [1] "call"
</syntaxhighlight>
</syntaxhighlight>
* eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.
** The '''parent.frame()''' is necessary in cases like the [https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/update stats::update()] function used by [https://github.com/cran/glmnet/blob/master/R/relax.glmnet.R#L66 relax.glmnet()].
** Example:
::<syntaxhighlight lang='rsplus'>
sample_df <- data.frame(a = 1:5, b = 5:1, c = c(5, 3, 1, 4, 1))


[https://datascienceplus.com/mastering-r-plot-part-3-outer-margins/ Mastering R plot – Part 3: Outer margins] '''mtext()''' & '''par(xpd)'''.
subset1 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x)
  x[r, ]
}
x <- 4
condition <- 4
subset1(sample_df, a== 4) # same as subset(sample_df, a >= 4)
subset1(sample_df, a== x) # WRONG!
subset1(sample_df, a == condition) # ERROR


=== Non-standard fonts in postscript and pdf graphics ===
subset2 <- function(x, condition) {
https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41
  condition_call <- substitute(condition)
  r <- eval(condition_call, x, parent.frame())
  x[r, ]
}
subset2(sample_df, a == 4) # same as subset(sample_df, a >= 4)
subset2(sample_df, a == x) # 👌
subset2(sample_df, a == condition) # 👍
</syntaxhighlight>
* deparse(expr) - turns unevaluated expressions into character strings. For example,
:<syntaxhighlight lang='rsplus'>
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", "
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                   
[4] "NULL"   


=== Suppress warnings ===
> deparse(args(lm), width=20)
Use [https://www.rdocumentation.org/packages/base/versions/3.4.1/topics/options options()]. If ''warn'' is negative all warnings are ignored. If ''warn'' is zero (the default) warnings are stored until the top--level function returns.
[1] "function (formula, data, "        "    subset, weights, "         
<syntaxhighlight lang='rsplus'>
[3] "    na.action, method = \"qr\", " "   model = TRUE, x = FALSE, "  
op <- options("warn")
[5] "    y = FALSE, qr = TRUE, "      "    singular.ok = TRUE, "       
options(warn = -1)
[7] "    contrasts = NULL, "          "    offset, ...) "             
....
[9] "NULL"
options(op)
 
# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)
</syntaxhighlight>
</syntaxhighlight>
* parse(text) - returns the parsed but unevaluated expressions in a list. See [[R#Create_a_Simple_Socket_Server_in_R|Create a Simple Socket Server in R]] for the application of '''eval(parse(text))'''. Be cautious!
** [http://r.789695.n4.nabble.com/using-eval-parse-paste-in-a-loop-td849207.html eval(parse...)) should generally be avoided]
** [https://stackoverflow.com/questions/13649979/what-specifically-are-the-dangers-of-evalparse What specifically are the dangers of eval(parse(…))?]


=== NULL, NA, NaN, Inf ===
Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).  
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/
{{Pre}}
f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3


=== save() vs saveRDS() ===
f1(1:3)
# saveRDS() can only save one R object while save() does not have this constraint.
f2(1:3)
# saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See [http://www.fromthebottomoftheheap.net/2012/04/01/saving-and-loading-r-objects/ this post].
f3(1:3)
<pre>
x <- 5
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)
</pre>


=== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ===
# Or
* ==: exact match
myfun <- function(f, a) {
* all.equal: compare R objects x and y testing ‘near equality’
    eval(parse(text = f))(a)
* identical: The safe and reliable way to test two objects for being exactly equal.
}
<syntaxhighlight lang='rsplus'>
myfun("f1", 1:3)
x <- 1.0; y <- 0.99999999999
myfun("f2", 1:3)
all.equal(x, y)
myfun("f3", 1:3)
# [1] TRUE
identical(x, y)
# [1] FALSE
</syntaxhighlight>


See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.
# Or with lapply
 
method <- c("f1", "f2", "f3")
=== Numerical Pitfall ===
res <- lapply(method, function(M) {
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
                    Mres <- eval(parse(text = M))(1:3)
<syntaxhighlight lang='bash'>
                    return(Mres)
.1 - .3/3
})
## [1] 0.00000000000000001388
names(res) <- method
</syntaxhighlight>
</pre>


=== Sys.getpid() ===
=== library() accept both quoted and unquoted strings ===
This can be used to monitor R process memory usage or stop the R process. See [https://stat.ethz.ch/pipermail/r-devel/2016-November/073360.html this post].
[https://stackoverflow.com/a/25210607 How can library() accept both quoted and unquoted strings]. The key lines are
 
=== How to debug an R code ===
==== Using assign() in functions ====
For example, insert the following line to your function
<pre>
<pre>
assign(envir=globalenv(), "GlobalVar", localvar)
  if (!character.only)
    package <- as.character(substitute(package))
</pre>
</pre>


=== Debug lapply()/sapply() ===
=== Lexical scoping ===
* https://stackoverflow.com/questions/1395622/debugging-lapply-sapply-calls
* [https://lgreski.github.io/dsdepot/2020/06/28/rObjectsSObjectsAndScoping.html R Objects, S Objects, and Lexical Scoping]
* https://stat.ethz.ch/R-manual/R-devel/library/utils/html/recover.html. Use options(error=NULL) to turn it off.
* [http://www.biostat.jhsph.edu/~rpeng/docs/R-classes-scope.pdf#page=31 Dynamic scoping vs Lexical scoping] and the example of [http://www.biostat.jhsph.edu/~rpeng/docs/R-classes-scope.pdf#page=41 optimization]
* [https://www.r-bloggers.com/2024/03/indicating-local-functions-in-r-scripts/ Indicating local functions in R scripts]


=== Debugging with RStudio ===
== The ‘…’ argument ==
* https://www.rstudio.com/resources/videos/debugging-techniques-in-rstudio/
* See [http://cran.r-project.org/doc/manuals/R-intro.html#The-three-dots-argument Section 10.4 of An Introduction to R]. Especially, the expression '''list(...)''' evaluates all such arguments and returns them in a named list
* https://github.com/ajmcoqui/debuggingRStudio/blob/master/RStudio_Debugging_Cheatsheet.pdf
* [https://statisticaloddsandends.wordpress.com/2020/11/15/some-notes-when-using-dot-dot-dot-in-r/ Some notes when using dot-dot-dot (…) in R]
* https://support.rstudio.com/hc/en-us/articles/205612627-Debugging-with-RStudio
* [https://stackoverflow.com/questions/26684509/how-to-check-if-any-arguments-were-passed-via-ellipsis-in-r-is-missing How to check if any arguments were passed via “…” (ellipsis) in R? Is missing(…) valid?]


=== Debug R source code ===
== Functions ==
==== Build R with debug information ====
* https://adv-r.hadley.nz/functions.html
* [[R#Build_R_from_its_source|R -> Build R from its source on Windows]]
* [https://towardsdatascience.com/writing-better-r-functions-best-practices-and-tips-d48ef0691c24 Writing Better R Functions — Best Practices and Tips]. The [https://cran.r-project.org/web/packages/docstring/index.html docstring] package and "?" is interesting!
* http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/ (defunct)
* http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/gdb.shtml (defunct)
* Build R with debug information (see the discussion [https://stackoverflow.com/a/30001096 here]). Cf [https://github.com/arraytools/r-build-output output messages] from running ./configure and make using the default options.
: <syntaxhighlight lang='bash'>
$ ./configure --help
$ ./configure --enable-R-shlib --with-valgrind-instrumentation=2 \
                              --with-system-valgrind-headers \
              CFLAGS='-g -O0 -fPIC' \
              FFLAGS='-g -O0 -fPIC' \
              CXXFLAGS='-g -O0 -fPIC' \
              FCFLAGS='-g -O0 -fPIC'
$ make -j4
$ sudo make install
</syntaxhighlight>
* [https://github.com/arraytools/r-debug My note of debugging cor() function]
* [https://vimeo.com/11937905 Using gdb to debug R packages with native code] (Video) The steps to debug is given below.
: <syntaxhighlight lang='bash'>
# Make sure to create a file <src/Makevars> with something like: CFLAGS=-ggdb -O0
# Or more generally
# CFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
# CXXFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
# FFLAGS=-Wall -Wextra -pedantic -O0 -ggdb
 
$ tree nidemo
$ R CMD INSTALL nidemo
$ cat bug.R
$ R -f bug.R
$ R -d gdb
(gdb) r
> library(nidemo)
> Ctrl+C
(gdb) b nid_buggy_freq
(gdb) c  # continue
> buggy_freq("nidemo/DESCRIPTION") # stop at breakpoint 1
(gdb) list
(gdb) n # step through
(gdb) # press RETURN a few times until you see the bug
(gdb) d 1 # delete the first break point
(gdb) b Rf_error # R's C entry point for the error function
(gdb) c
> buggy_freq("nidemo/DESCRIPTION")
(gdb) bt 5 # last 5 stack frames
(gdb) frame 2
(gdb) list
(gdb) p freq_data
(gdb) p ans
(gdb) call Rf_PrintValues(ans)
(gdb) call Rf_PrintValues(fname)
(gdb) q
# Edit buggy.c
 
$ R CMD INSTALL nidemo # re-install the package
$ R -f bug.R
$ R -d gdb
(gdb) run
> source("bug.R") # error happened
(gdb) bt 5 # show the last 5 frames
(gdb) frame 2
(gdb) list
(gdb) frame 1
(gdb) list
(gdb) p file
(gdb) p fh
(gdb) q
# Edit buggy.c


$ R CMD INSTALL nidemo
=== Function argument ===
$ R -f bug.R
[https://cran.r-project.org/doc/manuals/r-release/R-lang.html#Argument-matching Argument matching] from [https://cran.r-project.org/doc/manuals/r-release/R-lang.html R Language Definition] manual.
</syntaxhighlight>
* [http://r-pkgs.had.co.nz/src.html Compiled code] from "R packages" by Hadley Wickham
* [https://www.bioconductor.org/developers/how-to/c-debugging/ Debugging C/C++ code] from Bioconductor (case study)
* Same idea for the Rcpp situation. See [https://stackoverflow.com/questions/21226337/what-are-productive-ways-to-debug-rcpp-compiled-code-loaded-in-r-on-os-x-maveri What are productive ways to debug Rcpp compiled code loaded in R (on OS X Mavericks)?]


==== .Call ====
Argument matching is augmented by the functions
* [https://cran.rstudio.com/doc/manuals/r-release/R-exts.html#Calling-_002eCall Writing R Extensions] manual.
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.arg match.arg],
* [http://adv-r.had.co.nz/C-interface.html R’s C interface] from Advanced R by Hadley Wickham
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.call match.call]
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.fun match.fun].


==== Registering native routines ====
Access to the partial matching algorithm used by R is via [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/pmatch pmatch].
https://cran.rstudio.com/doc/manuals/r-release/R-exts.html#Registering-native-routines


Pay attention to the prefix argument '''.fixes''' (eg .fixes = "C_") in '''useDynLib()''' function in the NAMESPACE file.
=== Check function arguments ===
[https://blog.r-hub.io/2022/03/10/input-checking/ Checking the inputs of your R functions]: '''match.arg()''' , '''stopifnot()'''  


==== Example of debugging cor() function ====
'''stopifnot()''': function argument sanity check
Note that R's cor() function called a C function cor().
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/stopifnot stopifnot()]. ''stopifnot'' is a quick way to check multiple conditions on the input. so for instance. The code stops when either of the three conditions are not satisfied. However, it doesn't produce pretty error messages.  
<pre>
<pre>
stats::cor
stopifnot(condition1, condition2, ...)
....
.Call(C_cor, x, y, na.method, method == "kendall")
</pre>
</pre>
</li>
<li>[https://rud.is/b/2020/05/19/mining-r-4-0-0-changelog-for-nuggets-of-gold-1-stopifnot/ Mining R 4.0.0 Changelog for Nuggets of Gold] </li>
</ul>


A step-by-step screenshot of debugging using the GNU debugger '''gdb''' can be found on my Github repository https://github.com/arraytools/r-debug.
=== Lazy evaluation in R functions arguments ===
* http://adv-r.had.co.nz/Functions.html
* https://stat.ethz.ch/pipermail/r-devel/2015-February/070688.html
* https://twitter.com/_wurli/status/1451459394009550850


=== Locale bug (grep did not handle UTF-8 properly PR#16264) ===
'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.  
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264


=== Path length in dir.create() (PR#17206) ===
* Example 1. By default, R function arguments are lazy.
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)
 
=== install.package() error, R_LIBS_USER is empty in R 3.4.1 ===
* https://support.rstudio.com/hc/en-us/community/posts/115008369408-Since-update-to-R-3-4-1-R-LIBS-USER-is-empty and http://r.789695.n4.nabble.com/R-LIBS-USER-on-Ubuntu-16-04-td4740935.html. Modify '''/etc/R/Renviron''' (if you have a sudo right) by uncomment out line 43.
<pre>
<pre>
R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
f <- function(x) {
  999
}
f(stop("This is an error!"))
#> [1] 999
</pre>
</pre>
* https://stackoverflow.com/questions/44873972/default-r-personal-library-location-is-null. Modify '''$HOME/.Renviron''' by adding a line
 
* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
<pre>
<pre>
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"
add <- function(x) {
  force(x)
  function(y) x + y
}
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
</pre>
</pre>
* http://stat.ethz.ch/R-manual/R-devel/library/base/html/libPaths.html. Play with .libPaths()


On Mac & R 3.4.0 (it's fine)
* Example 3. Default arguments are evaluated inside the function.
<syntaxhighlight lang='rsplus'>
<pre>
> Sys.getenv("R_LIBS_USER")
f <- function(x = ls()) {
[1] "~/Library/R/3.4/library"
  a <- 1
> .libPaths()
  x
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"
}
</syntaxhighlight>
 
# ls() evaluated inside f:
f()
# [1] "a" "x"
 
# ls() evaluated in global environment:
f(ls())
# [1] "add"    "adders" "f"
</pre>
 
* Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
<pre>
x <- NULL
if (!is.null(x) && x > 0) {
 
}
</pre>
 
=== Use of functions as arguments ===
[https://www.njtierney.com/post/2019/09/29/unexpected-function/ Just Quickly: The unexpected use of functions as arguments]
 
=== body() ===
[https://stackoverflow.com/a/51548945 Remove top axis title base plot]
 
=== Return functions in R ===
* [https://win-vector.com/2015/04/03/how-and-why-to-return-functions-in-r/ How and why to return functions in R]
* See the doc & example from [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/taskCallback taskCallback - Create an R-level task callback manager]. [https://developer.r-project.org/TaskHandlers.pdf Top-level Task Callbacks in R].
* [https://purrple.cat/blog/2017/05/28/turn-r-users-insane-with-evil/ Turn R users insane with evil]


On Linux & R 3.3.1 (ARM)
=== anonymous function ===
<syntaxhighlight lang='rsplus'>
In R, the main difference between a lambda function (also known as an anonymous function) and a regular function is that a '''lambda function is defined without a name''', while a regular function is defined with a name.
> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"
</syntaxhighlight>


On Linux & R 3.4.1 (*Problem*)
<ul>
<li>See [[Tidyverse#Anonymous_functions|Tidyverse]] page
<li>But defining functions to use them only once is kind of overkill. That's why you can use so-called anonymous functions in R. For example, '''lapply(list(1,2,3), function(x) { x * x }) '''
<li>you can use lambda functions with many other functions in R that take a function as an argument. Some examples include '''sapply, apply, vapply, mapply, Map, Reduce, Filter''', and '''Find'''. These functions all work in a similar way to lapply by applying a function to elements of a list or vector.
<pre>
Reduce(function(x, y) x*y, list(1, 2, 3, 4)) # 24
</pre>
<li>[https://coolbutuseless.github.io/2019/03/13/anonymous-functions-in-r-part-1/ purrr anonymous function]
<li>[https://towardsdatascience.com/the-new-pipe-and-anonymous-function-syntax-in-r-54d98861014c The new pipe and anonymous function syntax in R 4.1.0]
<li>[http://adv-r.had.co.nz/Functional-programming.html#anonymous-functions Functional programming] from Advanced R
<li>[https://www.projectpro.io/recipes/what-are-anonymous-functions-r What are anonymous functions in R].
<syntaxhighlight lang='rsplus'>
<syntaxhighlight lang='rsplus'>
> Sys.getenv("R_LIBS_USER")
> (function(x) x * x)(3)
[1] ""
[1] 9
> .libPaths()
> (\(x) x * x)(3)
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[1] 9
[3] "/usr/lib/R/library"
</syntaxhighlight>
</syntaxhighlight>
</ul>


I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
== Backtick sign, infix/prefix/postfix operators ==
<syntaxhighlight lang='rsplus'>
The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in [http://adv-r.had.co.nz/Functions.html Advanced R] and [https://stackoverflow.com/a/36229703 What do backticks do in R?].
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
<pre>
> library(devtools)
iris %>%  `[[`("Species")
Error in library(devtools) : there is no package called 'devtools'
</pre>
 
'''[http://en.wikipedia.org/wiki/Infix_notation infix]''' operator.
<pre>
1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix
</pre>
 
Use with functions like sapply, e.g. '''sapply(1:5, `+`, 3) ''' .
 
== Error handling and exceptions, tryCatch(), stop(), warning() and message() ==
<ul>
<li>http://adv-r.had.co.nz/Exceptions-Debugging.html </li>
<li>[https://www.r-bloggers.com/2023/11/catch-me-if-you-can-exception-handling-in-r/ Catch Me If You Can: Exception Handling in R] </li>
<li>Temporarily disable warning messages
<pre>
# Method1:
suppressWarnings(expr)
 
# Method 2:
<pre>
defaultW <- getOption("warn")  
options(warn = -1)
[YOUR CODE]
options(warn = defaultW)
</pre>
</li>
<li>try() allows execution to continue even after an error has occurred. You can suppress the message with '''try(..., silent = TRUE)'''.
<pre>
out <- try({
  a <- 1
  b <- "x"
  a + b
})


# Specify lib.loc parameter will not help with the dependency package
elements <- list(1:10, c(-1, 10), c(T, F), letters)
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
results <- lapply(elements, log)
Error: package or namespace load failed for 'devtools':
is.error <- function(x) inherits(x, "try-error")
  .onLoad failed in loadNamespace() for 'devtools', details:
succeeded <- !sapply(results, is.error)
   call: loadNamespace(name)
</pre>
   error: there is no package called 'withr'
</li>
<li>tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
<pre>
tryCatch(expr, ..., finally)
 
show_condition <- function(code) {
  tryCatch(code,
    error = function(c) "error",
    warning = function(c) "warning",
    message = function(c) "message"
  )
}
show_condition(stop("!"))
#> [1] "error"
show_condition(warning("?!"))
#> [1] "warning"
show_condition(message("?"))
#> [1] "message"
show_condition(10)
#> [1] 10
</pre>
Below is another snippet from available.packages() function,
{{Pre}}
z <- tryCatch(download.file(....), error = identity)
if (!inherits(z, "error")) STATEMENTS
</pre>
</li>
<li>The return class from tryCatch() may not be fixed.
<pre>
result <- tryCatch({
  # Code that might generate an error or warning
  log(99)
}, warning = function(w) {
  # Code to handle warnings
  print(paste("Warning:", w))
}, error = function(e) {
  # Code to handle errors
  print(paste("Error:", e))
}, finally = {
  # Code to always run, regardless of whether an error or warning occurred
  print("Finished")
}) 
# character type. But if we remove 'finally', it will be numeric.
</pre>
<li>[https://www.bangyou.me/post/capture-logs/ Capture message, warnings and errors from a R function]
</li>
</ul>
 
=== suppressMessages() ===
suppressMessages(expression)
 
== List data type ==
=== Create an empty list ===
<pre>
out <- vector("list", length=3L) # OR out <- list()
for(j in 1:3) out[[j]] <- myfun(j)
 
outlist <- as.list(seq(nfolds))
</pre>
 
=== Nested list of data frames ===
An array can only hold data of a single type. read.csv() returns a data frame, which can contain both numerical and character data.
<pre>
res <- vector("list", 3)
names(res) <- paste0("m", 1:3)
for (i in seq_along(res)) {
  res[[i]] <- vector("list", 2)  # second-level list with 2 elements
  names(res[[i]]) <- c("fc", "pre")
}
 
res[["m1"]][["fc"]] <- read.csv()
 
head(res$m1$fc) # Same as res[["m1"]][["fc"]]
</pre>
 
=== Using $ in R on a List ===
[https://www.statology.org/dollar-sign-in-r/ How to Use Dollar Sign ($) Operator in R]
 
=== [http://adv-r.had.co.nz/Functions.html Calling a function given a list of arguments] ===
<pre>
> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5
</pre>
 
=== Descend recursively through lists ===
<nowiki>x[[c(5,3)]] </nowiki> is the same as <nowiki>x[[5]][[3]]</nowiki>. See [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Extract ?Extract].
 
=== Avoid if-else or switch ===
?plot.stepfun.
<pre>
y0 <- c(1,2,4,3)
sfun0  <- stepfun(1:3, y0, f = 0)
sfun.2 <- stepfun(1:3, y0, f = .2)
sfun1  <- stepfun(1:3, y0, right = TRUE)
 
tt <- seq(0, 3, by = 0.1)
op <- par(mfrow = c(2,2))
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
##-- This is  revealing :
plot(sfun0, verticals = FALSE,
    main = "stepfun(x, y0, f=f)  for f = 0, .2, 1")
 
for(i in 1:3)
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)
 
par(op)
</pre>
[[:File:StepfunExample.svg]]
 
== Open a new Window device ==
X11() or dev.new()
 
== par() ==
?par
 
=== text size (cex) and font size on main, lab & axis ===
* [https://www.statmethods.net/advgraphs/parameters.html Graphical Parameters] from statmethods.net.
* [https://designdatadecisions.wordpress.com/2015/06/09/graphs-in-r-overlaying-data-summaries-in-dotplots/ Overlaying Data Summaries in Dotplots]
 
Examples (default is 1 for each of them):
* cex.main=0.9
* cex.sub
* cex.lab=0.8, font.lab=2 (x/y axis labels)
* cex.axis=0.8, font.axis=2 (axis/tick text/labels)
* col.axis="grey50"
 
An quick example to increase font size ('''cex.lab''', '''cex.axis''', '''cex.main''') and line width ('''lwd''') in a line plot and '''cex''' & '''lwd''' in the legend.
<pre>
plot(x=x$mids, y=x$density, type="l",
    xlab="p-value", ylab="Density", lwd=2,
    cex.lab=1.5, cex.axis=1.5,
    cex.main=1.5, main = "")
lines(y$mids, y$density, lty=2, pwd=2)
lines(z$mids, z$density, lty=3, pwd=2)
legend('topright',legend = c('Method A','Method B','Method C'),
      lty=c(2,1,3), lwd=c(2,2,2), cex = 1.5, xjust = 0.5, yjust = 0.5)
</pre>
 
ggplot2 case (default font size is [https://ggplot2.tidyverse.org/articles/faq-customising.html 11 points]):
* plot.title
* plot.subtitle
* axis.title.x, axis.title.y: (x/y axis labels)
* axis.text.x & axis.text.y: (axis/tick text/labels)
<pre>
ggplot(df, aes(x, y)) +
  geom_point() +
  labs(title = "Title", subtitle = "Subtitle", x = "X-axis", y = "Y-axis") +
  theme(plot.title = element_text(size = 20),
        plot.subtitle = element_text(size = 15),
        axis.title.x = element_text(size = 15),
        axis.title.y = element_text(size = 15),
        axis.text.x = element_text(size = 10),
        axis.text.y = element_text(size = 10))
</pre>
 
=== Default font ===
* [https://stat.ethz.ch/R-manual/R-devel/library/grDevices/html/png.html ?png].  The default font family is '''Arial''' on Windows and '''Helvetica''' otherwise.
* ''sans''. See [https://www.r-bloggers.com/2015/08/changing-the-font-of-r-base-graphic-plots/ Changing the font of R base graphic plots]
* [http://www.cookbook-r.com/Graphs/Fonts/ Fonts] from ''Cookbook for R''. It seems ggplot2 also uses '''sans''' as the default font.
* [https://www.r-bloggers.com/2021/07/using-different-fonts-with-ggplot2/ Using different fonts with ggplot2]
* [https://r-coder.com/plot-r/#Font_family R plot font family]
* [https://r-coder.com/custom-fonts-r/ Add custom fonts in R]
 
=== layout ===
* [https://blog.rsquaredacademy.com/data-visualization-with-r-combining-plots/ Data Visualization with R - Combining Plots]
* http://datascienceplus.com/adding-text-to-r-plot/
 
=== reset the settings ===
{{Pre}}
op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1)
....
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1
</pre>
 
=== mtext (margin text) vs title ===
* https://datascienceplus.com/adding-text-to-r-plot/
* https://datascienceplus.com/mastering-r-plot-part-2-axis/
 
=== mgp (axis tick label locations or axis title) ===
# The margin line (in ‘mex’ units) for the axis title, axis labels and axis line.  Note that ‘mgp[1]’ affects the axis ‘title’ whereas ‘mgp[2:3]’ affect tick mark labels.  The default is ‘c(3, 1, 0)’. If we like to make the axis labels closer to an axis, we can use mgp=c(1.5, .5, 0) for example.
#* the default is c(3,1,0) which specify the margin line for the '''axis title''', '''axis labels''' and '''axis line'''.
#* the axis title is drawn in the fourth line of the margin starting from the plot region, the axis labels are drawn in the second line and the axis line itself is the first line.
# [https://www.r-bloggers.com/2010/06/setting-graph-margins-in-r-using-the-par-function-and-lots-of-cow-milk/ Setting graph margins in R using the par() function and lots of cow milk]
# [https://statisticsglobe.com/move-axis-label-closer-to-plot-in-base-r Move Axis Label Closer to Plot in Base R (2 Examples)]
# http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the '''labels/axis title''' (i.e. xlab and ylab in plot), the second the '''tick-mark labels''', and third the '''tick marks'''. The default is c(3, 1, 0).
 
=== move axis title closer to axis ===
* [https://r-charts.com/base-r/title/ Setting a title and a subtitle]. Default is around 1.7(?). [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/title ?title].
* [https://stackoverflow.com/a/30265996 move axis label closer to axis] '''title(, line)'''. This is useful when we use '''xaxt='n' ''' to hide the ticks and labels.
<pre>
title(ylab="Within-cluster variance", line=0,
      cex.lab=1.2, family="Calibri Light")
</pre>
 
=== pch and point shapes ===
[[:File:R pch.png]]
 
See [https://www.statmethods.net/advgraphs/parameters.html here].
 
* Full circle: pch=16
* Display all possibilities: ggpubr::show_point_shapes()
 
=== lty (line type) ===
[[:File:R lty.png]]
 
[https://finnstats.com/index.php/2021/06/11/line-types-in-r-lty-for-r-baseplot-and-ggplot/ Line types in R: Ultimate Guide For R Baseplot and ggplot]
 
See [http://www.sthda.com/english/wiki/line-types-in-r-lty here].
 
ggpubr::show_line_types()
 
=== las (label style) ===
0: The default, parallel to the axis
 
1: Always horizontal <syntaxhighlight lang='r' inline>boxplot(y~x, las=1)</syntaxhighlight>
 
2: Perpendicular to the axis
 
3: Always vertical
 
=== oma (outer margin), xpd, common title for two plots, 3 types of regions, multi-panel plots ===
<ul>
<li>The following trick is useful when we want to draw multiple plots with a common title.
{{Pre}}
par(mfrow=c(1,2),oma = c(0, 0, 2, 0))  # oma=c(0, 0, 0, 0) by default
plot(1:10,  main="Plot 1")
plot(1:100,  main="Plot 2")
mtext("Title for Two Plots", outer = TRUE, cex = 1.5) # outer=FALSE by default
</pre>
<li>[[PCA#Visualization|PCA plot]] example (the plot in the middle)
<li>For scatterplot3d() function, '''oma''' is not useful and I need to use '''xpd'''.
<li>[https://datascienceplus.com/mastering-r-plot-part-3-outer-margins/ Mastering R plot – Part 3: Outer margins] '''mtext()''' & '''par(xpd)'''.
<li>[https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par ?par] about '''xpd''' option
* If FALSE (default), all plotting is clipped to the plot region,
* If TRUE, all plotting is clipped to the figure region,
* If NA, all plotting is clipped to the device region.
<li>3 types of regions. See [https://www.benjaminbell.co.uk/2018/02/creating-multi-panel-plots-and-figures.html Creating multi-panel plots and figures using layout()] & [https://www.seehuhn.de/blog/122 publication-quality figures with R, part 2]
* plot region,
* figure region,
* device region.
<li>[https://www.benjaminbell.co.uk/2018/02/creating-multi-panel-plots-and-figures.html Creating multi-panel plots and figures using layout()] includes several tricks including creating a picture-in-picture plot.
</ul>
 
=== no.readonly ===
[https://www.zhihu.com/question/54116933 R语言里par(no.readonly=TURE)括号里面这个参数什么意思?], [https://www.jianshu.com/p/a716db5d30ef R-par()]
 
== Non-standard fonts in postscript and pdf graphics ==
https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41
 
 
== NULL, NA, NaN, Inf ==
https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/
 
== save()/load() vs saveRDS()/readRDS() vs dput()/dget() vs dump()/source() ==
# saveRDS() can only save one R object while save() does not have this constraint.
# saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See [http://www.fromthebottomoftheheap.net/2012/04/01/saving-and-loading-r-objects/ this post].
<pre>
x <- 5
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)
</pre>
 
[https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/dput dput]: Writes an ASCII text representation of an R object. The object name is not written (unlike '''dump''').
{{Pre}}
$ data(pbc, package = "survival")
$ names(pbc)
$ dput(names(pbc))
c("id", "time", "status", "trt", "age", "sex", "ascites", "hepato",
"spiders", "edema", "bili", "chol", "albumin", "copper", "alk.phos",
"ast", "trig", "platelet", "protime", "stage")
 
> iris2 <- iris[1:2, ]
> dput(iris2)
structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5,
3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L,
1L), .Label = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")
</pre>
 
=== User 'verbose = TRUE' in load() ===
When we use load(), it is helpful to add 'verbose =TRUE' to see what objects get loaded.
 
=== What are RDS files anyways ===
[https://www.statworx.com/de/blog/archive-existing-rds-files/ Archive Existing RDS Files]
 
== [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/all.equal ==, all.equal(), identical()] ==
* ==: exact match
* '''all.equal''': compare R objects x and y testing ‘near equality’
* identical: The safe and reliable way to test two objects for being exactly equal.
{{Pre}}
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
# [1] TRUE
identical(x, y)
# [1] FALSE
</pre>
 
Be careful about using "==" to return an index of matches in the case of data with missing values.
<pre>
R> c(1,2,NA)[c(1,2,NA) == 1]
[1]  1 NA
R> c(1,2,NA)[which(c(1,2,NA) == 1)]
[1] 1
</pre>
 
See also the [http://cran.r-project.org/web/packages/testthat/index.html testhat] package.
 
I found a case when I compare two objects where 1 is generated in ''Linux'' and the other is generated in ''macOS'' that identical() gives FALSE but '''all.equal()''' returns TRUE. The difference has a magnitude only e-17.
 
=== waldo ===
* https://waldo.r-lib.org/ or [https://cloud.r-project.org/web/packages/waldo/index.html CRAN]. Find and concisely describe the difference between a pair of R objects.
* [https://predictivehacks.com/how-to-compare-objects-in-r/ How To Compare Objects In R]
 
=== diffobj: Compare/Diff R Objects ===
https://cran.r-project.org/web/packages/diffobj/index.html
 
== testthat ==
* https://github.com/r-lib/testthat
* [http://www.win-vector.com/blog/2019/03/unit-tests-in-r/ Unit Tests in R]
* [https://davidlindelof.com/machine-learning-in-r-start-with-an-end-to-end-test/ Start with an End-to-End Test]
* [https://www.r-bloggers.com/2023/12/a-beautiful-mind-writing-testable-r-code/ A Beautiful Mind: Writing Testable R Code]
 
== tinytest ==
[https://cran.r-project.org/web/packages/tinytest/index.html tinytest]: Lightweight but Feature Complete Unit Testing Framework
 
[https://cran.r-project.org/web/packages/ttdo/index.html ttdo] adds support of the 'diffobj' package for 'diff'-style comparison of R objects.
 
== Numerical Pitfall ==
[http://bayesfactor.blogspot.com/2016/05/numerical-pitfalls-in-computing-variance.html Numerical pitfalls in computing variance]
{{Pre}}
.1 - .3/3
## [1] 0.00000000000000001388
</pre>
 
== Sys.getpid() ==
This can be used to monitor R process memory usage or stop the R process. See [https://stat.ethz.ch/pipermail/r-devel/2016-November/073360.html this post].
 
== Sys.getenv() & make the script more portable ==
Replace all the secrets from the script and replace them with '''Sys.getenv("secretname")'''. You can save the secrets in an '''.Renviron''' file next to the script in the same project.
<pre>
$ for v in 1 2; do MY=$v Rscript -e "Sys.getenv('MY')"; done
[1] "1"
[1] "2"
$ echo $MY
2
</pre>
 
== How to write R codes ==
* [https://youtu.be/7oyiPBjLAWY Code smells and feels] from R Consortium
** write simple conditions,
** handle class properly,
** return and exit early,
** polymorphism,
** switch() [e.g., switch(var, value1=out1, value2=out2, value3=out3). Several examples in [https://github.com/cran/glmnet/blob/master/R/assess.glmnet.R#L103 glmnet] ]
** case_when(),
** %||%.
* [https://appsilon.com/write-clean-r-code/ 5 Tips for Writing Clean R Code] – Leave Your Code Reviewer Commentless
** Comments
** Strings
** Loops
** Code Sharing
**Good Programming Practices
 
== How to debug an R code ==
[[Debug#R|Debug R]]
 
== Locale bug (grep did not handle UTF-8 properly PR#16264) ==
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264
 
== Path length in dir.create() (PR#17206) ==
https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)
 
== install.package() error, R_LIBS_USER is empty in R 3.4.1 & .libPaths() ==
* https://support.rstudio.com/hc/en-us/community/posts/115008369408-Since-update-to-R-3-4-1-R-LIBS-USER-is-empty and http://r.789695.n4.nabble.com/R-LIBS-USER-on-Ubuntu-16-04-td4740935.html. Modify '''/etc/R/Renviron''' (if you have a sudo right) by uncomment out line 43.
<pre>
R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
</pre>
* https://stackoverflow.com/questions/44873972/default-r-personal-library-location-is-null. Modify '''$HOME/.Renviron''' by adding a line
<pre>
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"
</pre>
* http://stat.ethz.ch/R-manual/R-devel/library/base/html/libPaths.html. Play with .libPaths()
 
On Mac & R 3.4.0 (it's fine)
{{Pre}}
> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"
</pre>
 
On Linux & R 3.3.1 (ARM)
{{Pre}}
> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"
</pre>
 
On Linux & R 3.4.1 (*Problematic*)
{{Pre}}
> Sys.getenv("R_LIBS_USER")
[1] ""
> .libPaths()
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[3] "/usr/lib/R/library"
</pre>
 
I need to specify the '''lib''' parameter when I use the '''install.packages''' command.
{{Pre}}
> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> library(devtools)
Error in library(devtools) : there is no package called 'devtools'
 
# Specify lib.loc parameter will not help with the dependency package
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
Error: package or namespace load failed for 'devtools':
  .onLoad failed in loadNamespace() for 'devtools', details:
   call: loadNamespace(name)
   error: there is no package called 'withr'


# A solution is to redefine .libPaths
# A solution is to redefine .libPaths
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works
> library(devtools) # Works
</syntaxhighlight>
</pre>
 
 
A better solution is to specify R_LIBS_USER in '''~/.Renviron''' file or '''~/.bash_profile'''; see [http://stat.ethz.ch/R-manual/R-patched/library/base/html/Startup.html ?Startup].
A better solution is to specify R_LIBS_USER in '''~/.Renviron''' file or '''~/.bash_profile'''; see [http://stat.ethz.ch/R-manual/R-patched/library/base/html/Startup.html ?Startup].
 
 
=== Using external data from within another package ===
== Using external data from within another package ==
https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/
https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/
 
 
=== How to run R scripts from the command line ===
== How to run R scripts from the command line ==
https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line
https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line
 
 
=== How to exit a sourced R script ===
== How to exit a sourced R script ==
* [http://stackoverflow.com/questions/25313406/how-to-exit-a-sourced-r-script How to exit a sourced R script]
* [http://stackoverflow.com/questions/25313406/how-to-exit-a-sourced-r-script How to exit a sourced R script]
* [http://r.789695.n4.nabble.com/Problem-using-the-source-function-within-R-functions-td907180.html Problem using the source-function within R-functions] ''' ''The best way to handle the generic sort of problem you are describing is to take those source'd files, and rewrite their content as functions to be called from your other functions.'' '''
* [http://r.789695.n4.nabble.com/Problem-using-the-source-function-within-R-functions-td907180.html Problem using the source-function within R-functions] ''' ''The best way to handle the generic sort of problem you are describing is to take those source'd files, and rewrite their content as functions to be called from your other functions.'' '''
 
* ‘source()’ and ‘example()’ have a new optional argument ‘catch.aborts’ which allows continued evaluation of the R code after an error. [https://developer.r-project.org/blosxom.cgi/R-devel/2023/10/11 4-devel] 2023/10/11.
=== Decimal point & decimal comma ===
 
Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark
== Decimal point & decimal comma ==
 
Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark
=== setting seed locally (not globally) in R ===
 
https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r
== setting seed locally (not globally) in R ==
 
https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r
=== R's internal C API ===
 
https://github.com/hadley/r-internals
== R's internal C API ==
 
https://github.com/hadley/r-internals
=== Random numbers: multivariate normal ===
 
Why [https://www.rdocumentation.org/packages/MASS/versions/7.3-49/topics/mvrnorm MASS::mvrnorm()] gives different result on Mac and Linux/Windows?
== cleancall package for C resource cleanup ==
[https://www.tidyverse.org/articles/2019/05/resource-cleanup-in-c-and-the-r-api/ Resource Cleanup in C and the R API]
 
== Random number generator ==
* https://cran.r-project.org/doc/manuals/R-exts.html#Random-numbers
* [https://stackoverflow.com/a/14555220 C code from R with .C(): random value is the same every time]
* [https://arxiv.org/pdf/2003.08009v2.pdf Random number generators produce collisions: Why, how many and more] Marius Hofert 2020 and the published paper in [https://www.tandfonline.com/doi/full/10.1080/00031305.2020.1782261 American Statistician] (including R code).
* R package examples. [https://github.com/cran/party/blob/5ddbd382f01fef2ab993401b43d1fc78d0b061fb/src/RandomForest.c party] package.
 
{{Pre}}
#include <R.h>
 
void myunif(){
  GetRNGstate();
  double u = unif_rand();
  PutRNGstate();
  Rprintf("%f\n",u);
}
</pre>
 
<pre>
$ R CMD SHLIB r_rand.c
$ R
R> dyn.load("r_rand.so")
R> set.seed(1)
R> .C("myunif")
0.265509
list()
R> .C("myunif")
0.372124
list()
R> set.seed(1)
R> .C("myunif")
0.265509
list()
</pre>
 
=== Test For Randomness ===
* [https://predictivehacks.com/how-to-test-for-randomness/ How To Test For Randomness]
* [https://www.r-bloggers.com/2021/08/test-for-randomness-in-r-how-to-check-dataset-randomness/ Test For Randomness in R-How to check Dataset Randomness]
 
== Different results in Mac and Linux ==
=== Random numbers: multivariate normal ===
Why [https://www.rdocumentation.org/packages/MASS/versions/7.3-49/topics/mvrnorm MASS::mvrnorm()] gives different result on Mac and Linux/Windows?
 
The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
* https://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
<ul>
<li>[https://stats.stackexchange.com/questions/61719/cholesky-versus-eigendecomposition-for-drawing-samples-from-a-multivariate-norma Cholesky versus eigendecomposition for drawing samples from a multivariate normal distribution]
 
See [https://gist.github.com/arraytools/0d7f0a02c233aefb9cefc6eb5f7b7754 this example]. A little more investigation shows the eigen values differ a little bit on macOS and Linux. See [https://gist.github.com/arraytools/0d7f0a02c233aefb9cefc6eb5f7b7754#file-mvtnorm_debug-r here].
</li>
</ul>
 
== rle() running length encoding ==
* https://en.wikipedia.org/wiki/Run-length_encoding
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
* https://rosettacode.org/wiki/Run-length_encoding
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
* R's [https://www.rdocumentation.org/packages/S4Vectors/versions/0.10.2/topics/Rle-class Rle class] from S4Vectors package which was used in for example [http://genomicsclass.github.io/book/pages/iranges_granges.html IRanges/GRanges/GenomicRanges] package
 
== citation() ==
{{Pre}}
citation()
citation("MASS")
toBibtex(citation())
</pre>
[https://www.r-bloggers.com/2024/05/notes-on-citing-r-and-r-packages/ Notes on Citing R and R Packages] with examples.
 
== R not responding request to interrupt stop process ==
[https://stackoverflow.com/a/43172530 R not responding request to interrupt stop process]. ''R is executing (for example) a C / C++ library call that doesn't provide R an opportunity to check for interrupts.'' It seems to match with the case I'm running (''dist()'' function).
 
== Monitor memory usage ==
* x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
* Windows: memory.size(max=TRUE)
* Linux
** RStudio: '''htop -p PID''' where PID is the process ID of ''/usr/lib/rstudio/bin/rsession'', not ''/usr/lib/rstudio/bin/rstudio''. This is obtained by running ''x <- rnorm(2*1e8)''. The object size can be obtained through ''print(object.size(x), units = "auto")''. Note that 1e8*8/2^20 = 762.9395.
** R: '''htop -p PID''' where PID is the process ID of ''/usr/lib/R/bin/exec/R''. Alternatively, use '''htop -p `pgrep -f /usr/lib/R/bin/exec/R`'''
** To find the peak memory usage '''grep VmPeak /proc/$PID/status'''
* '''mem_used()''' function from [https://cran.r-project.org/web/packages/pryr/index.html pryr] package. It is not correct or useful if I use it to check the value compared to the memory returned by '''jobload''' in biowulf. So I cannot use it to see the memory used in running mclapply().
* [https://cran.r-project.org/web/packages/peakRAM/index.html peakRAM]: Monitor the Total and Peak RAM Used by an Expression or Function
* [https://www.zxzyl.com/archives/1456/ Error: protect () : protection stack overflow] and [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/Memory ?Memory]
 
References:
* [https://unix.stackexchange.com/questions/554/how-to-monitor-cpu-memory-usage-of-a-single-process How to monitor CPU/memory usage of a single process?]. ''htop -p $PID'' is recommended. It only shows the percentage of memory usage.
* [https://stackoverflow.com/questions/774556/peak-memory-usage-of-a-linux-unix-process '''Peak''' memory usage of a linux/unix process] ''grep VmPeak /proc/$PID/status'' is recommended.
* [https://serverfault.com/a/264856 How can I see the memory usage of a Linux process?] ''pmap $PID | tail -n 1'' is recommended. It shows the memory usage in absolute value (eg 1722376K).
* [https://stackoverflow.com/a/6457769 How to check the amount of RAM in R] '''memfree <- as.numeric(system("awk '/MemFree/ {print $2}' /proc/meminfo", intern=TRUE)); memfree '''
 
== Monitor Data ==
[https://www.jstatsoft.org/article/view/v098i01?s=09 Monitoring Data in R with the lumberjack Package]
 
== Pushover ==
[https://rud.is/b/2020/01/29/monitoring-website-ssl-tls-certificate-expiration-times-with-r-openssl-pushoverr-and-dt/ Monitoring Website SSL/TLS Certificate Expiration Times with R, {openssl}, {pushoverr}, and {DT}]
 
[https://cran.r-project.org/web/packages/pushoverr/ pushoverr]
 
= Resource =
== Books ==
* [https://forwards.github.io/rdevguide/ R Development Guide] R Contribution Working Group
* [https://rviews.rstudio.com/2021/11/04/bookdown-org/ An R Community Public Library] 2011-11-04
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
* [http://statisticalestimation.blogspot.com/2016/11/learning-r-programming-by-reading-books.html Learning R programming by reading books: A book list]
* [http://www.stats.ox.ac.uk/pub/MASS4/ Modern Applied Statistics with S] by William N. Venables and Brian D. Ripley
* [http://dirk.eddelbuettel.com/code/rcpp.html Seamless R and C++ Integration with Rcpp] by Dirk Eddelbuettel
* [http://www.amazon.com/Advanced-Chapman-Hall-CRC-Series/dp/1466586966/ref=pd_sim_b_6?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB Advanced R] by Hadley Wickham 2014
** http://brettklamer.com/diversions/statistical/compile-hadleys-advanced-r-programming-to-a-pdf/ Compile Hadley's Advanced R to a PDF
* [https://b-rodrigues.github.io/fput/ Functional programming and unit testing for data munging with R] by Bruno Rodrigues
* [http://www.amazon.com/Cookbook-OReilly-Cookbooks-Paul-Teetor/dp/0596809158/ref=pd_sim_b_3?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB R Cookbook] by Paul Teetor
* [http://www.amazon.com/Machine-Learning-R-Brett-Lantz/dp/1782162143/ref=pd_sim_b_13?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 Machine Learning with R] by Brett Lantz
* [http://www.amazon.com/Everyone-Advanced-Analytics-Graphics-Addison-Wesley/dp/0321888030/ref=pd_sim_b_3?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 R for Everyone] by [http://www.jaredlander.com/r-for-everyone/ Jared P. Lander]
* [http://www.amazon.com/The-Art-Programming-Statistical-Software/dp/1593273843/ref=pd_sim_b_2?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 The Art of R Programming] by Norman Matloff
* [http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=pd_sim_b_3?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q Applied Predictive Modeling] by Max Kuhn
* [http://www.amazon.com/R-Action-Robert-Kabacoff/dp/1935182390/ref=pd_sim_b_17?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q R in Action] by Robert Kabacoff
* [http://www.amazon.com/The-Book-Michael-J-Crawley/dp/0470973927/ref=pd_sim_b_6?ie=UTF8&refRID=0CNF2XK8VBGF5A6W3NE3 The R Book] by Michael J. Crawley
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
* Data Manipulation with R by Phil Spector
* [https://www.datanovia.com/en/courses/data-manipulation-in-r/ DATA MANIPULATION IN R] by ALBOUKADEL KASSAMBARA
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
<ul>
<li>[https://github.com/csgillespie/efficientR Efficient R programming] by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the [https://csgillespie.github.io/efficientR/building-the-book-from-source.html Appendix]. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
{{Pre}}
# R 3.4.1
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
setwd("/tmp/efficientR/")
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
# generated pdf file is located _book/_main.pdf
 
bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
# generated epub file is located _book/_main.epub.
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
# but it is OK to run in an R terminal
</pre>
</li>
</ul>
* [https://learningstatisticswithr.com/book/ Learning statistics with R: A tutorial for psychology students and other beginners] by Danielle Navarro
* [https://rstats.wtf/ What They Forgot to Teach You About R] Jennifer Bryan & Jim Hester
* [http://knosof.co.uk/ESEUR/ Evidence-based Software Engineering] by Derek M. Jones
* [https://www.bigbookofr.com/index.html Big Book of R]
* [https://epirhandbook.com/?s=09 R for applied epidemiology and public health]
* [http://bendixcarstensen.com/EwR/ Epidemiology with R] and the [https://cran.r-project.org/web/packages/Epi/ Epi] package. [https://rdrr.io/cran/Epi/man/ci.lin.html ci.lin()] function to return the CI from glm() fit.
* [https://education.rstudio.com/learn/ RStudio &rarr; Finding Your Way To R]. Beginners/Intermediates/Experts
* [https://deepr.gagolewski.com/index.html Deep R Programming]
 
== Videos ==
* [https://www.infoworld.com/article/3411819/do-more-with-r-video-tutorials.html “Do More with R” video tutorials]. Search for R video tutorials by task, topic, or package. Most videos are shorter than 10 minutes.
* [https://www.youtube.com/@RLadiesGlobal/videos R-Ladies Global] (youtube)
 
=== Webinar ===
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
 
== useR! ==
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
* [https://www.youtube.com/watch?v=JacpQdj1Vfc&list=PL4IzsxWztPdnyAKQQLxA4ucpaCLdsKvZw UseR 2018 workshop and tutorials]
* [http://www.user2019.fr/ UseR! 2019], [https://github.com/sowla/useR2019-materials tutorial], [https://www.mango-solutions.com/blog/user2019-roundup-workflow-reproducibility-and-friends Better workflow]
* [https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/playlists UseR! 2020 & 2021]
* [https://rviews.rstudio.com/2021/09/09/a-guide-to-binge-watching-r-medicine/ A Guide to Binge Watching R / Medicine 2021]
* [https://t.co/QBZwNoPJsC UseR! 2022]
 
== R consortium ==
https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured
 
== Blogs, Tips, Socials, Communities ==
* Google: revolutionanalytics In case you missed it
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
* [http://onetipperday.sterding.com/2016/02/my-15-practical-tips-for.html My 15 practical tips for a bioinformatician]
* [http://blog.revolutionanalytics.com/2017/06/r-community.html The R community is one of R's best features]
* [https://hbctraining.github.io/main/ Bioinformatics Training at the Harvard Chan Bioinformatics Core]
* The R Blog <s>https://developer.r-project.org/Blog/public/</s> https://blog.r-project.org/
* [https://www.dataquest.io/blog/top-tips-for-learning-r-from-africa-rs-shelmith-kariuki/ Top Tips for Learning R from Africa R’s Shelmith Kariuki]
* [https://smach.github.io/R4JournalismBook/HowDoI.html How Do I? …(do that in R)] by Sharon Machlis
* [https://www.t4rstats.com/ Twitter for R programmers]
 
== Bug Tracking System ==
https://bugs.r-project.org/bugzilla3/ and [https://bugs.r-project.org/bugzilla3/query.cgi Search existing bug reports]. Remember to select 'All' in the Status drop-down list.
 
Use '''sessionInfo()'''.


The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See
== License ==
* https://stackoverflow.com/questions/11567613/different-random-number-generation-between-os
[http://www.win-vector.com/blog/2019/07/some-notes-on-gnu-licenses-in-r-packages/ Some Notes on GNU Licenses in R Packages]
* https://stats.stackexchange.com/questions/149321/generating-and-working-with-random-vectors-in-r
* [https://stats.stackexchange.com/questions/61719/cholesky-versus-eigendecomposition-for-drawing-samples-from-a-multivariate-norma Cholesky versus eigendecomposition for drawing samples from a multivariate normal distribution]
<syntaxhighlight lang='rsplus'>
set.seed(1234)
junk <- biospear::simdata(n=500, p=500, q.main = 10, q.inter = 10,
                          prob.tt = .5, m0=1, alpha.tt= -.5,
                          beta.main= -.5, beta.inter= -.5, b.corr = .7, b.corr.by=25,
                          wei.shape = 1, recr=3, fu=2, timefactor=1)
## Method 1: MASS::mvrnorm()
## This is simdata() has used. It gives different numbers on different OS.
##
library(MASS)
set.seed(1234)
m0 <-1
n <- 500
prob.tt <- .5
p <- 500
b.corr.by <- 25
b.corr <- .7
data <- data.frame(treat = rbinom(n, 1, prob.tt) - 0.5)
n.blocks <- p%/%b.corr.by
covMat <- diag(n.blocks) %x%
  matrix(b.corr^abs(matrix(1:b.corr.by, b.corr.by, b.corr.by, byrow = TRUE) -
                    matrix(1:b.corr.by, b.corr.by, b.corr.by)), b.corr.by, b.corr.by)
diag(covMat) <- 1
data <- cbind(data, mvrnorm(n, rep(0, p), Sigma = covMat))
range(data)
# Mac: -4.963827  4.133723
# Linux/Windows: -4.327635  4.408097
packageVersion("MASS")
# Mac: [1] ‘7.3.49’
# Linux: [1] ‘7.3.49’
# Windows: [1] ‘7.3.47’


R.version$version.string
[https://moderndata.plot.ly/why-dash-uses-the-mit-license/ Why Dash uses the mit license (and not a copyleft gpl license)]
# Mac: [1] "R version 3.4.3 (2017-11-30)"
# Linux: [1] "R version 3.4.4 (2018-03-15)"
# Windows: [1] "R version 3.4.3 (2017-11-30)"


## Method 2: mvtnorm::rmvnorm()
== Interview questions ==
library(mvtnorm)
* Does R store matrices in column-major order or row-major order?
set.seed(1234)
** Matrices are stored in column-major order, which means that elements are arranged and accessed by columns. This is in contrast to languages like Python, where matrices (or arrays) are typically stored in row-major order.
sigma <- matrix(c(4,2,2,3), ncol=2)
x <- rmvnorm(n=n, rep(0, p), sigma=covMat)
range(x)
# Mac: [1] -4.482566  4.459236
# Linux: [1] -4.482566  4.459236
 
## Method 3: mvnfast::rmvn()
set.seed(1234)
x <- mvnfast::rmvn(n, rep(0, p), covMat)
range(x)
# Mac: [1] -4.323585  4.355666
# Linux: [1] -4.323585  4.355666
 
library(microbenchmark)
library(MASS)
library(mvtnorm)
library(mvnfast)
microbenchmark(v1 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "eigen"),
              v2 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "svd"),
              v3 <- rmvnorm(n=n, rep(0, p), sigma=covMat, "chol"),
              v4 <- rmvn(n, rep(0, p), covMat),
              v5 <- mvrnorm(n, rep(0, p), Sigma = covMat))
Unit: milliseconds
expr      min        lq
v1 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "eigen") 296.55374 300.81089
v2 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "svd") 461.81867 466.98806
v3 <- rmvnorm(n = n, rep(0, p), sigma = covMat, "chol") 118.33759 120.01829
v4 <- rmvn(n, rep(0, p), covMat)  66.64675  69.89383
v5 <- mvrnorm(n, rep(0, p), Sigma = covMat) 291.19826 294.88038
mean    median        uq      max neval  cld
306.72485 301.99339 304.46662 335.6137  100    d
478.58536 470.44085 493.89041 571.7990  100    e
125.85427 121.26185 122.21361 151.1658  100  b 
71.67996  70.52985  70.92923 100.2622  100 a   
301.88144 296.76028 299.50839 346.7049  100  c 
</syntaxhighlight>
A little more investigation shows the eigen values differ a little bit on macOS and Linux.
<syntaxhighlight lang='rsplus'>
set.seed(1234); x <- mvrnorm(n, rep(0, p), Sigma = covMat)
debug(mvrnorm)
# eS --- macOS
# eS2 -- Linux
Browse[2]> range(abs(eS$values - eS2$values))
# [1] 0.000000e+00 1.776357e-15
Browse[2]> var(as.vector(eS$vectors))
[1] 0.002000006
Browse[2]> var(as.vector(eS2$vectors))
[1] 0.001999987
Browse[2]> all.equal(eS$values, eS2$values)
[1] TRUE
Browse[2]> which(eS$values != eS2$values)
  [1]  6  7  8  9  10  11  12  13  14  20  22  23  24  25  26  27  28  29
  ...
[451] 494 495 496 497 499 500
Browse[2]> range(abs(eS$vectors - eS2$vectors))
[1] 0.0000000 0.5636919
</syntaxhighlight>
 
=== rle() running length encoding ===
* https://en.wikipedia.org/wiki/Run-length_encoding
* [https://masterr.org/r/how-to-find-consecutive-repeats-in-r/ How to Find Consecutive Repeats in R]
* [https://www.r-bloggers.com/r-function-of-the-day-rle-2/amp/ R Function of the Day: rle]
* [https://blogs.reed.edu/ed-tech/2015/10/creating-nice-tables-using-r-markdown/ Creating nice tables using R Markdown]
* https://rosettacode.org/wiki/Run-length_encoding
* R's [https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/rle base::rle()] function
* R's [https://www.rdocumentation.org/packages/S4Vectors/versions/0.10.2/topics/Rle-class Rle class] from S4Vectors package which was used in for example [http://genomicsclass.github.io/book/pages/iranges_granges.html IRanges/GRanges/GenomicRanges] package


== Resource ==
* Explain the difference between == and === in R. Provide an example to illustrate their use.
=== Books ===
** The == operator is used for testing equality of values in R. It returns TRUE if the values on the left and right sides are equal, otherwise FALSE. The === operator does not exist in base R.  
* A list of recommended books http://blog.revolutionanalytics.com/2015/11/r-recommended-reading.html
* [http://statisticalestimation.blogspot.com/2016/11/learning-r-programming-by-reading-books.html Learning R programming by reading books: A book list]
* [http://www.stats.ox.ac.uk/pub/MASS4/ Modern Applied Statistics with S] by William N. Venables and Brian D. Ripley
* [http://dirk.eddelbuettel.com/code/rcpp.html Seamless R and C++ Integration with Rcpp] by Dirk Eddelbuettel
* [http://www.amazon.com/Advanced-Chapman-Hall-CRC-Series/dp/1466586966/ref=pd_sim_b_6?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB Advanced R] by Hadley Wickham 2014
** http://brettklamer.com/diversions/statistical/compile-hadleys-advanced-r-programming-to-a-pdf/ Compile Hadley's Advanced R to a PDF
* [http://www.brodrigues.co/functional_programming_and_unit_testing_for_data_munging/ Functional programming and unit testing for data munging with R] by Bruno Rodrigues
* [http://www.amazon.com/Cookbook-OReilly-Cookbooks-Paul-Teetor/dp/0596809158/ref=pd_sim_b_3?ie=UTF8&refRID=0C98YDK5MRSTRY0ZX1DB R Cookbook] by Paul Teetor
* [http://www.amazon.com/Machine-Learning-R-Brett-Lantz/dp/1782162143/ref=pd_sim_b_13?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 Machine Learning with R] by Brett Lantz
* [http://www.amazon.com/Everyone-Advanced-Analytics-Graphics-Addison-Wesley/dp/0321888030/ref=pd_sim_b_3?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 R for Everyone] by [http://www.jaredlander.com/r-for-everyone/ Jared P. Lander]
* [http://www.amazon.com/The-Art-Programming-Statistical-Software/dp/1593273843/ref=pd_sim_b_2?ie=UTF8&refRID=1851BAX3M17CK00VSMA6 The Art of R Programming] by Norman Matloff
* [http://www.amazon.com/Applied-Predictive-Modeling-Max-Kuhn/dp/1461468485/ref=pd_sim_b_3?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q Applied Predictive Modeling] by Max Kuhn
* [http://www.amazon.com/R-Action-Robert-Kabacoff/dp/1935182390/ref=pd_sim_b_17?ie=UTF8&refRID=0H3NMWX7KTRAEB32902Q R in Action] by Robert Kabacoff
* [http://www.amazon.com/The-Book-Michael-J-Crawley/dp/0470973927/ref=pd_sim_b_6?ie=UTF8&refRID=0CNF2XK8VBGF5A6W3NE3 The R Book] by Michael J. Crawley
* Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression by Frank E. Harrell
* Data Manipulation with R by Phil Spector
* [https://rviews.rstudio.com/2017/05/19/efficient_r_programming/ Review of Efficient R Programming]
* [http://r-pkgs.had.co.nz/ R packages: Organize, Test, Document, and Share Your Code] by Hadley Wicklam 2015
* [http://tidytextmining.com/ Text Mining with R: A Tidy Approach] and a [http://pacha.hk/2017-05-20_text_mining_with_r.html blog]
* [https://github.com/csgillespie/efficientR Efficient R programming] by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the [https://csgillespie.github.io/efficientR/building-the-book-from-source.html Appendix]. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
<syntaxhighlight lang='rsplus'>
# R 3.4.1
.libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
setwd("/tmp/efficientR/")
bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
# generated pdf file is located _book/_main.pdf


bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
* What is the purpose of the apply() function in R? How does it differ from the for loop?
# generated epub file is located _book/_main.epub.
** The apply() function in R is used to apply a function over the margins of an array or matrix. It is often used as an alternative to loops for applying a function to each row or column of a matrix.
# This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
# but it is OK to run in an R terminal
</syntaxhighlight>


=== Webinar ===
* Describe the concept of factors in R. How are they used in data manipulation and analysis?
* [https://www.rstudio.com/resources/webinars/ RStudio] & its [https://github.com/rstudio/webinars github] repository
** Factors in R are used to represent categorical data. They are an essential data type for statistical modeling and analysis. Factors store both the unique values that occur in a dataset and the corresponding integer codes used to represent those values.


=== useR! ===
* What is the significance
* http://blog.revolutionanalytics.com/2017/07/revisiting-user2017.html
of the attach() and detach() functions in R? When should they be used?
** A: The attach() function is used to add a data frame to the search path in R, making it easier to access variables within the data frame. The detach() function is used to remove a data frame from the search path, which can help avoid naming conflicts and reduce memory usage.


=== Blogs, Tips, Socials, Communities ===
* Explain the concept of vectorization in R. How does it impact the performance of R code?
* Google: revolutionanalytics In case you missed it
** Vectorization in R refers to the ability to apply operations to entire vectors or arrays at once, without needing to write explicit loops. This can significantly improve the performance of R code, as it allows operations to be performed in a more efficient, vectorized manner by taking advantage of R's underlying C code.
* [http://r4stats.com/articles/why-r-is-hard-to-learn/ Why R is hard to learn] by Bob Musenchen.
* [http://onetipperday.sterding.com/2016/02/my-15-practical-tips-for.html My 15 practical tips for a bioinformatician]
* [http://blog.revolutionanalytics.com/2017/06/r-community.html The R community is one of R's best features]
* [https://hbctraining.github.io/main/ Bioinformatics Training at the Harvard Chan Bioinformatics Core]


=== Bug Tracking System ===
* Describe the difference between data.frame and matrix in R. When would you use one over the other?
https://bugs.r-project.org/bugzilla3/ and [https://bugs.r-project.org/bugzilla3/query.cgi Search existing bug reports]. Remember to select 'All' in the Status drop-down list.
** A data.frame in R is a two-dimensional structure that can store different types of data (e.g., numeric, character, factor) in its columns. It is similar to a table in a database.
** A matrix in R is also a two-dimensional structure, but it can only store elements of the same data type. It is more like a mathematical matrix.
** You would use a data.frame when you have heterogeneous data (i.e., different types of data) and need to work with it as a dataset. You would use a matrix when you have homogeneous data (i.e., the same type of data) and need to perform matrix operations.


Use '''sessionInfo()'''.
* What are the benefits of using the dplyr package in R for data manipulation? Provide an example of how you would use dplyr to filter a data frame.
** The dplyr package provides a set of functions that make it easier to manipulate data frames in R.
** It uses a syntax that is easy to read and understand, making complex data manipulations more intuitive.
** To filter a data frame using dplyr, you can use the filter() function. For example, filter(df, column_name == value) would filter df to include only rows where column_name is equal to value.

Latest revision as of 13:31, 18 October 2024

Install and upgrade R

Here

New release

Online Editor

We can run R on web browsers without installing it on local machines (similar to [/ideone.com Ideone.com] for C++. It does not require an account either (cf RStudio).

rdrr.io

It can produce graphics too. The package I am testing (cobs) is available too.

rstudio.cloud

RDocumentation

The interactive engine is based on DataCamp Light

For example, tbl_df function from dplyr package.

The website DataCamp allows to run library() on the Script window. After that, we can use the packages on R Console.

Here is a list of (common) R packages that users can use on the web.

The packages on RDocumentation may be outdated. For example, the current stringr on CRAN is v1.2.0 (2/18/2017) but RDocumentation has v1.1.0 (8/19/2016).

Web Applications

R web applications

Creating local repository for CRAN and Bioconductor

R repository

Parallel Computing

See R parallel.

Cloud Computing

Install R on Amazon EC2

http://randyzwitch.com/r-amazon-ec2/

Bioconductor on Amazon EC2

http://www.bioconductor.org/help/bioconductor-cloud-ami/

Big Data Analysis

bigmemory, biganalytics, bigtabulate

ff, ffbase

biglm

data.table

See data.table.

disk.frame

Split-apply-combine for Maximum Likelihood Estimation of a linear model

Apache arrow

Reproducible Research

Reproducible Environments

https://rviews.rstudio.com/2019/04/22/reproducible-environments/

checkpoint package

Some lessons in R coding

  1. don't use rand() and srand() in c. The result is platform dependent. My experience is Ubuntu/Debian/CentOS give the same result but they are different from macOS and Windows. Use Rcpp package and R's random number generator instead.
  2. don't use list.files() directly. The result is platform dependent even different Linux OS. An extra sorting helps!

Useful R packages

Rcpp

http://cran.r-project.org/web/packages/Rcpp/index.html. See more here.

RInside : embed R in C++ code

Ubuntu

With RInside, R can be embedded in a graphical application. For example, $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt directory includes source code of a Qt application to show a kernel density plot with various options like kernel functions, bandwidth and an R command text box to generate the random data. See my demo on Youtube. I have tested this qtdensity example successfully using Qt 4.8.5.

  1. Follow the instruction cairoDevice to install required libraries for cairoDevice package and then cairoDevice itself.
  2. Install Qt. Check 'qmake' command becomes available by typing 'whereis qmake' or 'which qmake' in terminal.
  3. Open Qt Creator from Ubuntu start menu/Launcher. Open the project file $HOME/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/qt/qtdensity.pro in Qt Creator.
  4. Under Qt Creator, hit 'Ctrl + R' or the big green triangle button on the lower-left corner to build/run the project. If everything works well, you shall see the interactive program qtdensity appears on your desktop.

File:qtdensity.png

With RInside + Wt web toolkit installed, we can also create a web application. To demonstrate the example in examples/wt directory, we can do

cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
make
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080

Then we can go to the browser's address bar and type http://localhost:8080 to see how it works (a screenshot is in here).

Windows 7

To make RInside works on Windows OS, try the following

  1. Make sure R is installed under C:\ instead of C:\Program Files if we don't want to get an error like g++.exe: error: Files/R/R-3.0.1/library/RInside/include: No such file or directory.
  2. Install RTools
  3. Instal RInside package from source (the binary version will give an error )
  4. Create a DOS batch file containing necessary paths in PATH environment variable
@echo off
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set R_HOME=C:\R\R-3.0.1
echo Setting environment for using R
cmd

In the Windows command prompt, run

cd C:\R\R-3.0.1\library\RInside\examples\standard
make -f Makefile.win

Now we can test by running any of executable files that make generates. For example, rinside_sample0.

rinside_sample0

As for the Qt application qdensity program, we need to make sure the same version of MinGW was used in building RInside/Rcpp and Qt. See some discussions in

So the Qt and Wt web tool applications on Windows may or may not be possible.

GUI

Qt and R

tkrplot

On Ubuntu, we need to install tk packages, such as by

sudo apt-get install tk-dev

reticulate - Interface to 'Python'

Python -> reticulate

Hadoop (eg ~100 terabytes)

See also HighPerformanceComputing

RHadoop

Snowdoop: an alternative to MapReduce algorithm

XML

On Ubuntu, we need to install libxml2-dev before we can install XML package.

sudo apt-get update
sudo apt-get install libxml2-dev

On CentOS,

yum -y install libxml2 libxml2-devel

XML

library(XML)

# Read and parse HTML file
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)

# Extract all the paragraphs (HTML tag is p, starting at
# the root of the document). Unlist flattens the list to
# create a character vector.
doc.text = unlist(xpathApply(doc.html, '//p', xmlValue))

# Replace all by spaces
doc.text = gsub('\n', ' ', doc.text)

# Join all the elements of the character vector into a single
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')

This post http://stackoverflow.com/questions/25315381/using-xpathsapply-to-scrape-xml-attributes-in-r can be used to monitor new releases from github.com.

> library(RCurl) # getURL()
> library(XML)   # htmlParse and xpathSApply
> xData <- getURL("https://github.com/alexdobin/STAR/releases")
> doc = htmlParse(xData)
> plain.text <- xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
  # I look at the source code and search 2.5.3a and find the tag as
  # 2.5.3a
> plain.text
 [1] "2.5.3a"      "2.5.2b"      "2.5.2a"      "2.5.1b"      "2.5.1a"     
 [6] "2.5.0c"      "2.5.0b"      "STAR_2.5.0a" "STAR_2.4.2a" "STAR_2.4.1d"
>
> # try bwa
> > xData <- getURL("https://github.com/lh3/bwa/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
[1] "v0.7.15" "v0.7.13"

> # try picard
> xData <- getURL("https://github.com/broadinstitute/picard/releases")
> doc = htmlParse(xData)
> xpathSApply(doc, "//span[@class='css-truncate-target']", xmlValue)
 [1] "2.9.1" "2.9.0" "2.8.3" "2.8.2" "2.8.1" "2.8.0" "2.7.2" "2.7.1" "2.7.0"
[10] "2.6.0"

This method can be used to monitor new tags/releases from some projects like Cura, BWA, Picard, STAR. But for some projects like sratools the class attribute in the span element ("css-truncate-target") can be different (such as "tag-name").

xmlview

RCurl

On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)

# Test on Ubuntu 14.04
sudo apt-get install libxml2-dev
sudo apt-get install libcurl4-openssl-dev

Scrape google scholar results

https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R

No google ID is required

Seems not work

 Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  : 
  arguments imply differing number of rows: 2, 0 

devtools

devtools package depends on Curl. It actually depends on some system files. If we just need to install a package, consider the remotes package which was suggested by the BiocManager package.

# Ubuntu 14.04
sudo apt-get install libcurl4-openssl-dev

# Ubuntu 16.04, 18.04
sudo apt-get install build-essential libcurl4-gnutls-dev libxml2-dev libssl-dev

# Ubuntu 20.04
sudo apt-get install -y libxml2-dev libcurl4-openssl-dev libssl-dev

Lazy-load database XXX is corrupt. internal error -3. It often happens when you use install_github to install a package that's currently loaded; try restarting R and running the app again.

NB. According to the output of apt-cache show r-cran-devtools, the binary package is very old though apt-cache show r-base and supported packages like survival shows the latest version.

httr

httr imports curl, jsonlite, mime, openssl and R6 packages.

When I tried to install httr package, I got an error and some message:

Configuration failed because openssl was not found. Try installing:
 * deb: libssl-dev (Debian, Ubuntu, etc)
 * rpm: openssl-devel (Fedora, CentOS, RHEL)
 * csw: libssl_dev (Solaris)
 * brew: openssl (Mac OSX)
If openssl is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a openssl.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
--------------------------------------------------------------------
ERROR: configuration failed for package ‘openssl’

It turns out after I run sudo apt-get install libssl-dev in the terminal (Debian), it would go smoothly with installing httr package. Nice httr!

Real example: see this post. Unfortunately I did not get a table result; I only get an html file (R 3.2.5, httr 1.1.0 on Ubuntu and Debian).

Since httr package was used in many other packages, take a look at how others use it. For example, aRxiv package.

A package to download free Springer books during Covid-19 quarantine, An update to "An adventure in downloading books" (rvest package)

curl

curl is independent of RCurl package.

library(curl)
h <- new_handle()
handle_setform(h,
  name="aaa", email="bbb"
)
req <- curl_fetch_memory("http://localhost/d/phpmyql3_scripts/ch02/form2.html", handle = h)
rawToChar(req$content)

rOpenSci packages

rOpenSci contains packages that allow access to data repositories through the R statistical programming environment

remotes

Download and install R packages stored in 'GitHub', 'BitBucket', or plain 'subversion' or 'git' repositories. This package is a lightweight replacement of the 'install_*' functions in 'devtools'. Also remotes does not require any extra OS level library (at least on Ubuntu 16.04).

Example:

# https://github.com/henrikbengtsson/matrixstats
remotes::install_github('HenrikBengtsson/matrixStats@develop')

DirichletMultinomial

On Ubuntu, we do

sudo apt-get install libgsl0-dev

Create GUI

gWidgets

GenOrd: Generate ordinal and discrete variables with given correlation matrix and marginal distributions

here

json

R web -> json

Map

leaflet

choroplethr

ggplot2

How to make maps with Census data in R

googleVis

See an example from RJSONIO above.

googleAuthR

Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.

gtrendsR - Google Trends

quantmod

Maintaining a database of price files in R. It consists of 3 steps.

  1. Initial data downloading
  2. Update existing data
  3. Create a batch file

caret

Tool for connecting Excel with R

write.table

Output a named vector

vec <- c(a = 1, b = 2, c = 3)
write.csv(vec, file = "my_file.csv", quote = F)
x = read.csv("my_file.csv", row.names = 1)
vec2 <- x[, 1]
names(vec2) <- rownames(x)
all.equal(vec, vec2)

# one liner: row names of a 'matrix' become the names of a vector
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
all.equal(vec, vec3)

Avoid leading empty column to header

write.table writes unwanted leading empty column to header when has rownames

write.table(a, 'a.txt', col.names=NA)
# Or better by
write.table(data.frame("SeqId"=rownames(a), a), "a.txt", row.names=FALSE)

Add blank field AND column names in write.table

  • write.table(, row.names = TRUE) will miss one element on the 1st row when "row.names = TRUE" which is enabled by default.
    • Suppose x is (n x 2)
    • write.table(x, sep="\t") will generate a file with 2 element on the 1st row
    • read.table(file) will return an object with a size (n x 2)
    • read.delim(file) and read.delim2(file) will also be correct
  • Note that write.csv() does not have this issue that write.table() has
    • Suppose x is (n x 2)
    • Suppose we use write.csv(x, file). The csv file will be ((n+1) x 3) b/c the header row.
    • If we use read.csv(file), the object is (n x 3). So we need to use read.csv(file, row.names = 1)
  • adding blank field AND column names in write.table(); write.table writes unwanted leading empty column to header when has rownames
write.table(a, 'a.txt', col.names=NA)
  • readr::write_tsv() does not include row names in the output file

read.delim(, row.names=1) and write.table(, row.names=TRUE)

How to Use read.delim Function in R

Case 1: no row.names

write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=FALSE)
my_df <- read.delim('my_data.txt')  # the rownames will be 1, 2, 3, ...

Case 2: with row.names. Note: if we open the text file in Excel, we'll see the 1st row is missing one header at the end. It is actually missing the column name for the 1st column.

write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=TRUE)
my_df <- read.delim('my_data.txt')  # it will automatically assign the rownames

Read/Write Excel files package

  • http://www.milanor.net/blog/?p=779
  • flipAPI. One useful feature of DownloadXLSX, which is not supported by the readxl package, is that it can read Excel files directly from the URL.
  • xlsx: depends on Java
  • openxlsx: not depend on Java. Depend on zip application. On Windows, it seems to be OK without installing Rtools. But it can not read xls file; it works on xlsx file.
  • readxl: it does not depend on anything although it can only read but not write Excel files.
    • It is part of tidyverse package. The readxl website provides several articles for more examples.
    • readxl webinar.
    • One advantage of read_excel (as with read_csv in the readr package) is that the data imports into an easy to print object with three attributes a tbl_df, a tbl and a data.frame.
    • For writing to Excel formats, use writexl or openxlsx package.
library(readxl)
read_excel(path, sheet = NULL, range = NULL, col_names = TRUE, 
    col_types = NULL, na = "", trim_ws = TRUE, skip = 0, n_max = Inf, 
    guess_max = min(1000, n_max), progress = readxl_progress(), 
    .name_repair = "unique")
# Example
read_excel(path, range = cell_cols("c:cx"), col_types = "numeric")
  • writexl: zero dependency xlsx writer for R
library(writexl)
mylst <- list(sheet1name = df1, sheet2name = df2)
write_xlsx(mylst, "output.xlsx")

For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).

> head(read_excel("~/Downloads/BRCA.xls", 4)[ , -9], 3)
  UniqueID (Double-click) CloneID UGCluster
1                   HK1A1   21652 Hs.445981
2                   HK1A2   22012 Hs.119177
3                   HK1A4   22293 Hs.501376
                                                    Name Symbol EntrezID
1 Catenin (cadherin-associated protein), alpha 1, 102kDa CTNNA1     1495
2                              ADP-ribosylation factor 3   ARF3      377
3                          Uroporphyrinogen III synthase   UROS     7390
  Chromosome      Cytoband ChimericClusterIDs Filter
1   5.000000        5q31.2               <NA>      1
2  12.000000         12q13               <NA>      1
3       <NA> 10q25.2-q26.3               <NA>      1

The hidden worksheets become visible (Not sure what are those first rows mean in the output).

> excel_sheets("~/Downloads/BRCA.xls")
DEFINEDNAME: 21 00 00 01 0b 00 00 00 02 00 00 00 00 00 00 0d 3b 01 00 00 00 9a 0c 00 00 1a 00 
DEFINEDNAME: 21 00 00 01 0b 00 00 00 04 00 00 00 00 00 00 0d 3b 03 00 00 00 9b 0c 00 00 0a 00 
DEFINEDNAME: 21 00 00 01 0b 00 00 00 03 00 00 00 00 00 00 0d 3b 02 00 00 00 9a 0c 00 00 06 00 
[1] "Experiment descriptors" "Filtered log ratio"     "Gene identifiers"      
[4] "Gene annotations"       "CollateInfo"            "GeneSubsets"           
[7] "GeneSubsetsTemp"       

The Chinese character works too.

> read_excel("~/Downloads/testChinese.xlsx", 1)
   中文 B C
1     a b c
2     1 2 3

To read all worksheets we need a convenient function

read_excel_allsheets <- function(filename) {
    sheets <- readxl::excel_sheets(filename)
    sheets <- sheets[-1] # Skip sheet 1
    x <- lapply(sheets, function(X) readxl::read_excel(filename, sheet = X, col_types = "numeric"))
    names(x) <- sheets
    x
}
dcfile <- "table0.77_dC_biospear.xlsx"
dc <- read_excel_allsheets(dcfile)
# Each component (eg dc1) is a tibble.

readr

Compared to base equivalents like read.csv(), readr is much faster and gives more convenient output: it never converts strings to factors, can parse date/times, and it doesn’t munge the column names.

1.0.0 released. readr 2.0.0 adds built-in support for reading multiple files at once, fast multi-threaded lazy reading and automatic guessing of delimiters among other changes.

Consider a text file where the table (6100 x 22) has duplicated row names and the (1,1) element is empty. The column names are all unique.

  • read.delim() will treat the first column as rownames but it does not allow duplicated row names. Even we use row.names=NULL, it still does not read correctly. It does give warnings (EOF within quoted string & number of items read is not a multiple of the number of columns). The dim is 5177 x 22.
  • readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
  • data.table::fread(Filename, sep = "\t") will detect the number of column names is less than the number of columns. Added 1 extra default column name for the first column which is guessed to be row names or an index. The dim is 6100 x 22. (Winner!)

The readr::read_csv() function is as fast as data.table::fread() function. For files beyond 100MB in size fread() and read_csv() can be expected to be around 5 times faster than read.csv(). See 5.3 of Efficient R Programming book.

Note that data.table::fread() can read a selection of the columns.

Speed comparison

The Fastest Way To Read And Write Files In R. data.table >> readr >> base.

ggplot2

See ggplot2

Data Manipulation & Tidyverse

See Tidyverse.

Data Science

See Data science page

microbenchmark & rbenchmark

Plot, image

jpeg

If we want to create the image on this wiki left hand side panel, we can use the jpeg package to read an existing plot and then edit and save it.

We can also use the jpeg package to import and manipulate a jpg image. See Fun with Heatmaps and Plotly.

EPS/postscript format

  • Don't use postscript().
  • Use cairo_ps(). See aving High-Resolution ggplots: How to Preserve Semi-Transparency. It works on base R plots too.
    cairo_ps(filename = "survival-curves.eps",
             width = 7, height = 7, pointsize = 12,
             fallback_resolution = 300)
    print(p) # or any base R plots statements
    dev.off()
  • Export a graph to .eps file with R.
    • The results looks the same as using cairo_ps().
    • The file size by setEPS() + postscript() is quite smaller compared to using cairo_ps().
    • However, grep can find the characters shown on the plot generated by cairo_ps() but not setEPS() + postscript().
    setEPS()
    postscript("whatever.eps") # 483 KB
    plot(rnorm(20000))
    dev.off()
    # grep rnorm whatever.eps # Not found!
    
    cairo_ps("whatever_cairo.eps")   # 2.4 MB
    plot(rnorm(20000))
    dev.off()
    # grep rnorm whatever_cairo.eps  # Found!
    
  • View EPS files
    • Linux: evince. It is installed by default.
    • Mac: evince. brew install evince
    • Windows. Install ghostscript 9.20 (10.x does not work with ghostview/GSview) and ghostview/GSview (5.0). In Ghostview, open Options -> Advanced Configure. Change Ghostscript DLL path AND Ghostscript include Path according to the ghostscript location ("C:\.
  • Edit EPS files: Inkscape
    • Step 1: open the EPS file
    • Step 2: EPS Input: Determine page orientation from text direction 'Page by page' - OK
    • Step 3: PDF Import Settings: default is "Internal import", but we shall choose "Cairo import".
    • Step 4: Zoom in first.
    • Step 5: Click on Layers and Objects tab on the RHS. Now we can select any lines or letters and edit them as we like. The selected objects are highlighted in the "Layers and Objects" panel. That is, we can select multiple objects using object names. The selected objects can be rotated (Object -> Rotate 90 CW), for example.
    • Step 6: We can save the plot as any formats like svg, eps, pdf, html, pdf, ...

png and resolution

It seems people use res=300 as a definition of high resolution.

  • Bottom line: fix res=300 and adjust height/width as needed. The default is res=72, height=width=480. If we increase res=300, the text font size will be increased, lines become thicker and the plot looks like a zoom-in.
  • Saving high resolution plot in png.
    png("heatmap.png", width = 8, height = 6, units='in', res = 300) 
    # we can adjust width/height as we like
    # the pixel values will be width=8*300 and height=6*300 which is equivalent to 
    # 8*300 * 6*300/10^6 = 4.32 Megapixels (1M pixels = 10^6 pixels) in camera's term
    # However, if we use png(, width=8*300, height=6*300, units='px'), it will produce
    # a plot with very large figure body and tiny text font size.
    
    # It seems the following command gives the same result as above
    png("heatmap.png", width = 8*300, height = 6*300, res = 300) # default units="px"
    
  • Chapter 14.5 Outputting to Bitmap (PNG/TIFF) Files by R Graphics Cookbook
    • Changing the resolution affects the size (in pixels) of graphical objects like text, lines, and points.
  • 10 tips for making your R graphics look their best David Smith
    • In Word you can resize the graphic to an appropriate size, but the high resolution gives you the flexibility to choose a size while not compromising on the quality. I'd recommend at least 1200 pixels on the longest side for standard printers.
  • ?png. The png function has default settings ppi=72, height=480, width=480, units="px".
    • By default no resolution is recorded in the file, except for BMP.
    • BMP vs PNG format. If you need a smaller file size and don’t mind a lossless compression, PNG might be a better choice. If you need to retain as much detail as possible and don’t mind a larger file size, BMP could be the way to go.
      • Compression: BMP files are raw and uncompressed, meaning they’re large files that retain as much detail as possible. On the other hand, PNG files are compressed but still lossless. This means you can reduce or expand PNGs without losing any information.
      • File size: BMPs are larger than PNGs. This is because PNG files automatically compress, and can be compressed again to make the file even smaller.
      • Common uses: BMP contains a maximum amount of details while PNGs are good for small illustrations, sketches, drawings, logos and icons.
      • Quality: No difference
      • Transparency: PNG supports transparency while BMP doesn't
  • Some comparison about the ratio
    • 11/8.5=1.29 (A4 paper)
    • 8/6=1.33 (plot output)
    • 1440/900=1.6 (my display)
  • Setting resolution and aspect ratios in R
  • The difference of res parameter for a simple plot. How to change the resolution of a plot in base R?
  • High Resolution Figures in R.
  • High resolution graphics with R
  • R plot: size and resolution
  • How can I increase the resolution of my plot in R?, devEMF package
  • See Images -> Anti-alias.
  • How to check DPI on PNG
    • The width of a PNG file in terms of inches cannot be determined directly from the file itself, as the file contains pixel dimensions, not physical dimensions. However, you can calculate the width in inches if you know the resolution (DPI, dots per inch) of the image. Remember that converting pixel measurements to physical measurements like inches involves a specific resolution (DPI), and different devices may display the same image at different sizes due to having different resolutions.
  • Cairo case.

PowerPoint

  • For PP presentation, I found it is useful to use svg() to generate a small size figure. Then when we enlarge the plot, the text font size can be enlarged too. According to svg, by default, width = 7, height = 7, pointsize = 12, family = sans.
  • Try the following code. The font size is the same for both plots/files. However, the first plot can be enlarged without losing its quality.
    svg("svg4.svg", width=4, height=4)
    plot(1:10, main="width=4, height=4")
    dev.off()
    
    svg("svg7.svg", width=7, height=7) # default
    plot(1:10, main="width=7, height=7")
    dev.off()
    

magick

https://cran.r-project.org/web/packages/magick/

See an example here I created.

Cairo

See White strips problem in png() or tiff().

geDevices

cairoDevice

PS. Not sure the advantage of functions in this package compared to R's functions (eg. Cairo_svg() vs svg()).

For ubuntu OS, we need to install 2 libraries and 1 R package RGtk2.

sudo apt-get install libgtk2.0-dev libcairo2-dev

On Windows OS, we may got the error: unable to load shared object 'C:/Program Files/R/R-3.0.2/library/cairoDevice/libs/x64/cairoDevice.dll' . We need to follow the instruction in here.

dpi requirement for publication

For import into PDF-incapable programs (MS Office)

sketcher: photo to sketch effects

https://htsuda.net/sketcher/

httpgd

igraph

R web -> igraph

Identifying dependencies of R functions and scripts

https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts

library(mvbutils)
foodweb(where = "package:batr")

foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)

foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)

iterators

Iterator is useful over for-loop if the data is already a collection. It can be used to iterate over a vector, data frame, matrix, file

Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.

Colors

  • scales package. This is used in ggplot2 package.
  • colorspace: A Toolbox for Manipulating and Assessing Colors and Palettes. Popular! Many reverse imports/suggests; e.g. ComplexHeatmap. See my ggplot2 page.
    hcl_palettes(plot = TRUE) # a quick overview
    hcl_palettes(palette = "Dark 2", n=5, plot = T)
    q4 <- qualitative_hcl(4, palette = "Dark 3")
    
  • convert hex value to color names
    library(plotrix)
    sapply(rainbow(4), color.id) # color.id is a function
              # it is used to identify closest match to a color
    sapply(palette(), color.id)
    sapply(RColorBrewer::brewer.pal(4, "Set1"), color.id)
    

Below is an example using the option scale_fill_brewer(palette = "Paired"). See the source code at gist. Note that only set1 and set3 palettes in qualitative scheme can support up to 12 classes.

According to the information from the colorbrew website, qualitative schemes do not imply magnitude differences between legend classes, and hues are used to create the primary visual differences between classes.

File:GgplotPalette.svg

colortools

Tools that allow users generate color schemes and palettes

colourpicker

A Colour Picker Tool for Shiny and for Selecting Colours in Plots

eyedroppeR

Select colours from an image in R with {eyedroppeR}

rex

Friendly Regular Expressions

formatR

The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.

See also this discussion on stackoverflow talks about R code reformatting.

library(formatR)
tidy_source("Input.R", file = "output.R", width.cutoff=70)
tidy_source("clipboard") 
# default width is getOption("width") which is 127 in my case.

Some issues

  • Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
cat("abcd",
    # This is my comment
    "defg")

will result in

> tidy_source("clipboard")
Error in base::parse(text = code, srcfile = NULL) : 
  3:1: unexpected string constant
2: invisible(".BeGiN_TiDy_IdEnTiFiEr_HaHaHa# This is my comment.HaHaHa_EnD_TiDy_IdEnTiFiEr")
3: "defg"
   ^
  • Comments appearing at the end of a line within a long complete statement won't break tidy_source() but tidy_source() cannot re-locate/tidy the comma sign.
cat("abcd"
    ,"defg"   # This is my comment
  ,"ghij")

will become

cat("abcd", "defg"  # This is my comment
, "ghij") 

Still bad!!

  • Comments appearing at the end of a line within a long complete statement breaks tidy_source() function. For example,
cat("</p>",
	"<HR SIZE=5 WIDTH=\"100%\" NOSHADE>",
	ifelse(codeSurv == 0,"<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>", #4/9/09
	                     "<h3><a name='Genes'><b><u>Genes significantly associated with survival:</u></b></a></h3>"), 
	file=ExternalFileName, sep="\n", append=T)

will result in

> tidy_source("clipboard", width.cutoff=70)
Error in base::parse(text = code, srcfile = NULL) : 
  3:129: unexpected SPECIAL
2: "<HR SIZE=5 WIDTH=\"100%\" NOSHADE>" ,
3: ifelse ( codeSurv == 0 , "<h3><a name='Genes'><b><u>Genes which are differentially expressed among classes:</u></b></a></h3>" , %InLiNe_IdEnTiFiEr%
  • width.cutoff parameter is not always working. For example, there is no any change for the following snippet though I hope it will move the cat() to the next line.
if (codePF & !GlobalTest & !DoExactPermTest) cat(paste("Multivariate Permutations test was computed based on", 
    NumPermutations, "random permutations"), "<BR>", " ", file = ExternalFileName, 
    sep = "\n", append = T)
  • It merges lines though I don't always want to do that. For example
cat("abcd"
    ,"defg"  
  ,"ghij")

will become

cat("abcd", "defg", "ghij") 

styler

https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.

Download papers

biorxivr

Search and Download Papers from the bioRxiv Preprint Server (biology)

aRxiv

Interface to the arXiv API

pdftools

aside: set it aside

An RStudio addin to run long R commands aside your current session.

Teaching

  • smovie: Some Movies to Illustrate Concepts in Statistics

Organize R research project

How to save (and load) datasets in R (.RData vs .Rds file)

How to save (and load) datasets in R: An overview

Naming convention

Efficient Data Management in R

Efficient Data Management in R. .Rprofile, renv package and dplyr package.

Text to speech

Text-to-Speech with the googleLanguageR package

Speech to text

https://github.com/ggerganov/whisper.cpp and an R package audio.whisper

Weather data

logR

https://github.com/jangorecki/logR

Progress bar

https://github.com/r-lib/progress#readme

Configurable Progress bars, they may include percentage, elapsed time, and/or the estimated completion time. They work in terminals, in 'Emacs' 'ESS', 'RStudio', 'Windows' 'Rgui' and the 'macOS'.

cron

beepr: Play A Short Sound

https://www.rdocumentation.org/packages/beepr/versions/1.3/topics/beep. Try sound=3 "fanfare", 4 "complete", 5 "treasure", 7 "shotgun", 8 "mario".

utils package

https://www.rdocumentation.org/packages/utils/versions/3.6.2

tools package

Different ways of using R

Extending R by John M. Chambers (2016)

10 things R can do that might surprise you

https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/

R call C/C++

Mainly talks about .C() and .Call().

Note that scalars and arrays must be passed using pointers. So if we want to access a function not exported from a package, we may need to modify the function to make the arguments as pointers.

.Call

Be sure to add the PACKAGE parameter to avoid an error like

cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group, 
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
Error in .Call("standardize", X) : 
  "standardize" not resolved from current namespace (grpreg)

NAMESPACE file & useDynLib

(From Writing R Extensions manual) Loading is most often done automatically based on the useDynLib() declaration in the NAMESPACE file, but may be done explicitly via a call to library.dynam(). This has the form

library.dynam("libname", package, lib.loc) 

library.dynam.unload()

gcc

Coping with varying `gcc` versions and capabilities in R packages

Primitive functions

Primitive Functions List

SEXP

Some examples from packages

  • sva package has one C code function

R call Fortran

Embedding R

An very simple example (do not return from shell) from Writing R Extensions manual

The command-line R front-end, R_HOME/bin/exec/R, is one such example. Its source code is in file <src/main/Rmain.c>.

This example can be run by

R_HOME/bin/R CMD R_HOME/bin/exec/R

Note:

  1. R_HOME/bin/exec/R is the R binary. However, it couldn't be launched directly unless R_HOME and LD_LIBRARY_PATH are set up. Again, this is explained in Writing R Extension manual.
  2. R_HOME/bin/R is a shell-script front-end where users can invoke it. It sets up the environment for the executable. It can be copied to /usr/local/bin/R. When we run R_HOME/bin/R, it actually runs R_HOME/bin/R CMD R_HOME/bin/exec/R (see line 259 of R_HOME/bin/R as in R 3.0.2) so we know the important role of R_HOME/bin/exec/R.

More examples of embedding can be found in tests/Embedding directory. Read <index.html> for more information about these test examples.

An example from Bioconductor workshop

Example: Create embed.c file. Then build the executable. Note that I don't need to create R_HOME variable.

cd 
tar xzvf 
cd R-3.0.1
./configure --enable-R-shlib
make
cd tests/Embedding
make
~/R-3.0.1/bin/R CMD ./Rtest

nano embed.c
# Using a single line will give an error and cannot not show the real problem.
# ../../bin/R CMD gcc -I../../include -L../../lib -lR embed.c
# A better way is to run compile and link separately
gcc -I../../include -c embed.c
gcc -o embed embed.o -L../../lib -lR -lRblas
../../bin/R CMD ./embed

Note that if we want to call the executable file ./embed directly, we shall set up R environment by specifying R_HOME variable and including the directories used in linking R in LD_LIBRARY_PATH. This is based on the inform provided by Writing R Extensions.

export R_HOME=/home/brb/Downloads/R-3.0.2
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
./embed # No need to include R CMD in front.

Question: Create a data frame in C? Answer: Use data.frame() via an eval() call from C. Or see the code is stats/src/model.c, as part of model.frame.default. Or using Rcpp as here.

Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf

Create a Simple Socket Server in R

This example is coming from this paper.

Create an R function

simpleServer <- function(port=6543)
{
  sock <- socketConnection ( port=port , server=TRUE)
  on.exit(close( sock ))
  cat("\nWelcome to R!\nR>" ,file=sock )
  while(( line <- readLines ( sock , n=1)) != "quit")
  {
    cat(paste("socket >" , line , "\n"))
    out<- capture.output (try(eval(parse(text=line ))))
    writeLines ( out , con=sock )
    cat("\nR> " ,file =sock )
  }
}

Then run simpleServer(). Open another terminal and try to communicate with the server

$ telnet localhost 6543
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.

Welcome to R!
R> summary(iris[, 3:5])
  Petal.Length    Petal.Width          Species  
 Min.   :1.000   Min.   :0.100   setosa    :50  
 1st Qu.:1.600   1st Qu.:0.300   versicolor:50  
 Median :4.350   Median :1.300   virginica :50  
 Mean   :3.758   Mean   :1.199                  
 3rd Qu.:5.100   3rd Qu.:1.800                  
 Max.   :6.900   Max.   :2.500                  

R> quit
Connection closed by foreign host.

Rserve

Note the way of launching Rserve is like the way we launch C program when R was embedded in C. See Example from Bioconductor workshop.

See my Rserve page.

outsider

(Commercial) StatconnDcom

R.NET

rJava

Terminal

# jdk 7
sudo apt-get install openjdk-7-*
update-alternatives --config java
# oracle jdk 8
sudo add-apt-repository -y ppa:webupd8team/java
sudo apt-get update
echo debconf shared/accepted-oracle-license-v1-1 select true | sudo debconf-set-selections
echo debconf shared/accepted-oracle-license-v1-1 seen true | sudo debconf-set-selections
sudo apt-get -y install openjdk-8-jdk

and then run the following (thanks to http://stackoverflow.com/questions/12872699/error-unable-to-load-installed-packages-just-now) to fix an error: libjvm.so: cannot open shared object file: No such file or directory.

  • Create the file /etc/ld.so.conf.d/java.conf with the following entries:
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
  • And then run sudo ldconfig

Now go back to R

install.packages("rJava")

Done!

If above does not work, a simple way is by (under Ubuntu) running

sudo apt-get install r-cran-rjava

which will create new package 'default-jre' (under /usr/lib/jvm) and 'default-jre-headless'.

RCaller

RApache

Rscript, arguments and commandArgs()

Passing arguments to an R script from command lines Syntax:

$ Rscript --help
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]

Example:

args = commandArgs(trailingOnly=TRUE)
# test if there is at least one argument: if not, return an error
if (length(args)==0) {
  stop("At least one argument must be supplied (input file).n", call.=FALSE)
} else if (length(args)==1) {
  # default output file
  args[2] = "out.txt"
}
cat("args[1] = ", args[1], "\n")
cat("args[2] = ", args[2], "\n")
Rscript --vanilla sillyScript.R iris.txt out.txt
# args[1] =  iris.txt 
# args[2] =  out.txt

Rscript, #! Shebang and optparse package

littler

Provides hash-bang (#!) capability for R

FAQs:

root@ed5f80320266:/# ls -l /usr/bin/{r,R*}
# R 3.5.2 docker container
-rwxr-xr-x 1 root root 82632 Jan 26 18:26 /usr/bin/r        # binary, can be used for 'shebang' lines, r --help
                                              # Example: r --verbose -e "date()"

-rwxr-xr-x 1 root root  8722 Dec 20 11:35 /usr/bin/R        # text, R --help
                                              # Example: R -q -e "date()"

-rwxr-xr-x 1 root root 14552 Dec 20 11:35 /usr/bin/Rscript  # binary, can be used for 'shebang' lines, Rscript --help
                                              # It won't show the startup message when it is used in the command line.
                                              # Example: Rscript -e "date()"

We can install littler using two ways.

  • install.packages("littler"). This will install the latest version but the binary 'r' program is only available under the package/bin directory (eg ~/R/x86_64-pc-linux-gnu-library/3.4/littler/bin/r). You need to create a soft link in order to access it globally.
  • sudo apt install littler. This will install 'r' globally; however, the installed version may be old.

After the installation, vignette contains several examples. The off-line vignette has a table of contents. Nice! The web version of examples does not have the TOC.

r was not meant to run interactively like R. See man r.

RInside: Embed R in C++

See RInside

(From RInside documentation) The RInside package makes it easier to embed R in your C++ applications. There is no code you would execute directly from the R environment. Rather, you write C++ programs that embed R which is illustrated by some the included examples.

The included examples are armadillo, eigen, mpi, qt, standard, threads and wt.

To run 'make' when we don't have a global R, we should modify the file <Makefile>. Also if we just want to create one executable file, we can do, for example, 'make rinside_sample1'.

To run any executable program, we need to specify LD_LIBRARY_PATH variable, something like

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib 

The real build process looks like (check <Makefile> for completeness)

g++ -I/home/brb/Downloads/R-3.0.2/include \
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
    -I/home/brb/Downloads/R-3.0.2/library/RInside/include -g -O2 -Wall \
    -I/usr/local/include   \
    rinside_sample0.cpp  \
    -L/home/brb/Downloads/R-3.0.2/lib -lR  -lRblas -lRlapack \
    -L/home/brb/Downloads/R-3.0.2/library/Rcpp/lib -lRcpp \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/Rcpp/lib \
    -L/home/brb/Downloads/R-3.0.2/library/RInside/lib -lRInside \
    -Wl,-rpath,/home/brb/Downloads/R-3.0.2/library/RInside/lib \
    -o rinside_sample0

Hello World example of embedding R in C++.

#include <RInside.h>                    // for the embedded R via RInside

int main(int argc, char *argv[]) {

    RInside R(argc, argv);              // create an embedded R instance 

    R["txt"] = "Hello, world!\n";	// assign a char* (string) to 'txt'

    R.parseEvalQ("cat(txt)");           // eval the init string, ignoring any returns

    exit(0);
}

The above can be compared to the Hello world example in Qt.

#include <QApplication.h>
#include <QPushButton.h>

int main( int argc, char **argv )
{
    QApplication app( argc, argv );

    QPushButton hello( "Hello world!", 0 );
    hello.resize( 100, 30 );

    app.setMainWidget( &hello );
    hello.show();

    return app.exec();
}

RFortran

RFortran is an open source project with the following aim:

To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.

It works only on Windows platform with Microsoft Visual Studio installed:(

Call R from other languages

C

Using R from C/C++

Error: “not resolved from current namespace” error, when calling C routines from R

Solution: add getNativeSymbolInfo() around your C/Fortran symbols. Search Google:r dyn.load not resolved from current namespace

JRI

http://www.rforge.net/JRI/

ryp2

http://rpy.sourceforge.net/rpy2.html

Create a standalone Rmath library

R has many math and statistical functions. We can easily use these functions in our C/C++/Fortran. The definite guide of doing this is on Chapter 9 "The standalone Rmath library" of R-admin manual.

Here is my experience based on R 3.0.2 on Windows OS.

Create a static library <libRmath.a> and a dynamic library <Rmath.dll>

Suppose we have downloaded R source code and build R from its source. See Build_R_from_its_source. Then the following 2 lines will generate files <libRmath.a> and <Rmath.dll> under C:\R\R-3.0.2\src\nmath\standalone directory.

cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win

Use Rmath library in our code

set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
# It is not LD_LIBRARY_PATH in above.

# Created <RmathEx1.cpp> from the book "Statistical Computing in C++ and R" web site
# http://math.la.asu.edu/~eubank/CandR/ch4Code.cpp
# It is OK to save the cpp file under any directory.

# Force to link against the static library <libRmath.a>
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
# OR
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe

# Force to link against dynamic library <Rmath.dll>
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe

Test the executable program. Note that the executable program RmathEx1.exe can be transferred to and run in another computer without R installed. Isn't it cool!

c:\R>RmathEx1
Enter a argument for the normal cdf:
1
Enter a argument for the chi-squared cdf:
1
Prob(Z <= 1) = 0.841345
Prob(Chi^2 <= 1)= 0.682689

Below is the cpp program <RmathEx1.cpp>.

//RmathEx1.cpp
#define MATHLIB_STANDALONE 
#include <iostream>
#include "Rmath.h"

using std::cout; using std::cin; using std::endl;

int main()
{
  double x1, x2;
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;

  cout << "Prob(Z <= " << x1 << ") = " << 
    pnorm(x1, 0, 1, 1, 0)  << endl;
  cout << "Prob(Chi^2 <= " << x2 << ")= " << 
    pchisq(x2, 1, 1, 0) << endl;
  return 0;
}

Calling R.dll directly

See Chapter 8.2.2 of R Extensions. This is related to embedding R under Windows. The file <R.dll> on Windows is like <libR.so> on Linux.

Create HTML report

ReportingTools (Jason Hackney) from Bioconductor. See Genome->ReportingTools.

htmlTable package

The htmlTable package is intended for generating tables using HTML formatting. This format is compatible with Markdown when used for HTML-output. The most basic table can easily be created by just passing a matrix or a data.frame to the htmlTable-function.

formattable

htmltab package

This package is NOT used to CREATE html report but EXTRACT html table.

ztable package

Makes zebra-striped tables (tables with alternating row colors) in LaTeX and HTML formats easily from a data.frame, matrix, lm, aov, anova, glm or coxph objects.

Create academic report

reports package in CRAN and in github repository. The youtube video gives an overview of the package.

Create pdf and epub files

# Idea:
#        knitr        pdflatex
#   rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
  • A very simple example <002-minimal.Rnw> from yihui.name works fine on linux.
git clone https://github.com/yihui/knitr-examples.git
  • <knitr-minimal.Rnw>. I have no problem to create pdf file on Windows but still cannot generate pdf on Linux from tex file. Some people suggested to run sudo apt-get install texlive-fonts-recommended to install missing fonts. It works!

To see a real example, check out DESeq2 package (inst/doc subdirectory). In addition to DESeq2, I also need to install DESeq, BiocStyle, airway, vsn, gplots, and pasilla packages from Bioconductor. Note that, it is best to use sudo/admin account to install packages.

Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.

# Idea:
#        knitr        pandoc
#   rmd -------> md ----------> pdf

git clone https://github.com/yihui/knitr-examples.git
cd knitr-examples
R -e "library(knitr); knit('001-minimal.Rmd')"
pandoc 001-minimal.md -o 001-minimal.pdf # require pdflatex to be installed !!

To create an epub file (not success yet on Windows OS, missing figures on Linux OS)

# Idea:
#        knitr        pandoc
#   rnw -------> tex ----------> markdown or epub

library(knitr)
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")

Convert tex to epub

kable() for tables

Create Tables In LaTeX, HTML, Markdown And ReStructuredText

Create Word report

Using the power of Word

How to go from R to nice tables in Microsoft Word

knitr + pandoc

It is better to create rmd file in RStudio. Rstudio provides a template for rmd file and it also provides a quick reference to R markdown language.

# Idea:
#        knitr       pandoc
#   rmd -------> md --------> docx
library(knitr)
knit2html("example.rmd") #Create md and html files

and then

FILE <- "example"
system(paste0("pandoc -o ", FILE, ".docx ", FILE, ".md"))

Note. For example reason, if I play around the above 2 commands for several times, the knit2html() does not work well. However, if I click 'Knit HTML' button on the RStudio, it then works again.

Another way is

library(pander)
name = "demo"
knit(paste0(name, ".Rmd"), encoding = "utf-8")
Pandoc.brew(file = paste0(name, ".md"), output = paste0(-name, "docx"), convert = "docx")

Note that once we have used knitr command to create a md file, we can use pandoc shell command to convert it to different formats:

  • A pdf file: pandoc -s report.md -t latex -o report.pdf
  • A html file: pandoc -s report.md -o report.html (with the -c flag html files can be added easily)
  • Openoffice: pandoc report.md -o report.odt
  • Word docx: pandoc report.md -o report.docx

We can also create the epub file for reading on Kobo ereader. For example, download this file and save it as example.Rmd. I need to remove the line containing the link to http://i.imgur.com/RVNmr.jpg since it creates an error when I run pandoc (not sure if it is the pandoc version I have is too old). Now we just run these 2 lines to get the epub file. Amazing!

knit("example.Rmd")
pandoc("example.md", format="epub")

PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)

> pandoc("Rmd_to_Epub.md", format="epub")
executing pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"
pandoc.exe: .\.\http://i.imgur.com/RVNmr.jpg: openBinaryFile: invalid argument (Invalid argument)
Error in (function (input, format, ext, cfg)  : conversion failed
In addition: Warning message:
running command 'pandoc   -f markdown -t epub -o Rmd_to_Epub.epub "Rmd_to_Epub.utf8md"' had status 1

pander

Try pandoc[1] with a minimal reproducible example, you might give a try to my "pander" package [2] too:

library(pander)
Pandoc.brew(system.file('examples/minimal.brew', package='pander'),
            output = tempfile(), convert = 'docx')

Where the content of the "minimal.brew" file is something you might have got used to with Sweave - although it's using "brew" syntax instead. See the examples of pander [3] for more details. Please note that pandoc should be installed first, which is pretty easy on Windows.

  1. http://johnmacfarlane.net/pandoc/
  2. http://rapporter.github.com/pander/
  3. http://rapporter.github.com/pander/#examples

R2wd

Use R2wd package. However, only 32-bit R is allowed and sometimes it can not produce all 'table's.

> library(R2wd)
> wdGet()
Loading required package: rcom
Loading required package: rscproxy
rcom requires a current version of statconnDCOM installed.
To install statconnDCOM type
     installstatconnDCOM()

This will download and install the current version of statconnDCOM

You will need a working Internet connection
because installation needs to download a file.
Error in if (wdapp[["Documents"]][["Count"]] == 0) wdapp[["Documents"]]$Add() : 
  argument is of length zero 

The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.

Convert from pdf to word

The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert

rtf

Use rtf package for Rich Text Format (RTF) Output.

xtable

Package xtable will produce html output.

print(xtable(X), type="html")

If you save the file and then open it with Word, you will get serviceable results. I've had better luck copying the output from xtable and pasting it into Excel.

officer

  • CRAN. Microsoft Word, Microsoft Powerpoint and HTML documents generation from R.
  • The gist includes a comprehensive example that encompasses various elements such as sections, subsections, and tables. It also incorporates a detailed paragraph, along with visual representations created using base R plots and ggplots.
  • Add a line space
    doc <- body_add_par(doc, "")
    
    # Function to add n line spaces
    body_add_par_n <- function (doc, n) {
      for(i in 1:n){
        doc <- body_add_par(doc, "")
      }
      return(doc)
    }
    body_add_par_n(3)
    
  • Figures from the documentation of officeverse.
  • See Data frame to word table?.
  • See Office page for some code.
  • How to read and create Word Documents in R where we can extracting tables from Word Documents.
    x = read_docx("myfile.docx")
    content <- docx_summary(x) # a vector
    grep("nlme", content$text, ignore.case = T, value = T)
    

Powerpoint

PDF manipulation

staplr

R Graphs Gallery

COM client or server

Client

Server

RDCOMServer

Use R under proxy

http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy

RStudio

  • Github
  • Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
  • Preview

rstudio.cloud

https://rstudio.cloud/

Launch RStudio

Multiple versions of R

Create .Rproj file

If you have an existing package that doesn't have an .Rproj file, you can use devtools::use_rstudio("path/to/package") to add it.

With an RStudio project file, you can

  • Restore .RData into workspace at startup
  • Save workspace to .RData on exit (or save.image("Robj.RData") & load("Robj.RData"))
  • Always save history (even if no saving .RData, savehistory(".Rhistory") & loadhistory(".Rhistory"))
  • etc

package search

https://github.com/RhoInc/CRANsearcher

Git

Visual Studio

R and Python support now built in to Visual Studio 2017

List files using regular expression

  • Extension
list.files(pattern = "\\.txt$")

where the dot (.) is a metacharacter. It is used to refer to any character.

  • Start with
list.files(pattern = "^Something")

Using Sys.glob()"' as

> Sys.glob("~/Downloads/*.txt")
[1] "/home/brb/Downloads/ip.txt"       "/home/brb/Downloads/valgrind.txt"

Hidden tool: rsync in Rtools

c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
sending incremental file list
a.exe

sent 323142 bytes  received 31 bytes  646346.00 bytes/sec
total size is 1198416  speedup is 3.71

c:\Rtools\bin>

Unforunately, if the destination is a network drive, I could get a permission denied (13) error. See also rsync file permissions on windows.

Install rgdal package (geospatial Data) on ubuntu

Terminal

sudo apt-get install libgdal1-dev libproj-dev # https://stackoverflow.com/a/44389304
sudo apt-get install libgdal1i # Ubuntu 16.04 https://stackoverflow.com/a/12143411

R

install.packages("rgdal")

Install sf package

I got the following error even I have installed some libraries.

checking GDAL version >= 2.0.1... no
configure: error: sf is not compatible with GDAL versions below 2.0.1

Then I follow the instruction here

sudo apt remove libgdal-dev
sudo apt remove libproj-dev
sudo apt remove gdal-bin
sudo add-apt-repository ppa:ubuntugis/ubuntugis-stable

sudo apt update
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears 

sudo apt install libgdal-dev # works on ubuntu 20.04 too
                             # no need the previous lines

Database

RSQLite

Creating a new database:

library(DBI)

mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
dbDisconnect(mydb)
unlink("my-db.sqlite")

# temporary database
mydb <- dbConnect(RSQLite::SQLite(), "")
dbDisconnect(mydb)

Loading data:

mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)

dbListTables(mydb)

dbListFields(con, "mtcars")

dbReadTable(con, "mtcars")

Queries:

dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')

dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')

dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))

res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
dbFetch(res)

Batched queries:

dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}

dbClearResult(rs)

Multiple parameterised queries:

rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
nrow(dbFetch(rs))
#> [1] 4
dbClearResult(rs)

Statements:

dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
#> [1] 0
rs <- dbSendStatement(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < :x')
dbBind(rs, param = list(x = 4.5))
dbGetRowsAffected(rs)
#> [1] 4
dbClearResult(rs)

sqldf

Manipulate R data frames using SQL. Depends on RSQLite. A use of gsub, reshape2 and sqldf with healthcare data

RPostgreSQL

RMySQL

MongoDB

odbc

RODBC

DBI

dbplyr

Create a new SQLite database:

surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")

my_db_file <- "portal-database.sqlite"
my_db <- src_sqlite(my_db_file, create = TRUE)

copy_to(my_db, surveys)
copy_to(my_db, plots)
my_db

Connect to a database:

download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")

library(dbplyr)
library(dplyr)
mammals <- src_sqlite("portal_mammals.sqlite")

Querying the database with the SQL syntax:

tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))

Querying the database with the dplyr syntax:

surveys <- tbl(mammals, "surveys")
surveys %>%
    select(year, species_id, plot_id)
head(surveys, n = 10)

show_query(head(surveys, n = 10)) # show which SQL commands are actually sent to the database

Simple database queries:

surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)

Laziness (instruct R to stop being lazy):

data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()

Complex database queries:

plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table

surveys # The plot_id column also features in the surveys table

# Join databases method 1
plots %>%
  filter(plot_id == 1) %>%
  inner_join(surveys) %>%
  collect()

NoSQL

nodbi: the NoSQL Database Connector

Github

R source

https://github.com/wch/r-source/ Daily update, interesting, should be visited every day. Clicking 1000+ commits to look at daily changes.

If we are interested in a certain branch (say 3.2), look for R-3-2-branch.

R packages (only) source (metacran)

Bioconductor packages source

Announcement, https://github.com/Bioconductor-mirror

Send local repository to Github in R by using reports package

http://www.youtube.com/watch?v=WdOI_-aZV0Y

My collection

How to download

Clone ~ Download.

  • Command line
git clone https://gist.github.com/4484270.git

This will create a subdirectory called '4484270' with all cloned files there.

  • Within R
library(devtools)
source_gist("4484270")

or First download the json file from

https://api.github.com/users/MYUSERLOGIN/gists

and then

library(RJSONIO)
x <- fromJSON("~/Downloads/gists.json")
setwd("~/Downloads/")
gist.id <- lapply(x, "[[", "id")
lapply(gist.id, function(x){
  cmd <- paste0("git clone https://gist.github.com/", x, ".git")
  system(cmd)
})

Jekyll

An Easy Start with Jekyll, for R-Bloggers

Connect R with Arduino

Android App

Common plots tips

Create an empty plot

plot.new()

Overlay plots

How to Overlay Plots in R-Quick Guide with Example.

#Step1:-create scatterplot
plot(x1, y1)
#Step 2:-overlay line plot
lines(x2, y2)
#Step3:-overlay scatterplot
points(x2, y2)

Save the par() and restore it

Example 1: Don't use old.par <- par() directly. no.readonly = FALSE by default. * The `no.readonly = TRUE` argument in the par() function in R is used to get the full list of graphical parameters that can be restored.

  • When you call `par()` with no arguments or `par(no.readonly = TRUE)`, it returns an invisible named list of all the graphical parameters. This includes both parameters that can be set and those that are read-only.
  • If we use par(old.par) where old.par <- par(), we will get several warning messages like 'In par(op) : graphical parameter "cin" cannot be set'.
old.par <- par(no.readonly = TRUE); par(mar = c(5, 4, 4, 2) - 2)  # OR in one step
old.par <- par(mar = c(5, 4, 4, 2) - 2)
## do plotting stuff with new settings
par(old.par)

Example 2: Use it inside a function with the on.exit(0 function.

ex <- function() {
   old.par <- par(no.readonly = TRUE) # all par settings which
                                      # could be changed.
   on.exit(par(old.par))
   ## ... do lots of par() settings and plots
   ## ...
   invisible() #-- now,  par(old.par)  will be executed
}

Example 3: It seems par() inside a function will affect the global environment. But if we use dev.off(), it will reset all parameters.

ex <- function() { par(mar=c(5,4,4,1)) }
ex()
par()$mar
ex = function() { png("~/Downloads/test.png"); par(mar=c(5,4,4,1)); dev.off()}
ex()
par()$mar

Grouped boxplots

Weather Time Line

The plot looks similar to a boxplot though it is not. See a screenshot on Android by Sam Ruston.

Horizontal bar plot

library(ggplot2)
dtf <- data.frame(x = c("ETB", "PMA", "PER", "KON", "TRA", 
                        "DDR", "BUM", "MAT", "HED", "EXP"),
                  y = c(.02, .11, -.01, -.03, -.03, .02, .1, -.01, -.02, 0.06))
ggplot(dtf, aes(x, y)) +
  geom_bar(stat = "identity", aes(fill = x), show.legend = FALSE) + 
  coord_flip() + xlab("") + ylab("Fold Change")   

File:Ggplot2bar.svg

Include bar values in a barplot

Use text().

Or use geom_text() if we are using the ggplot2 package. See an example here or this.

For stacked barplot, see this post.

Grouped barplots

library(ggplot2)
# mydata <- data.frame(OUTGRP, INGRP, value)
ggplot(mydata, aes(fill=INGRP, y=value, x=OUTGRP)) + 
       geom_bar(position="dodge", stat="identity")
> 1 - 2*(1-pnorm(1))
[1] 0.6826895
> 1 - 2*(1-pnorm(1.96))
[1] 0.9500042

Unicode symbols

Mind reader game, and Unicode symbols

Math expression

# Expressions
plot(x,y, xlab = expression(hat(x)[t]),
     ylab = expression(phi^{rho + a}),
     main = "Pure Expressions")

# Superscript
plot(1:10, main = expression("My Title"^2)) 
# Subscript
plot(1:10, main = expression("My Title"[2]))  

# Expressions with Spacing
# '~' is to add space and '*' is to squish characters together
plot(1:10, xlab= expression(Delta * 'C'))
plot(x,y, xlab = expression(hat(x)[t] ~ z ~ w),
     ylab = expression(phi^{rho + a} * z * w),
     main = "Pure Expressions with Spacing")

# Expressions with Text
plot(x,y, 
     xlab = expression(paste("Text here ", hat(x), " here ", z^rho, " and here")), 
     ylab = expression(paste("Here is some text of ", phi^{rho})), 
     main = "Expressions with Text")

# Substituting Expressions
plot(x,y, 
     xlab = substitute(paste("Here is ", pi, " = ", p), list(p = py)), 
     ylab = substitute(paste("e is = ", e ), list(e = ee)), 
     main = "Substituted Expressions")

Impose a line to a scatter plot

  • abline + lsfit # least squares
plot(cars)
abline(lsfit(cars[, 1], cars[, 2]))
# OR
abline(lm(cars[,2] ~ cars[,1]))
  • abline + line # robust line fitting
plot(cars)
(z <- line(cars))
abline(coef(z), col = 'green')
  • lines
plot(cars)
fit <- lm(cars[,2] ~ cars[,1])
lines(cars[,1], fitted(fit), col="blue")
lines(stats::lowess(cars), col='red')

How to actually make a quality scatterplot in R: axis(), mtext()

How to actually make a quality scatterplot in R

3D scatterplot

Rotating x axis labels for barplot

https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot

barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)

Set R plots x axis to show at y=0

https://stackoverflow.com/questions/3422203/set-r-plots-x-axis-to-show-at-y-0

plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")

Different colors of axis labels in barplot

See Vary colors of axis labels in R based on another variable

Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.

tN <- table(Ni <- stats::rpois(100, lambda = 5))
r <- barplot(tN, col = rainbow(20))
axis(1, 1, LETTERS[1], col.axis="red", col="red")
axis(1, 2, LETTERS[2], col.axis="blue", col = "blue")

Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.

barplot(tN, col = rainbow(20), axisnames = F)
text(4:6, par("usr")[3]-2 , LETTERS[4:6], col=c("black","red","blue"), xpd=TRUE)

Use text() to draw labels on X/Y-axis including rotation

par(mar = c(5, 6, 4, 5) + 0.1)
plot(..., xaxt = "n") # "n" suppresses plotting of the axis; need mtext() and axis() to supplement
text(x = barCenters, y = par("usr")[3] - 1, srt = 45,
     adj = 1, labels = myData$names, xpd = TRUE)

Vertically stacked plots with the same x axis

https://stackoverflow.com/questions/11794436/stacking-multiple-plots-vertically-with-the-same-x-axis-but-different-y-axes-in

Include labels on the top axis/margin: axis() and mtext()

plot(1:4, rnorm(4), axes = FALSE)
axis(3, at=1:4, labels = LETTERS[1:4], tick = FALSE, line = -0.5) # las, cex.axis
box()
mtext("Groups selected", cex = 0.8, line = 1.5) # default side = 3

See also 15_Questions_All_R_Users_Have_About_Plots

This can be used to annotate each plot with the script name, date, ...

mtext(text=paste("Prepared on", format(Sys.time(), "%d %B %Y at %H:%M")), 
      adj=.99,  # text align to right 
      cex=.75, side=3, las=1, line=2)

ggplot2 uses breaks instead of at parameter. See ggplot2 → Add axis on top or right hand side, ggplot2 → scale_x_continus(name, breaks, labels) and the scale_continuous documentation.

Legend tips

Add legend to a plot in R

Increase/decrease legend font size cex & ggplot2 package case.

plot(rnorm(100))
# op <- par(cex=2)
legend("topleft", legend = 1:4, col=1:4, pch=1, lwd=2, lty = 1, cex =2)
# par(op)

legend inset. Default is 0. % (from 0 to 1) to draw the legend away from x and y axis. The inset argument with negative values moves the legend outside the plot.

legend("bottomright", inset=.05, )

legend without a box

legend(, bty = "n")

Add a legend title

legend(, title = "")

Add a common legend to multiple plots. Use the layout function.

Superimpose a density plot or any curves

Use lines().

Example 1

plot(cars, main = "Stopping Distance versus Speed")
lines(stats::lowess(cars))

plot(density(x), col = "#6F69AC", lwd = 3)
lines(density(y), col = "#95DAC1", lwd = 3)
lines(density(z), col = "#FFEBA1", lwd = 3)

Example 2

require(survival)
n = 10000
beta1 = 2; beta2 = -1
lambdaT = 1 # baseline hazard
lambdaC = 2  # hazard of censoring
set.seed(1234)
x1 = rnorm(n,0)
x2 = rnorm(n,0)
# true event time
T = rweibull(n, shape=1, scale=lambdaT*exp(-beta1*x1-beta2*x2)) 
C <- rweibull(n, shape=1, scale=lambdaC)   
time = pmin(T,C)  
status <- 1*(T <= C) 
status2 <- 1-status
plot(survfit(Surv(time, status2) ~ 1), 
     ylab="Survival probability",
     main = 'Exponential censoring time')
xseq <- seq(.1, max(time), length =100)
func <- function(x) 1-pweibull(x, shape = 1, scale = lambdaC)
lines(xseq, func(xseq), col = 'red') # survival function of Weibull

Example 3. Use ggplot(df, aes(x = x, color = factor(grp))) + geom_density(). Then each density curve will represent data from each "grp".

log scale

If we set y-axis to use log-scale, then what we display is the value log(Y) or log10(Y) though we still label the values using the input. For example, when we plot c(1, 10, 100) using the log scale, it is like we draw log10(c(1, 10, 100)) = c(0,1,2) on the plot but label the axis using the true values c(1, 10, 100).

File:Logscale.png

Custom scales

Using custom scales with the 'scales' package

Time series

Time series stock price plot

library(quantmod)
getSymbols("AAPL")
getSymbols("IBM") # similar to AAPL
getSymbols("CSCO") # much smaller than AAPL, IBM
getSymbols("DJI") # Dow Jones, huge 
chart_Series(Cl(AAPL), TA="add_TA(Cl(IBM), col='blue', on=1); add_TA(Cl(CSCO), col = 'green', on=1)", 
    col='orange', subset = '2017::2017-08')

tail(Cl(DJI))

tidyquant: Getting stock data

The 'largest stock profit or loss' puzzle: efficient computation in R

Timeline plot

Clockify

Clockify

Circular plot

Word cloud

Text mining

World map

Visualising SSH attacks with R (rworldmap and rgeolocate packages)

Diagram/flowchart/Directed acyclic diagrams (DAGs)

DiagrammeR

diagram

Functions for Visualising Simple Graphs (Networks), Plotting Flow Diagrams

DAGitty (browser-based and R package)

dagR

Gmisc

Easiest flowcharts eveR?

Concept Maps

concept-maps where the diagrams are generated from https://app.diagrams.net/.

flow

flow, How To Draw Flow Diagrams In R

Venn Diagram

Venn diagram

hexbin plot

Bump chart/Metro map

https://dominikkoch.github.io/Bump-Chart/

Amazing/special plots

See Amazing plot.

Google Analytics

GAR package

http://www.analyticsforfun.com/2015/10/query-your-google-analytics-data-with.html

Linear Programming

http://www.r-bloggers.com/modeling-and-solving-linear-programming-with-r-free-book/

Linear Algebra

Amazon Alexa

R and Singularity

https://rviews.rstudio.com/2017/03/29/r-and-singularity/

Teach kids about R with Minecraft

http://blog.revolutionanalytics.com/2017/06/teach-kids-about-r-with-minecraft.html

Secure API keys

Securely store API keys in R scripts with the "secret" package

Credentials and secrets

How to manage credentials and secrets safely in R

Hide a password

keyring package

getPass

getPass

Vision and image recognition

Creating a Dataset from an Image

Creating a Dataset from an Image in R Markdown using reticulate

Turn pictures into coloring pages

https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0

Numerical optimization

CRAN Task View: Numerical Mathematics, CRAN Task View: Optimization and Mathematical Programming

Ryacas: R Interface to the 'Yacas' Computer Algebra System

Doing Maths Symbolically: R as a Computer Algebra System (CAS)

Game

Music

  • gm. Require to install MuseScore, an open source and free notation software.

SAS

sasMap Static code analysis for SAS scripts

R packages

R packages

Tricks

Getting help

Better Coder/coding, best practices

E-notation

6.022E23 (or 6.022e23) is equivalent to 6.022×10^23

Getting user's home directory

See What are HOME and working directories?

# Windows
normalizePath("~")   # "C:\\Users\\brb\\Documents"
Sys.getenv("R_USER") # "C:/Users/brb/Documents"
Sys.getenv("HOME")   # "C:/Users/brb/Documents"

# Mac
normalizePath("~")   # [1] "/Users/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")   # "/Users/brb"

# Linux
normalizePath("~")   # [1] "/home/brb"
Sys.getenv("R_USER") # [1] ""
Sys.getenv("HOME")   # [1] "/home/brb"

tempdir()

  • The path is a per-session temporary directory. On parallel use, R processes forked by functions such as mclapply and makeForkCluster in package parallel share a per-session temporary directory.
  • Set temporary folder for R in Rstudio server

Distinguish Windows and Linux/Mac, R.Version()

identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.

get_os <- function(){
  sysinf <- Sys.info()
  if (!is.null(sysinf)){
    os <- sysinf['sysname']
    if (os == 'Darwin')
      os <- "osx"
  } else { ## mystery machine
    os <- .Platform$OS.type
    if (grepl("^darwin", R.version$os))
      os <- "osx"
    if (grepl("linux-gnu", R.version$os))
      os <- "linux"
  }
  tolower(os)
}
names(R.Version())
#  [1] "platform"       "arch"           "os"             "system"        
#  [5] "status"         "major"          "minor"          "year"          
#  [9] "month"          "day"            "svn rev"        "language"      
# [13] "version.string" "nickname" 
getRversion()
# [1] ‘4.3.0’

Rprofile.site, Renviron.site (all platforms) and Rconsole (Windows only)

If we like to install R packages to a personal directory, follow this. Just add the line

R_LIBS_SITE=F:/R/library

to the file R_HOME/etc/x64/Renviron.site. In R, run Sys.getenv("R_LIBS_SITE") or Sys.getenv("R_LIBS_USER") to query the environment variable. See Environment Variables.

What is the best place to save Rconsole on Windows platform

Put/create the file <Rconsole> under C:/Users/USERNAME/Documents folder so no matter how R was upgraded/downgraded, it always find my preference.

My preferred settings:

  • Font: Consolas (it will be shown as "TT Consolas" in Rconsole)
  • Size: 12
  • background: black
  • normaltext: white
  • usertext: GreenYellow or orange (close to RStudio's Cobalt theme) or sienna1 or SpringGreen or tan1 or yellow

and others (default options)

  • pagebg: white
  • pagetext: navy
  • highlight: DarkRed
  • dataeditbg: white
  • dataedittext: navy (View() function)
  • dataedituser: red
  • editorbg: white (edit() function)
  • editortext: black

A copy of the Rconsole is saved in github.

How R starts up

https://rstats.wtf/r-startup.html

startup - Friendly R Startup Configuration

https://github.com/henrikbengtsson/startup

Saving and loading history automatically: .Rprofile & local()

  • savehistory("filename"). It will save everything from the beginning to the command savehistory() to a text file.
  • .Rprofile will automatically be loaded when R has started from that directory
  • Don't do things in your .Rprofile that affect how R code runs, such as loading a package like dplyr or ggplot or setting an option such as stringsAsFactors = FALSE. See Project-oriented workflow.
  • .Rprofile has been created/used by the packrat package to restore a packrat environment. See the packrat/init.R file and R packages → packrat.
  • Customizing Startup from R in Action, Fun with .Rprofile and customizing R startup
    • You can also place a .Rprofile file in any directory that you are going to run R from or in the user home directory.
    • At startup, R will source the Rprofile.site file. It will then look for a .Rprofile file to source in the current working directory. If it doesn't find it, it will look for one in the user's home directory.
    options(continue="  ") # default is "+ "
    options(prompt="R> ", continue=" ")
    options(editor="nano") # default is "vi" on Linux
    # options(htmlhelp=TRUE) 
    
    local({r <- getOption("repos")
          r["CRAN"] <- "https://cran.rstudio.com"
          options(repos=r)})
    
    .First <- function(){
     # library(tidyverse)
     cat("\nWelcome at", date(), "\n")
    }
    
    .Last <- function(){
     cat("\nGoodbye at ", date(), "\n")
    }  
    
  • https://stackoverflow.com/questions/16734937/saving-and-loading-history-automatically
  • The history file will always be read from the $HOME directory and the history file will be overwritten by a new session. These two problems can be solved if we define R_HISTFILE system variable.
  • local() function can be used in .Rprofile file to set up the environment even no new variables will be created (change repository, install packages, load libraries, source R files, run system() function, file/directory I/O, etc)

Linux or Mac

In ~/.profile or ~/.bashrc I put:

export R_HISTFILE=~/.Rhistory

In ~/.Rprofile I put:

if (interactive()) {
  if (.Platform$OS.type == "unix")  .First <- function() try(utils::loadhistory("~/.Rhistory")) 
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}

Windows

If you launch R by clicking its icon from Windows Desktop, the R starts in C:\User\$USER\Documents directory. So we can create a new file .Rprofile in this directory.

if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}

Disable "Save workspace image?" prompt when exit R?

How to disable "Save workspace image?" prompt in R?

R release versions

rversions: Query the main 'R' 'SVN' repository to find the released versions & dates.

getRversion()

getRversion()
[1] ‘4.3.0’

Detect number of running R instances in Windows

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rscript.exe"
INFO: No tasks are running which match the specified criteria.

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rgui.exe"

Image Name                     PID Session Name        Session#    Mem Usage
============================================================================
Rgui.exe                      1096 Console                    1     44,712 K

C:\Program Files\R>tasklist /FI "IMAGENAME eq Rserve.exe"

Image Name                     PID Session Name        Session#    Mem Usage
============================================================================
Rserve.exe                    6108 Console                    1    381,796 K

In R, we can use

> system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE)
[1] ""                                                                            
[2] "Image Name                     PID Session Name        Session#    Mem Usage"
[3] "============================================================================"
[4] "Rgui.exe                      1096 Console                    1     44,804 K"

> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3

Editor

http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs

  • Emacs + ESS. The ESS is useful in the case I want to tidy R code (the tidy_source() function in the formatR package sometimes gives errors; eg when I tested it on an R file like <GetComparisonResults.R> from BRB-ArrayTools v4.4 stable).
    • Edit the file C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el with something like
    (setq-default inferior-R-program-name
                  "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
    

GUI for Data Analysis

Update to Data Science Software Popularity 6/7/2023

BlueSky Statistics

Rcmdr

http://cran.r-project.org/web/packages/Rcmdr/index.html. After loading a dataset, click Statistics -> Fit models. Then select Linear regression, Linear model, GLM, Multinomial logit model, Ordinal regression model, Linear mixed model, and Generalized linear mixed model. However, Rcmdr does not include, e.g. random forest, SVM, glmnet, et al.

Deducer

http://cran.r-project.org/web/packages/Deducer/index.html

jamovi

Scope

See

source()

## foo.R ##
cat(ArrayTools, "\n")
## End of foo.R

# 1. Error
predict <- function() {
  ArrayTools <- "C:/Program Files" # or through load() function 
  source("foo.R")                  # or through a function call; foo()
}
predict()   # Object ArrayTools not found

# 2. OK. Make the variable global
predict <- function() {
  ArrayTools <<- "C:/Program Files'
  source("foo.R")
}
predict()  
ArrayTools

# 3. OK. Create a global variable
ArrayTools <- "C:/Program Files"
predict <- function() {
  source("foo.R")
}
predict()

Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.

Example 1.

> ttt <- data.frame(type=letters[1:5], JpnTest=rep("999", 5), stringsAsFactors = F)
> ttt
  type JpnTest
1    a     999
2    b     999
3    c     999
4    d     999
5    e     999
> jpntest <- function() { ttt$JpnTest[1] ="N5"; print(ttt)}
> jpntest()
  type JpnTest
1    a      N5
2    b     999
3    c     999
4    d     999
5    e     999
> ttt
  type JpnTest
1    a     999
2    b     999
3    c     999
4    d     999
5    e     999

Example 2. How can we set global variables inside a function? The answer is to use the "<<-" operator or assign(, , envir = .GlobalEnv) function.

Other resource: Advanced R by Hadley Wickham.

Example 3. Writing functions in R, keeping scoping in mind

New environment

Run the same function on a bunch of R objects

mye = new.env()
load(<filename>, mye)
for(n in names(mye)) n = as_tibble(mye[[n]])

Just look at the contents of rda file without saving to anywhere (?load)

local({
   load("myfile.rda")
   ls()
})

Or use attach() which is a wrapper of load(). It creates an environment and slots it into the list right after the global environment, then populates it with the objects we're attaching.

attach("all.rda") # safer and will warn about masked objects w/ same name in .GlobalEnv
ls(pos = 2)
##  also typically need to cleanup the search path:
detach("file:all.rda")

If we want to read data from internet, load() works but not attach().

con <- url("http://some.where.net/R/data/example.rda")
## print the value to see what objects were created.
print(load(con))
close(con)
# Github example
# https://stackoverflow.com/a/62954840

source() case.

myEnv <- new.env()    
source("some_other_script.R", local=myEnv)
attach(myEnv, name="sourced_scripts")
search()
ls(2)
ls(myEnv)
with(myEnv, print(x))

str( , max) function

Use max.level parameter to avoid a long display of the structure of a complex R object. Use give.head = FALSE to hide the attributes. See ?str

If we use str() on a function like str(lm), it is equivalent to args(lm)

For a complicated list object, it is useful to use the max.level argument; e.g. str(, max.level = 1)

For a large data frame, we can use the tibble() function; e.g. mydf %>% tibble()

tidy() function

broom::tidy() provides a simplified form of an R object (obtained from running some analysis). See here.

View all objects present in a package, ls()

https://stackoverflow.com/a/30392688. In the case of an R package created by Rcpp.package.skeleton("mypackage"), we will get

> devtools::load_all("mypackage")
> search()
 [1] ".GlobalEnv"        "devtools_shims"    "package:mypackage"
 [4] "package:stats"     "package:graphics"  "package:grDevices"
 [7] "package:utils"     "package:datasets"  "package:methods"
[10] "Autoloads"         "package:base"

> ls("package:mypackage")
[1] "_mypackage_rcpp_hello_world" "evalCpp"                     "library.dynam.unload"       
[4] "rcpp_hello_world"            "system.file"

Note that the first argument of ls() (or detach()) is used to specify the environment. It can be

  • an integer (the position in the ‘search’ list);
  • the character string name of an element in the search list;
  • an explicit ‘environment’ (including using ‘sys.frame’ to access the currently active function calls).

Speedup R code

Profiler

&& vs &

See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.

  • The shorter form performs elementwise comparisons in much the same way as arithmetic operators. The return is a vector.
  • The longer form evaluates left to right examining only the first element of each vector. The return is one value.
  • The longer form evaluates left to right examining only the first element of each vector. Evaluation proceeds only until the result is determined.
  • The idea of the longer form && in R seems to be the same as the && operator in linux shell; see here.
  • Single or double?: AND operator and OR operator in R. The confusion might come from the inconsistency when choosing these operators in different languages. For example, in C, & performs bitwise AND, while && does Boolean logical AND.
  • Think of && as a stricter &
c(T,F,T) & c(T,T,T)
# [1]  TRUE FALSE  TRUE
c(T,F,T) && c(T,T,T)
# [1] TRUE
c(T,F,T) && c(F,T,T)
# [1] FALSE
c(T,F,T) && c(NA,T,T)
# [1] NA
# Assume 'b' is not defined
> if (TRUE && b==3) cat("end")
Error: object 'b' not found
> if (FALSE && b==3) cat("end")
> # No error since the 2nd condition is never evaluated

It's useful in functions(). We don't need nested if statements. In this case if 'arg' is missing, the argument 'L' is not needed so there is not syntax error.

> foo <- function(arg, L) {
   # Suppose 'L' is meaningful only if 'arg' is provided
   # 
   # Evaluate 'L' only if 'arg' is provided
   #
   if (!missing(arg) && L) {
     print("L is true")
   } else {
     print("Either arg is missing or L is FALSE")
   }
 }
> foo()
[1] "arg is missing or L is FALSE"
> foo("a", F)
[1] "arg is missing or L is FALSE"
> foo("a", T)
[1] "L is true"

Other examples: && is more flexible than &.

nspot <- ifelse(missing(rvm) || !rvm, nrow(exprTrain), sum(filter))

if (!is.null(exprTest) && any(is.na(exprTest))) { ... }

for-loop, control flow

Vectorization

sapply vs vectorization

Speed test: sapply vs vectorization

lapply vs for loop

split() and sapply()

split() can be used to split a vector, columns or rows. See How to split a data frame?

  • Split divides the data in the vector or data frame x into the groups defined by f. The syntax is
    split(x, f, drop = FALSE, …)
    
  • split() + cut(). How to Split Data into Equal Sized Groups in R: A Comprehensive Guide for Beginners
  • Split a vector into chunks. split() returns a vector/indices and the indices can be used in lapply() to subset the data. Useful for the split() + lapply() + do.call() or split() + sapply() operations.
    d <- 1:10
    chunksize <- 4
    ceiling(1:10/4)
    # [1] 1 1 1 1 2 2 2 2 3 3
    split(d, ceiling(seq_along(d)/chunksize))
    # $`1`
    # [1] 1 2 3 4
    #
    # $`2`
    # [1] 5 6 7 8
    #
    # $`3`
    # [1]  9 10
    do.call(c, lapply(split(d, ceiling(seq_along(d)/4)), function(x) sum(x)) ) 
    #  1  2  3 
    # 10 26 19
    
    # bigmemory vignette
    planeindices <- split(1:nrow(x), x[,'TailNum'])
    planeStart <- sapply(planeindices,
                         function(i) birthmonth(x[i, c('Year','Month'),
                                                drop=FALSE]))
    
  • Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.
    split(mtcars,mtcars$cyl)
    
    split(data.frame(matrix(1:20, nr=10) ), ceiling(1:10/chunksize)) # data.frame/tibble works
    split.data.frame(matrix(1:20, nr=10), ceiling(1:10/chunksize))   # split.data.frame() works for matrices
    
  • Split columns of a data frame/matrix.
    ma <- cbind(x = 1:10, y = (-4:5)^2, z = 11:20)
    split(ma, cbind(rep(1,10), rep(2, 10), rep(1,10))) # not an interesting example
    # $`1`
    #  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20
    #
    # $`2`
    #  [1] 16  9  4  1  0  1  4  9 16 25
    
  • split() + sapply() to merge columns. See below Mean of duplicated columns for more detail.
  • split() + sapply() to split a vector. See nsFilter() function which can remove duplicated probesets/rows using unique Entrez Gene IDs (genefilter package). The source code of nsFilter() and findLargest().
    tSsp = split.default(testStat, lls) 
    # testStat is a vector of numerics including probeset IDs as names
    # lls is a vector of entrez IDs (same length as testStat)
    # tSSp is a list of the same length as unique elements of lls.
    
    sapply(tSsp, function(x) names(which.max(x))) 
    # return a vector of probset IDs of length of unique entrez IDs
    

strsplit and sapply

> namedf <- c("John ABC", "Mary CDE", "Kat FGH")
> strsplit(namedf, " ")
1
[1] "John" "ABC" 

2
[1] "Mary" "CDE" 

3
[1] "Kat" "FGH"

> sapply(strsplit(namedf, " "), "[", 1)
[1] "John" "Mary" "Kat" 
> sapply(strsplit(namedf, " "), "[", 2)
[1] "ABC" "CDE" "FGH"

Mean of duplicated columns: rowMeans; compute Means by each row

  • Reduce columns of a matrix by a function in R. To use rowMedians() instead of rowMeans(), we need to install matrixStats from CRAN.
    set.seed(1)
    x <- matrix(1:60, nr=10); x[1, 2:3] <- NA
    colnames(x) <- c("b", "b", "b", "c", "a", "a"); x
    res <- sapply(split(1:ncol(x), colnames(x)), 
                  function(i) rowMeans(x[, i, drop=F], na.rm = TRUE))
    res  # notice the sorting of columns
           a  b  c
     [1,] 46  1 31
     [2,] 47 12 32
     [3,] 48 13 33
     [4,] 49 14 34
     [5,] 50 15 35
     [6,] 51 16 36
     [7,] 52 17 37
     [8,] 53 18 38
     [9,] 54 19 39
    [10,] 55 20 40
    
    # vapply() is safter than sapply(). 
    # The 3rd arg in vapply() is a template of the return value.
    res2 <- vapply(split(1:ncol(x), colnames(x)), 
                   function(i) rowMeans(x[, i, drop=F], na.rm = TRUE),
                   rep(0, nrow(x)))
  • colSums, rowSums, colMeans, rowMeans (no group variable). These functions are equivalent to use of ‘apply’ with ‘FUN = mean’ or ‘FUN = sum’ with appropriate margins, but are a lot faster.
    rowMeans(x, na.rm=T)
    # [1] 31 27 28 29 30 31 32 33 34 35
    
    apply(x, 1, mean, na.rm=T)
    # [1] 31 27 28 29 30 31 32 33 34 35
    
  • matrixStats: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
  • From for() loops to the split-apply-combine paradigm for column-wise tasks: the transition for a dinosaur

Mean of duplicated rows: colMeans and rowsum

  • colMeans(x, na.rm = FALSE, dims = 1), take mean per columns & sum over rows. It returns a vector. Other similar idea functions include colSums, rowSums, rowMeans.
    x <- matrix(1:60, nr=10); x[1, 2:3] <- NA; x
    rownames(x) <- c(rep("b", 2), rep("c", 3), rep("d", 4), "a") # move 'a' to the last
    res <- sapply(split(1:nrow(x), rownames(x)), 
                  function(i) colMeans(x[i, , drop=F], na.rm = TRUE))
    res <- t(res) # transpose is needed since sapply() will form the resulting matrix by columns
    res  # still a matrix, rows are ordered
    #   [,1] [,2] [,3] [,4] [,5] [,6]
    # a 10.0 20.0 30.0 40.0 50.0 60.0
    # b  1.5 12.0 22.0 31.5 41.5 51.5
    # c  4.0 14.0 24.0 34.0 44.0 54.0
    # d  7.5 17.5 27.5 37.5 47.5 57.5
    table(rownames(x))
    # a b c d
    # 1 2 3 4
    
    aggregate(x, list(rownames(x)), FUN=mean, na.rm = T) # EASY, but it becomes a data frame, rows are ordered
    #   Group.1   V1   V2   V3   V4   V5   V6
    # 1       a 10.0 20.0 30.0 40.0 50.0 60.0
    # 2       b  1.5 12.0 22.0 31.5 41.5 51.5
    # 3       c  4.0 14.0 24.0 34.0 44.0 54.0
    # 4       d  7.5 17.5 27.5 37.5 47.5 57.5
    
  • Reduce multiple probes by the maximally expressed probe (set) measured by average intensity across arrays
  • rowsum(x, group, reorder = TRUE, …). Sum over rows. It returns a matrix. This is very special. It's not the same as rowSums. There is no "colsum" function. It has the speed advantage over sapply+colSums OR aggregate.
    group <- rownames(x)
    rowsum(x, group, na.rm=T)/as.vector(table(group))
    #   [,1] [,2] [,3] [,4] [,5] [,6]
    # a 10.0 20.0 30.0 40.0 50.0 60.0
    # b  1.5  6.0 11.0 31.5 41.5 51.5
    # c  4.0 14.0 24.0 34.0 44.0 54.0
    # d  7.5 17.5 27.5 37.5 47.5 57.5
    
  • by() function. Calculating change from baseline in R
  • See aggregate Function in R- A powerful tool for data frames & summarize in r, Data Summarization In R
  • aggregate() function. Too slow! http://slowkow.com/2015/01/28/data-table-aggregate/. Don't use aggregate post.
    > attach(mtcars)
    dim(mtcars)
    [1] 32 11
    > head(mtcars)
                       mpg cyl disp  hp drat    wt  qsec vs am gear carb
    Mazda RX4         21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
    Mazda RX4 Wag     21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
    Datsun 710        22.8   4  108  93 3.85 2.320 18.61  1  1    4    1
    Hornet 4 Drive    21.4   6  258 110 3.08 3.215 19.44  1  0    3    1
    Hornet Sportabout 18.7   8  360 175 3.15 3.440 17.02  0  0    3    2
    Valiant           18.1   6  225 105 2.76 3.460 20.22  1  0    3    1
    > with(mtcars, table(cyl, vs))
       vs
    cyl  0  1
      4  1 10
      6  3  4
      8 14  0
    > aggdata <-aggregate(mtcars, by=list(cyl,vs),  FUN=mean, na.rm=TRUE)
    > print(aggdata)
      Group.1 Group.2      mpg cyl   disp       hp     drat       wt     qsec vs
    1       4       0 26.00000   4 120.30  91.0000 4.430000 2.140000 16.70000  0
    2       6       0 20.56667   6 155.00 131.6667 3.806667 2.755000 16.32667  0
    3       8       0 15.10000   8 353.10 209.2143 3.229286 3.999214 16.77214  0
    4       4       1 26.73000   4 103.62  81.8000 4.035000 2.300300 19.38100  1
    5       6       1 19.12500   6 204.55 115.2500 3.420000 3.388750 19.21500  1
             am     gear     carb
    1 1.0000000 5.000000 2.000000
    2 1.0000000 4.333333 4.666667
    3 0.1428571 3.285714 3.500000
    4 0.7000000 4.000000 1.500000
    5 0.0000000 3.500000 2.500000
    > detach(mtcars)
    
    # Another example: select rows with a minimum value from a certain column (yval in this case)
    > mydf <- read.table(header=T, text='
     id xval yval
     A 1  1
     A -2  2
     B 3  3
     B 4  4
     C 5  5
     ')
    > x = mydf$xval
    > y = mydf$yval
    > aggregate(mydf[, c(2,3)], by=list(id=mydf$id), FUN=function(x) x[which.min(y)])
      id xval yval
    1  A    1    1
    2  B    3    3
    3  C    5    5
    

Mean by Group

Mean by Group in R (2 Examples) | dplyr Package vs. Base R

aggregate(x = iris$Sepal.Length,                # Specify data column
          by = list(iris$Species),              # Specify group indicator
          FUN = mean)                           # Specify function (i.e. mean)
library(dplyr)
iris %>%                                        # Specify data frame
  group_by(Species) %>%                         # Specify group indicator
  summarise_at(vars(Sepal.Length),              # Specify column
               list(name = mean))               # Specify function
  • ave(x, ..., FUN),
  • aggregate(x, by, FUN),
  • by(x, INDICES, FUN): return is a list
  • tapply(): return results as a matrix or array. Useful for ragged array.

Apply family

Vectorize, aggregate, apply, by, eapply, lapply, mapply, rapply, replicate, scale, sapply, split, tapply, and vapply.

The following list gives a hierarchical relationship among these functions.

  • apply(X, MARGIN, FUN, ...) – Apply a Functions Over Array Margins
  • lapply(X, FUN, ...) – Apply a Function over a List (including a data frame) or Vector X.
    • sapply(X, FUN, ..., simplify = TRUE, USE.NAMES = TRUE) – Apply a Function over a List or Vector
      • replicate(n, expr, simplify = "array")
    • mapply(FUN, ..., MoreArgs = NULL, SIMPLIFY = TRUE, USE.NAMES = TRUE) – Multivariate version of sapply
      • Vectorize(FUN, vectorize.args = arg.names, SIMPLIFY = TRUE, USE.NAMES = TRUE) - Vectorize a Scalar Function
      • Map(FUN, ...) A wrapper to mapply with SIMPLIFY = FALSE, so it is guaranteed to return a list.
    • vapply(X, FUN, FUN.VALUE, ..., USE.NAMES = TRUE) – similar to sapply, but has a pre-specified type of return value
    • rapply(object, f, classes = "ANY", deflt = NULL, how = c("unlist", "replace", "list"), ...) – A recursive version of lapply
  • tapply(V, INDEX, FUN = NULL, ..., default = NA, simplify = TRUE) – Apply a Function Over a "Ragged" Array. V is typically a vector where split() will be applied. INDEX is a list of one or more factors.
    • aggregate(D, by, FUN, ..., simplify = TRUE, drop = TRUE) - Apply a function to each columns of subset data frame split by factors. FUN (such as mean(), weighted.mean(), sum()) is a simple function applied to a vector. D is typically a data frame. This is used to summarize data.
    • by(D, INDICES, FUN, ..., simplify = TRUE) - Apply a Function to each subset data frame split by factors. FUN (such as summary(), lm()) is applied to a data frame. D is typically a data frame.
  • eapply(env, FUN, ..., all.names = FALSE, USE.NAMES = TRUE) – Apply a Function over values in an environment

Difference between apply vs sapply vs lapply vs tapply?

  • apply - When you want to apply a function to the rows or columns or both of a matrix and output is a one-dimensional if only row or column is selected else it is a 2D-matrix
  • lapply - When you want to apply a function to each element of a list in turn and get a list back.
  • sapply - When you want to apply a function to each element of a list in turn, but you want a vector back, rather than a list.
  • tapply - When you want to apply a function to subsets of a vector and the subsets are defined by some other vector, usually a factor.

Some short examples:

Apply vs for loop

Note that, apply's performance is not always better than a for loop. See

Progress bar

What is the cost of a progress bar in R?

The package 'pbapply' creates a text-mode progress bar - it works on any platforms. On Windows platform, check out this post. It uses winProgressBar() and setWinProgressBar() functions.

e-Rum 2020 Slides on Progressr by Henrik Bengtsson. progressr 0.8.0: RStudio's progress bar, Shiny progress updates, and absolute progress, progressr 0.10.1: Plyr Now Supports Progress Updates also in Parallel

simplify option in sapply()

library(KEGGREST)

names1 <- keggGet(c("hsa05340", "hsa05410"))
names2 <- sapply(names1, function(x) x$GENE)
length(names2)  # same if we use lapply() above
# [1] 2

names3 <- keggGet(c("hsa05340"))
names4 <- sapply(names3, function(x) x$GENE)
length(names4)  # may or may not be what we expect
# [1] 76
names4 <- sapply(names3, function(x) x$GENE, simplify = FALSE)
length(names4)  # same if we use lapply() w/o simplify 
# [1] 1

lapply and its friends Map(), Reduce(), Filter() from the base package for manipulating lists

  • mapply() documentation. Use mapply() to merge lists.
    mapply(rep, 1:4, 4:1)
    mapply(rep, times = 1:4, x = 4:1)
    mapply(function(x, y) seq_len(x) + y,
           c(a =  1, b = 2, c = 3),  # names from first
           c(A = 10, B = 0, C = -10))
    mapply(c, firstList, secondList, SIMPLIFY=FALSE)
    
  • Finding the Expected value of the maximum of two Bivariate Normal variables with simulation sapply + mapply.
    z <- mapply(function(u, v) { max(u, v) }, 
                u = x[, 1], v = x[, 2])
    
  • Map() and Reduce() in functional programming
  • Map(), Reduce(), and Filter() from Advanced R by Hadley
    • If you have two or more lists (or data frames) that you need to process in parallel, use Map(). One good example is to compute the weighted.mean() function that requires two input objects. Map() is similar to mapply() function and is more concise than lapply(). Advanced R has a comment that Map() is better than mapply().
      # Syntax: Map(f, ...)
      
      xs <- replicate(5, runif(10), simplify = FALSE)
      ws <- replicate(5, rpois(10, 5) + 1, simplify = FALSE)
      Map(weighted.mean, xs, ws)
      
      # instead of a more clumsy way
      lapply(seq_along(xs), function(i) {
        weighted.mean(xsi, wsi)
      })
      
    • Reduce() reduces a vector, x, to a single value by recursively calling a function, f, two arguments at a time. A good example of using Reduce() function is to read a list of matrix files and merge them. See How to combine multiple matrix frames into one using R?
      # Syntax: Reduce(f, x, ...)
      
      > m1 <- data.frame(id=letters[1:4], val=1:4)
      > m2 <- data.frame(id=letters[2:6], val=2:6)
      > merge(m1, m2, "id", all = T)
        id val.x val.y
      1  a     1    NA
      2  b     2     2
      3  c     3     3
      4  d     4     4
      5  e    NA     5
      6  f    NA     6
      > m <- list(m1, m2)
      > Reduce(function(x,y) merge(x,y, "id",all=T), m)
        id val.x val.y
      1  a     1    NA
      2  b     2     2
      3  c     3     3
      4  d     4     4
      5  e    NA     5
      6  f    NA     6
      

sapply & vapply

See parallel::parSapply() for a parallel version of sapply(1:n, function(x)). We can this technique to speed up this example.

rapply - recursive version of lapply

replicate

https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r

> replicate(5, rnorm(3))
           [,1]       [,2]       [,3]      [,4]        [,5]
[1,]  0.2509130 -0.3526600 -0.3170790  1.064816 -0.53708856
[2,]  0.5222548  1.5343319  0.6120194 -1.811913 -1.09352459
[3,] -1.9905533 -0.8902026 -0.5489822  1.308273  0.08773477

See parSapply() for a parallel version of replicate().

Vectorize

> rep(1:4, 4:1)
 [1] 1 1 1 1 2 2 2 3 3 4
> vrep <- Vectorize(rep.int)
> vrep(1:4, 4:1)
1
[1] 1 1 1 1

2
[1] 2 2 2

3
[1] 3 3

4
[1] 4
> rweibull(1, 1, c(1, 2)) # no error but not sure what it gives?
[1] 2.17123
> Vectorize("rweibull")(n=1, shape = 1, scale = c(1, 2)) 
[1] 1.6491761 0.9610109
myfunc <- function(a, b) a*b
myfunc(1, 2) # 2
myfunc(3, 5) # 15
myfunc(c(1,3), c(2,5)) # 2 15
Vectorize(myfunc)(c(1,3), c(2,5)) # 2 15

myfunc2 <- function(a, b) if (length(a) == 1) a * b else NA
myfunc2(1, 2) # 2 
myfunc2(3, 5) # 15
myfunc2(c(1,3), c(2,5)) # NA
Vectorize(myfunc2)(c(1, 3), c(2, 5)) # 2 15
Vectorize(myfunc2)(c(1, 3, 6), c(2, 5)) # 2 15 12
                                        # parameter will be re-used

plyr and dplyr packages

Practical Data Science for Stats - a PeerJ Collection

The Split-Apply-Combine Strategy for Data Analysis (plyr package) in J. Stat Software.

A quick introduction to plyr with a summary of apply functions in R and compare them with functions in plyr package.

  1. plyr has a common syntax -- easier to remember
  2. plyr requires less code since it takes care of the input and output format
  3. plyr can easily be run in parallel -- faster

Tutorials

Examples of using dplyr:

tibble

Tidy DataFrames but not Tibbles

Tibble objects

  • it does not have row names (cf data frame),
  • it never changes the type of the inputs (e.g. it never converts strings to factors!),
  • it never changes the names of variables

To show all rows or columns of a tibble object,

print(tbObj, n= Inf)

print(tbObj, width = Inf)

If we try to do a match on some column of a tibble object, we will get zero matches. The issue is we cannot use an index to get a tibble column.

Subsetting: to extract a column from a tibble object, use [[ or $ or dplyr::pull(). Select Data Frame Columns in R.

TibbleObject$VarName
# OR
TibbleObject"VarName"
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore

# For multiple columns, use select()
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
# OR
dplyr::select(TibbleObject, 2:5) # 

Convert a data frame to a tibble See Tibble Data Format in R: Best and Modern Way to Work with Your Data

my_data <- as_tibble(iris)
class(my_data)

llply()

llply is equivalent to lapply except that it will preserve labels and can display a progress bar. This is handy if we want to do a crazy thing.

LLID2GOIDs <- lapply(rLLID, function(x) get("org.Hs.egGO")[[x]])

where rLLID is a list of entrez ID. For example,

get("org.Hs.egGO")[["6772"]]

returns a list of 49 GOs.

ddply()

http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html

ldply()

An R Script to Automatically download PubMed Citation Counts By Year of Publication

Performance/speed comparison

Performance comparison of converting list to data.frame with R language

Using R's set.seed() to set seeds for use in C/C++ (including Rcpp)

http://rorynolan.rbind.io/2018/09/30/rcsetseed/

get_seed()

See the same blog

get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}

Note: .Machine$integer.max = 2147483647 = 2^31 - 1.

Random seeds

By default, R uses the exact time in milliseconds of the computer's clock when R starts up to generate a seed. See ?Random.

set.seed(as.numeric(Sys.time()))

set.seed(as.numeric(Sys.Date()))  # same seed for each day

.Machine and the largest integer, double

See ?.Machine.

                          Linux/Mac  32-bit Windows 64-bit Windows
double.eps              2.220446e-16   2.220446e-16   2.220446e-16
double.neg.eps          1.110223e-16   1.110223e-16   1.110223e-16
double.xmin            2.225074e-308  2.225074e-308  2.225074e-308
double.xmax            1.797693e+308  1.797693e+308  1.797693e+308
double.base             2.000000e+00   2.000000e+00   2.000000e+00
double.digits           5.300000e+01   5.300000e+01   5.300000e+01
double.rounding         5.000000e+00   5.000000e+00   5.000000e+00
double.guard            0.000000e+00   0.000000e+00   0.000000e+00
double.ulp.digits      -5.200000e+01  -5.200000e+01  -5.200000e+01
double.neg.ulp.digits  -5.300000e+01  -5.300000e+01  -5.300000e+01
double.exponent         1.100000e+01   1.100000e+01   1.100000e+01
double.min.exp         -1.022000e+03  -1.022000e+03  -1.022000e+03
double.max.exp          1.024000e+03   1.024000e+03   1.024000e+03
integer.max             2.147484e+09   2.147484e+09   2.147484e+09
sizeof.long             8.000000e+00   4.000000e+00   4.000000e+00
sizeof.longlong         8.000000e+00   8.000000e+00   8.000000e+00
sizeof.longdouble       1.600000e+01   1.200000e+01   1.600000e+01
sizeof.pointer          8.000000e+00   4.000000e+00   8.000000e+00

NA when overflow

tmp <- 156287L
tmp*tmp
# [1] NA
# Warning message:
# In tmp * tmp : NAs produced by integer overflow
.Machine$integer.max
# [1] 2147483647

How to select a seed for simulation or randomization

set.seed() allow alphanumeric seeds

https://stackoverflow.com/a/10913336

set.seed(), for loop and saving random seeds

  • Detect When the Random Number Generator Was Used
    if (interactive()) {
      invisible(addTaskCallback(local({
        last <- .GlobalEnv$.Random.seed
        
        function(...) {
          curr <- .GlobalEnv$.Random.seed
          if (!identical(curr, last)) {
            msg <- "NOTE: .Random.seed changed"
            if (requireNamespace("crayon", quietly=TRUE)) msg <- crayon::blurred(msg)
            message(msg)
            last <<- curr
          }
          TRUE
        }
      }), name = "RNG tracker"))
    }
    
  • http://r.789695.n4.nabble.com/set-seed-and-for-loop-td3585857.html. This question is legitimate when we want to debug on a certain iteration.
    set.seed(1001) 
    data <- vector("list", 30) 
    seeds <- vector("list", 30) 
    for(i in 1:30) { 
      seeds[[i]] <- .Random.seed 
      data[[i]] <- runif(5) 
    } 
     
    # If we save and load .Random.seed from a file using scan(), make
    # sure to convert its type from doubles to integers.
    # Otherwise, .Random.seed will complain!
    
    .Random.seed <- seeds[[23]]  # restore 
    data.23 <- runif(5) 
    data.23 
    data[[23]] 
    
  • impute.knn
  • Duncan Murdoch: This works in this example, but wouldn't work with all RNGs, because some of them save state outside of .Random.seed. See ?.Random.seed for details.
  • Uwe Ligges's comment: set.seed() actually generates a seed. See ?set.seed that points us to .Random.seed (and relevant references!) which contains the actual current seed.
  • Petr Savicky's comment is also useful in the situation when it is not difficult to re-generate the data.
  • Local randomness in R.

sample()

sample() inaccurate on very large populations, fixed in R 3.6.0

# R 3.5.3
set.seed(123)
m <- (2/5)*2^32
m > 2^31
# [1] FALSE
log10(m)
# [1] 9.23502
x <- sample(m, 1000000, replace = TRUE)
table(x %% 2)
#      0      1 
# 400070 599930 
# R 3.5.3
# docker run --net=host -it --rm r-base:3.5.3
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5

# R 3.6.0
# docker run --net=host -it --rm r-base:3.6.0
> set.seed(1234)
> sample(5)
[1] 4 5 2 3 1
> RNGkind(sample.kind = "Rounding")
Warning message:
In RNGkind(sample.kind = "Rounding") : non-uniform 'Rounding' sampler used
> set.seed(1234)
> sample(5)
[1] 1 3 2 4 5

Getting different results with set.seed() in RStudio

Getting different results with set.seed(). It's possible that you're loading an R package that is changing the requested random number generator; RNGkind().

dplyr::sample_n()

The function has a parameter weight. For example if we have some download statistics for each day and we want to do sampling based on their download numbers, we can use this function.

Regular Expression

See here.

Read rrd file

on.exit()

Examples of using on.exit(). In all these examples, add = TRUE is used in the on.exit() call to ensure that each exit action is added to the list of actions to be performed when the function exits, rather than replacing the previous actions.

  • Database connections
    library(RSQLite)
    sqlite_get_query <- function(db, sql) {
      conn <- dbConnect(RSQLite::SQLite(), db)
      on.exit(dbDisconnect(conn), add = TRUE)
      dbGetQuery(conn, sql)
    }
    
  • File connections
    read_chars <- function(file_name) {
      conn <- file(file_name, "r")
      on.exit(close(conn), add = TRUE)
      readChar(conn, file.info(file_name)$size)
    }
    
  • Temporary files
    history_lines <- function() {
      f <- tempfile()
      on.exit(unlink(f), add = TRUE)
      savehistory(f)
      readLines(f, encoding = "UTF-8")
    }
    
  • Printing messages
    myfun = function(x) {
      on.exit(print("first"))
      on.exit(print("second"), add = TRUE)
      return(x)
    }
    

file, connection

  • cat() and scan() (read data into a vector or list from the console or file)
  • read() and write()
  • read.table() and write.table()
out = file('tmp.txt', 'w')
writeLines("abcd", out)
writeLines("eeeeee", out)
close(out)
readLines('tmp.txt')
unlink('tmp.txt')
args(writeLines)
# function (text, con = stdout(), sep = "\n", useBytes = FALSE)

foo <- function() {
  con <- file()
  ...
  on.exit(close(con))
  ...
}

Error in close.connection(f) : invalid connection. If we want to use close(con), we have to specify how to open the connection; such as

con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)

withr package

https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by languageserver.

Clipboard (?connections), textConnection(), pipe()

  • On Windows, we can use readClipboard() and writeClipboard().
    source("clipboard")
    read.table("clipboard")
    
  • Clipboard -> R. Reading/writing clipboard on macOS. Use textConnection() function:
    x <- read.delim(textConnection("<USE_KEYBOARD_TO_PASTE_FROM_CLIPBOARD>"))
    # Or on Mac
    x <- read.delim(pipe("pbpaste"))
    # safely ignore the warning: incomplete final line found by readTableHeader on 'pbpaste'
    

    An example is to copy data from this post. In this case we need to use read.table() instead of read.delim().

  • R -> clipboard on Mac. Note: pbcopy and pbpaste are macOS terminal commands. See pbcopy & pbpaste: Manipulating the Clipboard from the Command Line.
    • pbcopy: takes standard input and places it in the clipboard buffer
    • pbpaste: takes data from the clipboard buffer and writes it to the standard output
    clip <- pipe("pbcopy", "w")
    write.table(apply(x, 1, mean), file = clip, row.names=F, col.names=F)
    # write.table(data.frame(Var1, Var2), file = clip, row.names=F, quote=F, sep="\t")
    close(clip)
    
  • Clipboard -> Excel.
    • Method 1: Paste icon -> Text import wizard -> Delimit (Tab, uncheck Space) or Fixed width depending on the situation -> Finish.
    • Method 2: Ctrl+v first. Then choose Data -> Text to Columns. Fixed width -> Next -> Next -> Finish.
  • On Linux, we need to install "xclip". See R Copy from Clipboard in Ubuntu Linux. It seems to work.
    # sudo apt-get install xclip
    read.table(pipe("xclip -selection clipboard -o",open="r"))
    

clipr

clipr: Read and Write from the System Clipboard

read/manipulate binary data

  • x <- readBin(fn, raw(), file.info(fn)$size)
  • rawToChar(x[1:16])
  • See Biostrings C API

String Manipulation

format(): padding with zero

ngenes <- 10
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"

noquote()

noqute Print character strings without quotes.

stringr package

glue package

  • glue. Useful in a loop and some function like ggtitle() or ggsave(). Inside the curly braces {R-Expression}, the expression is evaluated.
    library(glue)
    name <- "John"
    age <- 30
    glue("My name is {name} and I am {age} years old.")
    # My name is John and I am 30 years old.
    
    price <- 9.99
    quantity <- 3
    total <- glue("The total cost is {round(price * quantity, 2)}.")
    # Inside the curly braces {}, the expression round(price * quantity, 2) is evaluated.
    print(total)
    # The total cost is 29.97.

    The syntax of glue() in R is quite similar to Python's print() function when using formatted strings. In Python, you typically use f-strings to embed variables inside strings.

    name = "John"
    age = 30
    print(f"My name is {name} and I am {age} years old.")
    # My name is John and I am 30 years old.
    
    price = 9.99
    quantity = 3
    total = f"The total cost is {price * quantity:.2f}."
    print(total)
    # The total cost is 29.97.
  • String interpolation

Raw data type

Fun with strings, Cyrillic alphabets

a1 <- "А"
a2 <- "A"
a1 == a2
# [1] FALSE
charToRaw("А")
# [1] d0 90
charToRaw("A")
# [1] 41

number of characters limit

It's a limit on a (single) input line in the REPL

Comparing strings to numeric

">" coerces the number to a string before comparing. "10" < 2 # TRUE

HTTPs connection

HTTPS connection becomes default in R 3.2.2. See

R 3.3.2 patched The internal methods of ‘download.file()’ and ‘url()’ now report if they are unable to follow the redirection of a ‘http://’ URL to a ‘https://’ URL (rather than failing silently)

setInternet2

There was a bug in ftp downloading in R 3.2.2 (r69053) Windows though it is fixed now in R 3.2 patch.

Read the discussion reported on 8/8/2015. The error only happened on ftp not http connection. The final solution is explained in this post. The following demonstrated the original problem.

url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)

It seems the bug was fixed in R 3.2-branch. See 8/16/2015 patch r69089 where a new argument INTERNET_FLAG_PASSIVE was added to InternetOpenUrl() function of wininet library. This article and this post explain differences of active and passive FTP.

The following R command will show the exact svn revision for the R you are currently using.

R.Version()$"svn rev"

If setInternet2(T), then https protocol is supported in download.file().

When setInternet(T) is enabled by default, download.file() does not work for ftp protocol (this is used in getGEO() function of the GEOquery package). If I use setInternet(F), download.file() works again for ftp protocol.

The setInternet2() function is defined in R> src> library> utils > R > windows > sysutils.R.

R up to 3.2.2

setInternet2 <- function(use = TRUE) .Internal(useInternet2(use))

See also

  • <src/include/Internal.h> (declare do_setInternet2()),
  • <src/main/names.c> (show do_setInternet2() in C)
  • <src/main/internet.c> (define do_setInternet2() in C).

Note that: setInternet2(T) becomes default in R 3.2.2. To revert to the previous default use setInternet2(FALSE). See the <doc/NEWS.pdf> file. If we use setInternet2(F), then it solves the bug of getGEO() error. But it disables the https file download using the download.file() function. In R < 3.2.2, it is also possible to download from https by setIneternet2(T).

R 3.3.0

setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}

Note that setInternet2.Rd says As from \R 3.3.0 it changes nothing, and only \code{use = NA} is accepted. Also NEWS.Rd says setInternet2() has no effect and will be removed in due course.

Finite, Infinite and NaN Numbers: is.finite(), is.infinite(), is.nan()

In R, basically all mathematical functions (including basic Arithmetic), are supposed to work properly with +/-, Inf and NaN as input or output.

See ?is.finite.

How to replace Inf with NA in All or Specific Columns of the Data Frame

replace() function

File/path operations

  • list.files(, include.dirs =F, recursive = T, pattern = "\\.csv$", all.files = TRUE)
  • file.info()
  • dir.create()
  • file.create()
  • file.copy()
  • file.exists()
  • basename() - remove the parent path, dirname() - returns the part of the path up to but excluding the last path separator
    > file.path("~", "Downloads")
    [1] "~/Downloads"
    > dirname(file.path("~", "Downloads"))
    [1] "/home/brb"
    > basename(file.path("~", "Downloads"))
    [1] "Downloads"
    
  • path.expand("~/.Renviron") # "/home/brb/.Renviron"
  • normalizePath() # Express File Paths in Canonical Form
    > cat(normalizePath(c(R.home(), tempdir())), sep = "\n")
    /usr/lib/R
    /tmp/RtmpzvDhAe
    
  • system.file() - Finds the full file names of files in packages etc
    > system.file("extdata", "ex1.bam", package="Rsamtools")
    [1] "/home/brb/R/x86_64-pc-linux-gnu-library/4.0/Rsamtools/extdata/ex1.bam"
    

read/download/source a file from internet

Simple text file http

retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)

Zip, RData, gz file and url() function

x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
source(con)
close(con)

Here url() function is like file(), gzfile(), bzfile(), xzfile(), unz(), pipe(), fifo(), socketConnection(). They are used to create connections. By default, the connection is not opened (except for ‘socketConnection’), but may be opened by setting a non-empty value of argument ‘open’. See ?url.

Another example is Read gzipped csv directly from a url in R

con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
                       "aa_combined-20110321.csv.gz", sep="")))
txt <- readLines(con)
dat <- read.csv(textConnection(txt))

Another example of using url() is

load(url("http:/www.example.com/example.RData"))

This does not work with load(), dget(), read.table() for files on OneDrive. In fact, I cannot use wget with shared files from OneDrive. The following trick works: How to configure a OneDrive file for use with wget.

Dropbox is easy and works for load(), wget, ...

R download .RData or Directly loading .RData from github from Github.

zip function

This will include 'hallmarkFiles' root folder in the files inside zip.

zip(zipfile = 'myFile.zip', 
    files = dir('hallmarkFiles', full.names = TRUE))

# Verify/view the files. 'list = TRUE' won't extract 
unzip('testZip.zip', list = TRUE) 

downloader package

This package provides a wrapper for the download.file function, making it possible to download files over https on Windows, Mac OS X, and other Unix-like platforms. The RCurl package provides this functionality (and much more) but can be difficult to install because it must be compiled with external dependencies. This package has no external dependencies, so it is much easier to install.

Google drive file based on https using RCurl package

require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
read.csv(textConnection(myCsv))

Google sheet file using googlesheets package

Reading data from google sheets into R

Github files https using RCurl package

x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt", 
            ssl.verifypeer = FALSE)
read.table(text=x)

data summary table

summarytools: create summary tables for vectors and data frames

https://github.com/dcomtois/summarytools. R Package for quickly and neatly summarizing vectors and data frames.

skimr: A frictionless, pipeable approach to dealing with summary statistics

skimr for useful and tidy summary statistics

modelsummary

modelsummary: Summary Tables and Plots for Statistical Models and Data: Beautiful, Customizable, and Publication-Ready

broom

Tidyverse->broom

Create publication tables using tables package

See p13 for example at here

R's tables packages is the best solution. For example,

> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
+          (Sepal.Length + Sepal.Width)*(mean + sd), data=iris )
                                                  
                Sepal.Length      Sepal.Width     
 Species    n   mean         sd   mean        sd  
 setosa      50 5.01         0.35 3.43        0.38
 versicolor  50 5.94         0.52 2.77        0.31
 virginica   50 6.59         0.64 2.97        0.32
 All        150 5.84         0.83 3.06        0.44
> str(iris)
'data.frame':   150 obs. of  5 variables:
 $ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
 $ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
 $ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
 $ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
 $ Species     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 ...

and

# This example shows some of the less common options         
> Sex <- factor(sample(c("Male", "Female"), 100, rep=TRUE))
> Status <- factor(sample(c("low", "medium", "high"), 100, rep=TRUE))
> z <- rnorm(100)+5
> fmt <- function(x) {
  s <- format(x, digits=2)
  even <- ((1:length(s)) %% 2) == 0
  s[even] <- sprintf("(%s)", s[even])
  s
}
> tabular( Justify(c)*Heading()*z*Sex*Heading(Statistic)*Format(fmt())*(mean+sd) ~ Status )
                  Status              
 Sex    Statistic high   low    medium
 Female mean       4.88   4.96   5.17 
        sd        (1.20) (0.82) (1.35)
 Male   mean       4.45   4.31   5.05 
        sd        (1.01) (0.93) (0.75)

fgsea example

vignette & source code

(archived) ClinReport: Statistical Reporting in Clinical Trials

https://cran.r-project.org/web/packages/ClinReport/index.html

Append figures to PDF files

How to append a plot to an existing pdf file. Hint: use the recordPlot() function.

Save base graphics as pseudo-objects

Save base graphics as pseudo-objects in R. Note there are some cons with this approach.

pdf(NULL)
dev.control(displaylist="enable")
plot(df$x, df$y)
text(40, 0, "Random")
text(60, 2, "Text")
lines(stats::lowess(df$x, df$y))
p1.base <- recordPlot()
invisible(dev.off())

# Display the saved plot
grid::grid.newpage()
p1.base

Extracting tables from PDFs

Print tables

addmargins()

tableone

Some examples

Cox models

finalfit package

table1

gtsummary

gt*

dplyr

https://stackoverflow.com/a/34587522. The output includes counts and proportions in a publication like fashion.

tables::tabular()

gmodels::CrossTable()

https://www.statmethods.net/stats/frequencies.html

base::prop.table(x, margin)

New function ‘proportions()’ and ‘marginSums()’. These should replace the unfortunately named ‘prop.table()’ and ‘margin.table()’. for R 4.0.0.

R> m <- matrix(1:4, 2)
R> prop.table(m, 1) # row percentage
          [,1]      [,2]
[1,] 0.2500000 0.7500000
[2,] 0.3333333 0.6666667
R> prop.table(m, 2) # column percentage
          [,1]      [,2]
[1,] 0.3333333 0.4285714
[2,] 0.6666667 0.5714286

stats::xtabs()

stats::ftable()

> ftable(Titanic, row.vars = 1:3)
                   Survived  No Yes
Class Sex    Age                   
1st   Male   Child            0   5
             Adult          118  57
      Female Child            0   1
             Adult            4 140
2nd   Male   Child            0  11
             Adult          154  14
      Female Child            0  13
             Adult           13  80
3rd   Male   Child           35  13
             Adult          387  75
      Female Child           17  14
             Adult           89  76
Crew  Male   Child            0   0
             Adult          670 192
      Female Child            0   0
             Adult            3  20
> ftable(Titanic, row.vars = 1:2, col.vars = "Survived")
             Survived  No Yes
Class Sex                    
1st   Male            118  62
      Female            4 141
2nd   Male            154  25
      Female           13  93
3rd   Male            422  88
      Female          106  90
Crew  Male            670 192
      Female            3  20
> ftable(Titanic, row.vars = 2:1, col.vars = "Survived")
             Survived  No Yes
Sex    Class                 
Male   1st            118  62
       2nd            154  25
       3rd            422  88
       Crew           670 192
Female 1st              4 141
       2nd             13  93
       3rd            106  90
       Crew             3  20
> str(Titanic)
 table [1:4, 1:2, 1:2, 1:2] 0 0 35 0 0 0 17 0 118 154 ...
 - attr(*, "dimnames")=List of 4
  ..$ Class   : chr [1:4] "1st" "2nd" "3rd" "Crew"
  ..$ Sex     : chr [1:2] "Male" "Female"
  ..$ Age     : chr [1:2] "Child" "Adult"
  ..$ Survived: chr [1:2] "No" "Yes"
> x <- ftable(mtcars[c("cyl", "vs", "am", "gear")])
> x
          gear  3  4  5
cyl vs am              
4   0  0        0  0  0
       1        0  0  1
    1  0        1  2  0
       1        0  6  1
6   0  0        0  0  0
       1        0  2  1
    1  0        2  2  0
       1        0  0  0
8   0  0       12  0  0
       1        0  0  2
    1  0        0  0  0
       1        0  0  0
> ftable(x, row.vars = c(2, 4))
        cyl  4     6     8   
        am   0  1  0  1  0  1
vs gear                      
0  3         0  0  0  0 12  0
   4         0  0  0  2  0  0
   5         0  1  0  1  0  2
1  3         1  0  2  0  0  0
   4         2  6  2  0  0  0
   5         0  1  0  0  0  0
> 
> ## Start with expressions, use table()'s "dnn" to change labels
> ftable(mtcars$cyl, mtcars$vs, mtcars$am, mtcars$gear, row.vars = c(2, 4),
         dnn = c("Cylinders", "V/S", "Transmission", "Gears"))

          Cylinders     4     6     8   
          Transmission  0  1  0  1  0  1
V/S Gears                               
0   3                   0  0  0  0 12  0
    4                   0  0  0  2  0  0
    5                   0  1  0  1  0  2
1   3                   1  0  2  0  0  0
    4                   2  6  2  0  0  0
    5                   0  1  0  0  0  0

tracemem, data type, copy

How to avoid copying a long vector

Tell if the current R is running in 32-bit or 64-bit mode

8 * .Machine$sizeof.pointer

where sizeof.pointer returns the number of *bytes* in a C SEXP type and '8' means number of bits per byte.

32- and 64-bit

See R-admin.html.

  • For speed you may want to use a 32-bit build, but to handle large datasets a 64-bit build.
  • Even on 64-bit builds of R there are limits on the size of R objects, some of which stem from the use of 32-bit integers (especially in FORTRAN code). For example, the dimensionas of an array are limited to 2^31 -1.
  • Since R 2.15.0, it is possible to select '64-bit Files' from the standard installer even on a 32-bit version of Windows (2012/3/30).

Handling length 2^31 and more in R 3.0.0

From R News for 3.0.0 release:

There is a subtle change in behaviour for numeric index values 2^31 and larger. These never used to be legitimate and so were treated as NA, sometimes with a warning. They are now legal for long vectors so there is no longer a warning, and x[2^31] <- y will now extend the vector on a 64-bit platform and give an error on a 32-bit one.

In R 2.15.2, if I try to assign a vector of length 2^31, I will get an error

> x <- seq(1, 2^31)
Error in from:to : result would be too long a vector

However, for R 3.0.0 (tested on my 64-bit Ubuntu with 16GB RAM. The R was compiled by myself):

> system.time(x <- seq(1,2^31))
   user  system elapsed
  8.604  11.060 120.815
> length(x)
[1] 2147483648
> length(x)/2^20
[1] 2048
> gc()
             used    (Mb) gc trigger    (Mb)   max used    (Mb)
Ncells     183823     9.9     407500    21.8     350000    18.7
Vcells 2147764406 16386.2 2368247221 18068.3 2148247383 16389.9
>

Note:

  1. 2^31 length is about 2 Giga length. It takes about 16 GB (2^31*8/2^20 MB) memory.
  2. On Windows, it is almost impossible to work with 2^31 length of data if the memory is less than 16 GB because virtual disk on Windows does not work well. For example, when I tested on my 12 GB Windows 7, the whole Windows system freezes for several minutes before I force to power off the machine.
  3. My slide in http://goo.gl/g7sGX shows the screenshots of running the above command on my Ubuntu and RHEL machines. As you can see the linux is pretty good at handling large (> system RAM) data. That said, as long as your linux system is 64-bit, you can possibly work on large data without too much pain.
  4. For large dataset, it makes sense to use database or specially crafted packages like bigmemory or ff or bigstatsr.
  5. [[<- for index 2^31 fails

NA in index

  • Question: what is seq(1, 3)[c(1, 2, NA)]?

Answer: It will reserve the element with NA in indexing and return the value NA for it.

  • Question: What is TRUE & NA?

Answer: NA

  • Question: What is FALSE & NA?

Answer: FALSE

  • Question: c("A", "B", NA) != "" ?

Answer: TRUE TRUE NA

  • Question: which(c("A", "B", NA) != "") ?

Answer: 1 2

  • Question: c(1, 2, NA) != "" & !is.na(c(1, 2, NA)) ?

Answer: TRUE TRUE FALSE

  • Question: c("A", "B", NA) != "" & !is.na(c("A", "B", NA)) ?

Answer: TRUE TRUE FALSE

Conclusion: In order to exclude empty or NA for numerical or character data type, we can use which() or a convenience function keep.complete(x) <- function(x) x != "" & !is.na(x). This will guarantee return logical values and not contain NAs.

Don't just use x != "" OR !is.na(x).

Some functions

Constant and 'L'

Add 'L' after a constant. For example,

for(i in 1L:n) { }

if (max.lines > 0L) { }

label <- paste0(n-i+1L, ": ")

n <- length(x);  if(n == 0L) { }

Vector/Arrays

R indexes arrays from 1 like Fortran, not from 0 like C or Python.

remove integer(0)

How to remove integer(0) from a vector?

Append some elements

append() and its after argument

setNames()

Assign names to a vector

z <- setNames(1:3, c("a", "b", "c"))
# OR
z <- 1:3; names(z) <- c("a", "b", "c")
# OR
z <- c("a"=1, "b"=2, "c"=3) # not work if "a", "b", "c" is like x[1], x[2], x[3].

Factor

labels argument

We can specify the factor levels and new labels using the factor() function.

sex <- factor(sex, levels = c("0", "1"), labels = c("Male", "Female"))
drug_treatment <- factor(drug_treatment, levels = c("Placebo", "Low dose", "High dose"))
health_status <- factor(health_status, levels = c("Healthy", "Alzheimer's"))

factor(rev(letters[1:3]), labels = c("A", "B", "C"))
# C B A
# Levels: A B C

Create a factor/categorical variable from a continuous variable: cut() and dplyr::case_when()

cut(
     c(0, 10, 30), 
     breaks = c(0, 30, 50, Inf), 
     labels = c("Young", "Middle-aged", "Elderly")
 )  # Default include.lowest = FALSE
# [1] <NA>  Young Young
  • ?cut
    set.seed(1)
    x <- rnorm(100)
    facVar <- cut(x, c(min(x), -1, 1, max(x)), labels = c("low", "medium", "high"))
    table(facVar, useNA = "ifany")
    facVar
    #   low medium   high   <NA> 
    #    10     74     15      1 
    

    Note the option include.lowest = TRUE is needed when we use cut() + quantile(); otherwise the smallest data will become NA since the intervals have the format (a, b].

    x2 <- cut(x, quantile(x, 0:2/2), include.lowest = TRUE) # split x into 2 levels
    x2 <- cut(x, quantile(x, 0:3/3), include.lowest = TRUE) # split x into 3 levels
    
    library(tidyverse); library(magrittr)
    set.seed(1)
    breaks <- quantile(runif(100), probs=seq(0, 1, len=20))
    x <- runif(50)
    bins <- cut(x, breaks=unique(breaks), include.lowest=T, right=T)
    
    data.frame(sc=x, bins=bins) %>% 
      group_by(bins) %>% 
      summarise(n=n()) %>% 
      ggplot(aes(x = bins, y = n)) + 
        geom_col(color = "black", fill = "#90AACB") + 
        theme_minimal() + 
        theme(axis.text.x = element_text(angle = 90)) + 
        theme(legend.position = "none") + coord_flip()
    
  • A Guide to Using the cut() Function in R
  • tibble object
    library(tidyverse)
    tibble(age_yrs = c(0, 4, 10, 15, 24, 55),
           age_cat = case_when(
              age_yrs < 2 ~ "baby",
              age_yrs < 13 ~ "kid",
              age_yrs < 20 ~ "teen",
              TRUE         ~ "adult")
    )
    
  • R tip: Learn dplyr’s case_when() function
    case_when(
      condition1 ~ value1, 
      condition2 ~ value2,
      TRUE ~ ValueAnythingElse
    )
    # Example
    case_when(
      x %%2 == 0 ~ "even",
      x %%2 == 1 ~ "odd",
      TRUE ~ "Neither even or odd"
    )
    

How to change one of the level to NA

https://stackoverflow.com/a/25354985. Note that the factor level is removed.

x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA

Creating missing values in factors

Concatenating two factor vectors

Not trivial. How to concatenate factors, without them being converted to integer level?.

unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))

droplevels()

droplevels(): drop unused levels from a factor or, more commonly, from factors in a data frame.

factor(x , levels = ...) vs levels(x) <-

Note levels(x) is to set/rename levels, not reorder. Use relevel() or factor() to reorder.

levels()
plyr::revalue()
forcats::fct_recode()
rename levels
factor(, levels) reorder levels
sizes <- factor(c("small", "large", "large", "small", "medium"))
sizes
#> [1] small  large  large  small  medium
#> Levels: large medium small

sizes2 <- factor(sizes, levels = c("small", "medium", "large")) # reorder levels but data is not changed
sizes2
# [1] small  large  large  small  medium
# Levels: small medium large

sizes3 <- sizes
levels(sizes3) <- c("small", "medium", "large") # rename, not reorder
                                                # large -> small
                                                # medium -> medium
                                                # small -> large 
sizes3
# [1] large  small  small  large  medium
# Levels: small medium large

A regression example.

set.seed(1)
x <- sample(1:2, 500, replace = TRUE)
y <- round(x + rnorm(500), 3)
x <- as.factor(x)
sample_data <- data.frame(x, y)
 
# create linear model
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  0.96804    0.06610   14.65   <2e-16 ***
# x2           0.99620    0.09462   10.53   <2e-16 ***

# Wrong way when we want to change the baseline level to '2'
# No change on the model fitting except the apparent change on the variable name in the printout
levels(sample_data$x) <- c("2", "1")
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  0.96804    0.06610   14.65   <2e-16 ***
# x1           0.99620    0.09462   10.53   <2e-16 ***

# Correct way if we want to change the baseline level to '2'
# The estimate was changed by flipping the sign from the original data
sample_data$x <- relevel(x, ref = "2")
summary(lm( y~x, sample_data))
# Coefficients:
#             Estimate Std. Error t value Pr(>|t|)    
# (Intercept)  1.96425    0.06770   29.01   <2e-16 ***
# x1          -0.99620    0.09462  -10.53   <2e-16 ***

stats::relevel()

relevel. This function can only be used to change the reference level of a factor variable. It does not directly create an arbitrary order of levels. That is, it is useful in lm() or aov(), etc.

reorder(), levels() and boxplot()

  • How to Reorder Boxplots in R: A Comprehensive Guide (tapply() method, simple & effective)
  • reorder().This is useful in barplot (ggplot2::geom_col()) where we want to sort the bars by a numerical variable.
    # Syntax:
    # newFac <- with(df, reorder(fac, vec, FUN=mean)) # newFac is like fac except it has a new order
    
    (bymedian <- with(InsectSprays, reorder(spray, count, median)) )
    class(bymedian)
    levels(bymedian)
    boxplot(count ~ bymedian, data = InsectSprays,
            xlab = "Type of spray", ylab = "Insect count",
            main = "InsectSprays data", varwidth = TRUE,
            col = "lightgray") # boxplots are sorted according to the new levels
    boxplot(count ~ spray, data = InsectSprays,
            xlab = "Type of spray", ylab = "Insect count",
            main = "InsectSprays data", varwidth = TRUE,
            col = "lightgray") # not sorted
    
  • Statistics Sunday: My 2019 Reading (reorder function)

factor() vs ordered()

factor(levels=c("a", "b", "c"), ordered=TRUE)
# ordered(0)
# Levels: a < b < c

factor(levels=c("a", "b", "c"))
# factor(0)
# Levels: a b c

ordered(levels=c("a", "b", "c"))
# Error in factor(x, ..., ordered = TRUE) : 
#  argument "x" is missing, with no default

Data frame

stringsAsFactors = FALSE

http://www.win-vector.com/blog/2018/03/r-tip-use-stringsasfactors-false/

We can use options(stringsAsFactors=FALSE) forces R to import character data as character objects.

In R 4.0.0, stringAsFactors=FALSE will be default. This also affects read.table() function.

check.names = FALSE

Note this option will not affect rownames. So if the rownames contains special symbols, like dash, space, parentheses, etc, they will not be modified.

> data.frame("1a"=1:2, "2a"=1:2, check.names = FALSE)
  1a 2a
1  1  1
2  2  2
> data.frame("1a"=1:2, "2a"=1:2) # default
  X1a X2a
1   1   1
2   2   2

Create unique rownames: make.unique()

groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
rownames(mydf) <- make.unique(groupCodes)

data.frame() will change rownames

class(df2)
# [1] "matrix" "array"
rownames(df2)[c(9109, 44999)]
# [1] "A1CF"     "A1BG-AS1"
rownames(data.frame(df2))[c(9109, 44999)]
# [1] "A1CF"     "A1BG.AS1"

Print a data frame without rownames

# Method 1. 
rownames(df1) <- NULL

# Method 2. 
print(df1, row.names = FALSE)

Convert data frame factor columns to characters

Convert data.frame columns from factors to characters

# Method 1:
bob <- data.frame(lapply(bob, as.character), stringsAsFactors=FALSE)

# Method 2:
bob[] <- lapply(bob, as.character)

To replace only factor columns:

# Method 1:
i <- sapply(bob, is.factor)
bob[i] <- lapply(bob[i], as.character)

# Method 2:
library(dplyr)
bob %>% mutate_if(is.factor, as.character) -> bob

Sort Or Order A Data Frame

How To Sort Or Order A Data Frame In R

  1. df[order(df$x), ], df[order(df$x, decreasing = TRUE), ], df[order(df$x, df$y), ]
  2. library(plyr); arrange(df, x), arrange(df, desc(x)), arrange(df, x, y)
  3. library(dplyr); df %>% arrange(x),df %>% arrange(x, desc(x)), df %>% arrange(x, y)
  4. library(doBy); order(~x, df), order(~ -x, df), order(~ x+y, df)

data.frame to vector

df <- data.frame(x = c(1, 2, 3), y = c(4, 5, 6))

class(df)
# [1] "data.frame"
class(t(df))
# [1] "matrix" "array"
class(unlist(df))
# [1] "numeric"

# Method 1: Convert data frame to matrix using as.matrix()
# and then Convert matrix to vector using as.vector() or c()
mat <- as.matrix(df)
vec1 <- as.vector(mat)   # [1] 1 2 3 4 5 6
vec2 <- c(mat)

# Method 2: Convert data frame to matrix using t()/transpose
# and then Convert matrix to vector using as.vector() or c()
vec3 <- as.vector(t(df)) # [1] 1 4 2 5 3 6
vec4 <- c(t(df))

# Not working
as.vector(df)
# $x
# [1] 1 2 3
# $y
# [1] 4 5 6

# Method 3: unlist() - easiest solution
unlist(df)
# x1 x2 x3 y1 y2 y3 
#  1  2  3  4  5  6 
unlist(data.frame(df), use.names = F) # OR dplyr::pull()
# [1] 1 2 3 4 5 6

Q: Why as.vector(df) cannot convert a data frame into a vector?

A: The as.vector function cannot be used directly on a data frame to convert it into a vector because a data frame is a list of vectors (i.e., its columns) and as.vector only removes the attributes of an object to create a vector. When you apply as.vector to a data frame, R does not know how to concatenate these independent columns (which could be of different types) into a single vector. Therefore, it doesn’t perform the operation. Therefore as.vector() returns the underlying list structure of the data frame instead of converting it into a vector.

However, when you transpose the data frame using t(), it gets converted into a matrix. A matrix in R is a vector with dimensions. Therefore, all elements of the matrix must be of the same type. If they are not, R will coerce them to be so. Once you have a matrix, as.vector() can easily convert it into a vector because all elements are of the same type.

Using cbind() to merge vectors together?

It’s a common mistake to try and create a data frame by cbind()ing vectors together. This doesn’t work because cbind() will create a matrix unless one of the arguments is already a data frame. Instead use data.frame() directly. See Advanced R -> Data structures chapter.

cbind NULL and data.frame

cbind can't combine NULL with dataframe. Add as.matrix() will fix the problem.

merge

Special character in the matched variable can create a trouble when we use merge() or dplyr::inner_join(). I guess R internally turns df2 (a matrix but not a data frame) to a data frame (so rownames are changed if they contain special character like "-"). This still does not explain the situation when I

class(df1); class(df2)
# [1] "data.frame"  # 2 x 2
# [1] "matrix" "array" # 52439 x 2
rownames(df1)
# [1] "A1CF"     "A1BG-AS1"
merge(df1, df2[c(9109, 44999), ], by=0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1  A1BG-AS1     0    0   7.321358   6.908333
# 2      A1CF     0    0   3.011470   1.189578
merge(df1, df2[c(9109, 38959:44999), ], by= 0) # still correct
merge(df1, df2[c(9109, 38958:44999), ], by= 0) # same as merge(df1, df2, by=0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1      A1CF     0    0    3.01147   1.189578
rownames(df2)[38958:38959]
# [1] "ITFG2-AS1"  "ADGRD1-AS1"

rownames(df1)[2] <- "A1BGAS1"
rownames(df2)[44999] <- "A1BGAS1"
merge(df1, df2, by= 0)
#   Row.names 786-0 A498 ACH-000001 ACH-000002
# 1   A1BGAS1     0    0   7.321358   6.908333
# 2      A1CF     0    0   3.011470   1.189578

is.matrix: data.frame is not necessarily a matrix

See ?matrix. is.matrix returns TRUE if x is a vector and has a "dim" attribute of length 2 and FALSE otherwise.

An example that is a data frame (is.data.frame() returns TRUE) but not a matrix (is.matrix() returns FALSE) is an object returned by

X <- data.frame(x=1:2, y=3:4)

The 'X' object is NOT a vector and it does NOT have the "dim" attribute. It has only 3 attributes: "names", "row.names" & "class". Note that dim() function works fine and returns correctly though there is not "dim" attribute.

Another example that is a data frame but not a matrix is the built-in object cars; see ?matrix. It is not a vector

Convert a data frame to a matrix: as.matrix() vs data.matrix()

If I have a data frame X which recorded the time of some files.

  • is.data.frame(X) shows TRUE but is.matrix(X) show FALSE
  • as.matrix(X) will keep the time mode. The returned object is not a data frame anymore.
  • data.matrix(X) will convert the time to numerical values. So use data.matrix() if the data is numeric. The returned object is not a data frame anymore.
# latex directory contains cache files from knitting an rmarkdown file
X <- list.files("latex/", full.names = T) %>%
     grep("RData", ., value=T) %>% 
     file.info() %>%  
     `[`("mtime")
X %>% is.data.frame() # TRUE
X %>% is.matrix() # FALSE
X %>% as.matrix() %>% is.matrix() # TRUE
X %>% data.matrix() %>% is.matrix() # TRUE
X %>% as.matrix() %>% "["(1:2, ) # timestamps
X %>% data.matrix() %>% "["(1:2, ) # numeric
  • The as.matrix() function is used to coerce an object into a matrix. It can be used with various types of R objects, such as vectors, data frames, and arrays.
  • The data.matrix() function is specifically designed for converting a data frame into a matrix by coercing all columns to numeric values. If the data frame contains non-numeric columns, such as character or factor columns, data.matrix() will convert them to numeric values if possible (e.g., by converting factors to their integer codes).
  • See the following example where as.matrix() and data.matrix() return different resuls.
df <- data.frame(a = c(1, 2, 3), b = c("x", "y", "z"))
mat <- as.matrix(df)
mat
#      a   b  
# [1,] "1" "x"
# [2,] "2" "y"
# [3,] "3" "z"
class(mat)
# [1] "matrix" "array" 
mat2 <- data.matrix(df)
mat2
#      a b
# [1,] 1 1
# [2,] 2 2
# [3,] 3 3
class(mat2)
# [1] "matrix" "array" 
typeof(mat)
# [1] "character"
typeof(mat2)
# [1] "double"

matrix vs data.frame

Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.

Browse[2]> names(res2$surv.data.new[[index]])
NULL
Browse[2]> colnames(res2$surv.data.new[[index]])
 [1] "time"   "status" "treat"  "AKT1"   "BRAF"   "FLOT2"  "MTOR"   "PCK2"   "PIK3CA"
[10] "RAF1"  
Browse[2]> mode(res2$surv.data.new[[index]])
[1] "numeric"
Browse[2]> is.matrix(res2$surv.data.new[[index]])
[1] TRUE
Browse[2]> dim(res2$surv.data.new[[index]])
[1] 991  10

Case 2:

ip1 <- installed.packages()[,c(1,3:4)] # class(ip1) = 'matrix'
unique(ip1$Priority)
# Error in ip1$Priority : $ operator is invalid for atomic vectors
unique(ip1[, "Priority"])   # OK

ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
unique(ip2$Priority)     # OK

The length of a matrix and a data frame is different.

> length(matrix(1:6, 3, 2))
[1] 6
> length(data.frame(matrix(1:6, 3, 2)))
[1] 2
> x[1]
  X1
1  1
2  2
3  3
4  4
5  5
6  6
> x1
[1] 1 2 3 4 5 6

So the length of a data frame is the number of columns. When we use sapply() function on a data frame, it will apply to each column of the data frame.

How to Remove Duplicates

How to Remove Duplicates in R with Example

Convert a matrix (not data frame) of characters to numeric

Just change the mode of the object

tmp <- cbind(a=c("0.12", "0.34"), b =c("0.567", "0.890")); tmp
     a     b
1 0.12 0.567
2 0.34 0.890
> is.data.frame(tmp) # FALSE
> is.matrix(tmp)     # TRUE
> sum(tmp)
Error in sum(tmp) : invalid 'type' (character) of argument
> mode(tmp)  # "character"

> mode(tmp) <- "numeric"
> sum(tmp)
[1] 1.917

Convert Data Frame Row to Vector

as.numeric() or c()

Convert characters to integers

mode(x) <- "integer"

Non-Standard Evaluation

Understanding Non-Standard Evaluation. Part 1: The Basics

Select Data Frame Columns in R

This is part of series of DATA MANIPULATION IN R from datanovia.com

  • pull(): Extract column values as a vector. The column of interest can be specified either by name or by index.
  • select(): Extract one or multiple columns as a data table. It can be also used to remove columns from the data frame.
  • select_if(): Select columns based on a particular condition. One can use this function to, for example, select columns if they are numeric.
  • Helper functions - starts_with(), ends_with(), contains(), matches(), one_of(): Select columns/variables based on their names

Another way is to the dollar sign $ operator (?"$") to extract rows or column from a data frame.

class(USArrests)  # "data.frame"
USArrests$"Assault"

Note that for both data frame and matrix objects, we need to use the [ operator to extract columns and/or rows.

USArrests[c("Alabama", "Alask"), c("Murder", "Assault")]
#         Murder Assault
# Alabama   13.2     236
# Alaska    10.0     263
USArrests[c("Murder", "Assault")]  # all rows

tmp <- data(package="datasets")
class(tmp$results)  # "matrix" "array" 
tmp$results[, "Item"]
# Same method can be used if rownames are available in a matrix

Note for a data.table object, we can extract columns using the column names without double quotes.

data.table(USArrests)[1:2, list(Murder, Assault)]

Add columns to a data frame

How to add columns to a data frame in R

Exclude/drop/remove data frame columns

# method 1
df = subset(mydata, select = -c(x,z) )

# method 2
drop <- c("x","z")
df = mydata[,!(names(mydata) %in% drop)]

# method 3: dplyr
mydata2 = select(mydata, -a, -x, -y)
mydata2 = select(mydata, -c(a, x, y))
mydata2 = select(mydata, -a:-y)
mydata2 = mydata[,!grepl("^INC",names(mydata))]

Remove Rows from the data frame

Remove Rows from the data frame in R

Danger of selecting rows from a data frame

> dim(cars)
[1] 50  2
> data.frame(a=cars[1,], b=cars[2, ])
  a.speed a.dist b.speed b.dist
1       4      2       4     10
> dim(data.frame(a=cars[1,], b=cars[2, ]))
[1] 1 4
> cars2 = as.matrix(cars)
> data.frame(a=cars2[1,], b=cars2[2, ])
      a  b
speed 4  4
dist  2 10

Creating data frame using structure() function

Creating data frame using structure() function in R

Create an empty data.frame

https://stackoverflow.com/questions/10689055/create-an-empty-data-frame

# the column types default as logical per vector(), but are then overridden
a = data.frame(matrix(vector(), 5, 3,
               dimnames=list(c(), c("Date", "File", "User"))),
               stringsAsFactors=F)
str(a) # NA but they are logical , not numeric.
a[1,1] <- rnorm(1)
str(a)

# similar to above
a <- data.frame(matrix(NA, nrow = 2, ncol = 3))

# different data type
a <- data.frame(x1 = character(),
                x2 = numeric(),
                x3 = factor(),
                stringsAsFactors = FALSE)

Objects from subsetting a row in a data frame vs matrix

  • Subsetting creates repeated rows. This will create unexpected rownames.
    R> z <- data.frame(x=1:3, y=2:4)
    R> rownames(z) <- letters[1:3]
    R> rownames(z)[c(1,1)]
    [1] "a" "a"
    R> rownames(z[c(1,1),])
    [1] "a"   "a.1"
    R> z[c(1,1), ]
        x y
    a   1 2
    a.1 1 2
    
  • Convert a dataframe to a vector (by rows) The solution is as.vector(t(mydf[i, ])) or c(mydf[i, ]). My example:
    str(trainData)
    # 'data.frame':	503 obs. of  500 variables:
    #  $ bm001: num  0.429 1 -0.5 1.415 -1.899 ...
    #  $ bm002: num  0.0568 1 0.5 0.3556 -1.16 ...
    # ...
    trainData[1:3, 1:3]
    #        bm001      bm002    bm003
    # 1  0.4289449 0.05676296 1.657966
    # 2  1.0000000 1.00000000 1.000000
    # 3 -0.5000000 0.50000000 0.500000
    o <- data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3, ], t(TData))
    # Warning message:
    # In data.frame(time = trainData[1, ], status = trainData[2, ], treat = trainData[3,  :
    #   row names were found from a short variable and have been discarded
    

    'trees' data from the 'datasets' package

    trees[1:3,]
    #   Girth Height Volume
    # 1   8.3     70   10.3
    # 2   8.6     65   10.3
    # 3   8.8     63   10.2
    
    # Wrong ways:
    data.frame(trees[1,] , trees[2,])
    #   Girth Height Volume Girth.1 Height.1 Volume.1
    # 1   8.3     70   10.3     8.6       65     10.3
    data.frame(time=trees[1,] , status=trees[2,])
    #   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    data.frame(time=as.vector(trees[1,]) , status=as.vector(trees[2,]))
    #   time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    data.frame(time=c(trees[1,]) , status=c(trees[2,]))
    # time.Girth time.Height time.Volume status.Girth status.Height status.Volume
    # 1        8.3          70        10.3          8.6            65          10.3
    
    # Right ways:
    # method 1: dropping row names
    data.frame(time=c(t(trees[1,])) , status=c(t(trees[2,]))) 
    # OR
    data.frame(time=as.numeric(trees[1,]) , status=as.numeric(trees[2,]))
    #   time status
    # 1  8.3    8.6
    # 2 70.0   65.0
    # 3 10.3   10.3
    # method 2: keeping row names
    data.frame(time=t(trees[1,]) , status=t(trees[2,]))
    #          X1   X2
    # Girth   8.3  8.6
    # Height 70.0 65.0
    # Volume 10.3 10.3
    data.frame(time=unlist(trees[1,]) , status=unlist(trees[2,]))
    #        time status
    # Girth   8.3    8.6
    # Height 70.0   65.0
    # Volume 10.3   10.3
    
    # Method 3: convert a data frame to a matrix
    is.matrix(trees)
    # [1] FALSE
    trees2 <- as.matrix(trees)
    data.frame(time=trees2[1,] , status=trees2[2,]) # row names are kept
    #        time status
    # Girth   8.3    8.6
    # Height 70.0   65.0
    # Volume 10.3   10.3
    
    dim(trees[1,])
    # [1] 1 3
    dim(trees2[1, ])
    # NULL
    trees[1, ]  # notice the row name '1' on the left hand side
    #   Girth Height Volume
    # 1   8.3     70   10.3
    trees2[1, ]
    #  Girth Height Volume
    #    8.3   70.0   10.3
    

Convert a list to data frame

How to Convert a List to a Data Frame in R.

# method 1
data.frame(t(sapply(my_list,c)))

# method 2
library(dplyr)
bind_rows(my_list) # OR bind_cols(my_list)

# method 3
library(data.table)
rbindlist(my_list)

tibble and data.table

Clean a dataset

How to clean the datasets in R

matrix

Define and subset a matrix

  • Matrix in R
    • It is clear when a vector becomes a matrix the data is transformed column-wisely (byrow = FALSE, by default).
    • When subsetting a matrix, it follows the format: X[rows, colums] or X[y-axis, x-axis].
data <- c(2, 4, 7, 5, 10, 1)
A <- matrix(data, ncol = 3)
print(A)
#      [,1] [,2] [,3]
# [1,]    2    7   10
# [2,]    4    5    1

A[1:1, 2:3, drop=F]
#      [,1] [,2]
# [1,]    7   10

Prevent automatic conversion of single column to vector

use drop = FALSE such as mat[, 1, drop = FALSE].

complete.cases(): remove rows with missing in any column

It works on a sequence of vectors, matrices and data frames.

NROW vs nrow

?nrow. Use NROW/NCOL instead of nrow/ncol to treat vectors as 1-column matrices.

matrix (column-major order) multiply a vector

> matrix(1:6, 3,2)
     [,1] [,2]
[1,]    1    4
[2,]    2    5
[3,]    3    6
> matrix(1:6, 3,2) * c(1,2,3) # c(1,2,3) will be recycled to form a matrix. Good quiz.
     [,1] [,2]
[1,]    1    4
[2,]    4   10
[3,]    9   18
> matrix(1:6, 3,2) * c(1,2,3,4) # c(1,2,3,4) will be recycled
     [,1] [,2]
[1,]    1   16
[2,]    4    5
[3,]    9   12

add a vector to all rows of a matrix

add a vector to all rows of a matrix. sweep() or rep() is the best.

sparse matrix

R convert matrix or data frame to sparseMatrix

To subset a vector from some column of a sparseMatrix, we need to convert it to a regular vector, as.vector().

Attributes

Names

Useful functions for dealing with object names. (Un)Setting object names: stats::setNames(), unname() and rlang::set_names()

Print a vector by suppressing names

Use unname. sapply(, , USE.NAMES = FALSE).

format.pval/print p-values/format p values

format.pval(). By default it will show 5 significant digits (getOption("digits")-2).

> set.seed(1); format.pval(c(stats::runif(5), pi^-100, NA))
[1] "0.26551" "0.37212" "0.57285" "0.90821" "0.20168" "< 2e-16" "NA"
> format.pval(c(0.1, 0.0001, 1e-27))
[1] "1e-01"  "1e-04"  "<2e-16"

R> pvalue
[1] 0.0004632104
R> print(pvalue, digits =20)
[1] 0.00046321036188223807528
R> format.pval(pvalue)
[1] "0.00046321"
R> format.pval(pvalue * 1e-1)
[1] "4.6321e-05"
R> format.pval(0.00004632)
[1] "4.632e-05"
R> getOption("digits")
[1] 7

Return type

The format.pval() function returns a string, so it’s not appropriate to use the returned object for operations like sorting.

Wrong number of digits in format.pval()

See here. The solution is to apply round() and then format.pval().

x <- c(6.25433625041843e-05, NA, 0.220313341361346, NA, 0.154029880744594, 
   0.0378437685448703, 0.023358329881356, NA, 0.0262561986351483, 
   0.000251274794673796) 
format.pval(x, digits=3)
# [1] "6.25e-05" "NA"       "0.220313" "NA"       "0.154030" "0.037844" "0.023358"
# [8] "NA"       "0.026256" "0.000251"

round(x, 3) |> format.pval(digits=3, eps=.001)
# [1] "<0.001" "NA"     "0.220"  "NA"     "0.154"  "0.038"  "0.023"  "NA"
# [9] "0.026"  "<0.001"

dplr::mutate_if()

library(dplyr)
df <- data.frame(
  char_var = c("A", "B", "C"),
  num_var1 = c(1.123456, 2.123456, 3.123456),
  num_var2 = c(4.654321, 5.654321, 6.654321),
  stringsAsFactors = FALSE
)

# Round numerical variables to 4 digits after the decimal point
df_rounded <- df %>%
  mutate_if(is.numeric, round, digits = 4)

Customize R: options()

Change the default R repository, my .Rprofile

Change R repository

Edit global Rprofile file. On *NIX platforms, it's located in /usr/lib/R/library/base/R/Rprofile although local .Rprofile settings take precedence.

For example, I can specify the R mirror I like by creating a single line .Rprofile file under my home directory. Another good choice of repository is cloud.r-project.org.

Type file.edit("~/.Rprofile")

local({
  r = getOption("repos")
  r["CRAN"] = "https://cran.rstudio.com/"
  options(repos = r)
})
options(continue = "  ", editor = "nano")
message("Hi MC, loading ~/.Rprofile")
if (interactive()) {
  .Last <- function() try(savehistory("~/.Rhistory"))
}

Change the default web browser for utils::browseURL()

When I run help.start() function in LXLE, it cannot find its default web browser (seamonkey). The solution is to put

options(browser='seamonkey')

in the .Rprofile of your home directory. If the browser is not in the global PATH, we need to put the full path above.

For one-time only purpose, we can use the browser option in help.start() function:

> help.start(browser="seamonkey")
If the browser launched by 'seamonkey' is already running, it is *not*
    restarted, and you must switch to its window.
Otherwise, be patient ...

We can work made a change (or create the file) ~/.Renviron or etc/Renviron. See

Change the default editor

On my Linux and mac, the default editor is "vi". To change it to "nano",

options(editor = "nano")

Change prompt and remove '+' sign

See https://stackoverflow.com/a/1448823.

options(prompt="R> ", continue=" ")

digits

  • signif() rounds x to n significant digits.
    R> signif(pi, 3)
    [1] 3.14
    R> signif(pi, 5)
    [1] 3.1416
    
  • The default digits 7 may be too small. For example, if a number is very large, then we may not be able to see (enough) value after the decimal point. The acceptable range is 1-22. See the following examples

In R,

> options()$digits # Default
[1] 7
> print(.1+.2, digits=18)
[1] 0.300000000000000044
> 100000.07 + .04
[1] 100000.1
> options(digits = 16)
> 100000.07 + .04
[1] 100000.11

In Python,

>>> 100000.07 + .04
100000.11

Disable scientific notation in printing: options(scipen)

How to Turn Off Scientific Notation in R?

This also helps with write.table() results. For example, 0.0003 won't become 3e-4 in the output file.

> numer = 29707; denom = 93874
> c(numer/denom, numer, denom) 
[1] 3.164561e-01 2.970700e+04 9.387400e+04

# Method 1. Without changing the global option
> format(c(numer/denom, numer, denom), scientific=FALSE)
[1] "    0.3164561" "29707.0000000" "93874.0000000"

# Method 2. Change the global option
> options(scipen=999)
> numer/denom
[1] 0.3164561
> c(numer/denom, numer, denom)
[1]     0.3164561 29707.0000000 93874.0000000
> c(4/5, numer, denom)
[1]     0.8 29707.0 93874.0

Suppress warnings: options() and capture.output()

Use options(). If warn is negative all warnings are ignored. If warn is zero (the default) warnings are stored until the top--level function returns.

op <- options("warn")
options(warn = -1)
....
options(op)

# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)

suppressWarnings()

suppressWarnings( foo() )

foo <- capture.output( 
 bar <- suppressWarnings( 
 {print( "hello, world" ); 
   warning("unwanted" )} ) ) 

capture.output()

str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")

Converts warnings into errors

options(warn=2)

demo() function

  • How to wait for a keypress in R? PS readline() is different from readLines().
    for(i in 1:2) { print(i); readline("Press [enter] to continue")}
    
  • Hit 'ESC' or Ctrl+c to skip the prompt "Hit <Return> to see next plot:"
  • demo() uses options() to ask users to hit Enter on each plot
    op <- options(device.ask.default = ask)  # ask = TRUE
    on.exit(options(op), add = TRUE)
    

sprintf

paste, paste0, sprintf

this post, 3 R functions that I enjoy

sep vs collapse in paste()

  • sep is used if we supply multiple separate objects to paste(). A more powerful function is tidyr::unite() function.
  • collapse is used to make the output of length 1. It is commonly used if we have only 1 input object
R> paste("a", "A", sep=",") # multi-vec -> multi-vec
[1] "a,A"
R> paste(c("Elon", "Taylor"), c("Mask", "Swift"))
[1] "Elon Mask"    "Taylor Swift"
# OR
R> sprintf("%s, %s", c("Elon", "Taylor"), c("Mask", "Swift"))

R> paste(c("a", "A"), collapse="-") # one-vec/multi-vec  -> one-scale
[1] "a-A"

# When use together, sep first and collapse second
R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
[1] "a,A - b,B - c,C"
R> paste(letters[1:3], LETTERS[1:3], sep=",")
[1] "a,A" "b,B" "c,C"
R> paste(letters[1:3], LETTERS[1:3], sep=",") |> paste(collapse=" - ")
[1] "a,A - b,B - c,C"

Format number as fixed width, with leading zeros

# sprintf()
a <- seq(1,101,25)
sprintf("name_%03d", a)
[1] "name_001" "name_026" "name_051" "name_076" "name_101"

# formatC()
paste("name", formatC(a, width=3, flag="0"), sep="_")
[1] "name_001" "name_026" "name_051" "name_076" "name_101"

# gsub()
paste0("bm", gsub(" ", "0", format(5:15)))
# [1] "bm05" "bm06" "bm07" "bm08" "bm09" "bm10" "bm11" "bm12" "bm13" "bm14" "bm15"

formatC and prettyNum (prettifying numbers)

R> (x <- 1.2345 * 10 ^ (-8:4))
 [1] 1.2345e-08 1.2345e-07 1.2345e-06 1.2345e-05 1.2345e-04 1.2345e-03
 [7] 1.2345e-02 1.2345e-01 1.2345e+00 1.2345e+01 1.2345e+02 1.2345e+03
[13] 1.2345e+04
R> formatC(x)
 [1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "0.0001234" "0.001234"
 [7] "0.01235"   "0.1235"    "1.234"     "12.34"     "123.4"     "1234"
[13] "1.234e+04"
R> formatC(x, digits=3)
 [1] "1.23e-08" "1.23e-07" "1.23e-06" "1.23e-05" "0.000123" "0.00123"
 [7] "0.0123"   "0.123"    "1.23"     "12.3"     " 123"     "1.23e+03"
[13] "1.23e+04"
R> formatC(x, digits=3, format="e")
 [1] "1.234e-08" "1.234e-07" "1.234e-06" "1.234e-05" "1.234e-04" "1.234e-03"
 [7] "1.235e-02" "1.235e-01" "1.234e+00" "1.234e+01" "1.234e+02" "1.234e+03"
[13] "1.234e+04"

R> x <- .000012345
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"

format(x, scientific = TRUE) vs round() vs format.pval()

Print numeric data in exponential format, so .0001 prints as 1e-4

format(c(0.00001156, 0.84134, 2.1669), scientific = T, digits=4)
# [1] "1.156e-05" "8.413e-01" "2.167e+00"
round(c(0.00001156, 0.84134, 2.1669), digits=4)
# [1] 0.0000 0.8413 2.1669

format.pval(c(0.00001156, 0.84134, 2.1669)) # output is char vector
# [1] "1.156e-05" "0.84134"   "2.16690"
format.pval(c(0.00001156, 0.84134, 2.1669), digits=4)
# [1] "1.156e-05" "0.8413"    "2.1669"

Creating publication quality graphs in R

HDF5 : Hierarchical Data Format

HDF5 is an open binary file format for storing and managing large, complex datasets. The file format was developed by the HDF Group, and is widely used in scientific computing.

Formats for writing/saving and sharing data

Efficiently Saving and Sharing Data in R

Write unix format files on Windows and vice versa

https://stat.ethz.ch/pipermail/r-devel/2012-April/063931.html

with() and within() functions

closePr <- with(mariokart, totalPr - shipPr)
head(closePr, 20)

mk <- within(mariokart, {
             closePr <- totalPr - shipPr
     })
head(mk) # new column closePr

mk <- mariokart
aggregate(. ~ wheels + cond, mk, mean)
# create mean according to each level of (wheels, cond)

aggregate(totalPr ~ wheels + cond, mk, mean)

tapply(mk$totalPr, mk[, c("wheels", "cond")], mean)

stem(): stem-and-leaf plot (alternative to histogram), bar chart on terminals

Plot histograms as lines

https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.

x2=invisible(hist(out2$EB))
y2=invisible(hist(out2$Bench))
z2=invisible(hist(out2$EB0.001))

plot(x=x2$mids, y=x2$density, type="l")
lines(y2$mids, y2$density, lty=2, pwd=2)
lines(z2$mids, z2$density, lty=3, pwd=2)

Histogram with density line

hist(x, prob = TRUE)
lines(density(x), col = 4, lwd = 2)

The overlayed density may looks strange in cases for example counts from single-cell RNASeq or p-values from RNASeq (there is a peak around x=0).

Graphical Parameters, Axes and Text, Combining Plots

statmethods.net

15 Questions All R Users Have About Plots

See 15 Questions All R Users Have About Plots. This is a tremendous post. It covers the built-in plot() function and ggplot() from ggplot2 package.

  1. How To Draw An Empty R Plot? plot.new()
  2. How To Set The Axis Labels And Title Of The R Plots?
  3. How To Add And Change The Spacing Of The Tick Marks Of Your R Plot? axis()
  4. How To Create Two Different X- or Y-axes? par(new=TRUE), axis(), mtext(). ?par.
  5. How To Add Or Change The R Plot’s Legend? legend()
  6. How To Draw A Grid In Your R Plot? grid()
  7. How To Draw A Plot With A PNG As Background? rasterImage() from the png package
  8. How To Adjust The Size Of Points In An R Plot? cex argument
  9. How To Fit A Smooth Curve To Your R Data? loess() and lines()
  10. How To Add Error Bars In An R Plot? arrows()
  11. How To Save A Plot As An Image On Disc
  12. How To Plot Two R Plots Next To Each Other? par(mfrow)[which means Multiple Figures (use ROW-wise)], gridBase package, lattice package
  13. How To Plot Multiple Lines Or Points? plot(), lines()
  14. How To Fix The Aspect Ratio For Your R Plots? asp parameter
  15. What Is The Function Of hjust And vjust In ggplot2?

jitter function

Jitterbox.png

Scatterplot with the "rug" function

require(stats)  # both 'density' and its default method
with(faithful, {
    plot(density(eruptions, bw = 0.15))
    rug(eruptions)
    rug(jitter(eruptions, amount = 0.01), side = 3, col = "light blue")
})

File:RugFunction.png

See also the stripchart() function which produces one dimensional scatter plots (or dot plots) of the given data.

Identify/Locate Points in a Scatter Plot

  • ?identify
  • Using the identify function in R
    plot(x, y)
    identify(x, y, labels = names, plot = TRUE) 
    # Use left clicks to select points we want to identify and "esc" to stop the process
    # This will put the labels on the plot and also return the indices of points
    # [1] 143
    names[143]
    

Draw a single plot with two different y-axes

Draw Color Palette

Default palette before R 4.0

palette() # black, red, green3, blue, cyan, magenta, yellow, gray

# Example from Coursera "Statistics for Genomic Data Science" by Jeff Leek
tropical = c('darkorange', 'dodgerblue', 'hotpink', 'limegreen', 'yellow')
palette(tropical)
plot(1:5, 1:5, col=1:5, pch=16, cex=5)

New palette in R 4.0.0

R 4.0: 3 new features, R 4.0.0 now available, and a look back at R's history. For example, we can select "ggplot2" palette to make the base graphics charts that match the color scheme of ggplot2.

R> palette() 
[1] "black"   "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "gray62"
R> palette.pals()
 [1] "R3"              "R4"              "ggplot2"        
 [4] "Okabe-Ito"       "Accent"          "Dark 2"         
 [7] "Paired"          "Pastel 1"        "Pastel 2"       
[10] "Set 1"           "Set 2"           "Set 3"          
[13] "Tableau 10"      "Classic Tableau" "Polychrome 36"  
[16] "Alphabet"
R> palette.colors(palette='R4') # same as palette()
[1] "#000000" "#DF536B" "#61D04F" "#2297E6" "#28E2E5" "#CD0BBC" "#F5C710"
[8] "#9E9E9E"
R> palette("R3")  # nothing return on screen but palette has changed
R> palette() 
[1] "black"   "red"     "green3"  "blue"    "cyan"    "magenta" "yellow" 
[8] "gray"  
R> palette("R4") # reset to the default color palette; OR palette("default")

R> scales::show_col(palette.colors(palette = "Okabe-Ito"))
R> for(id in palette.pals()) { 
     scales::show_col(palette.colors(palette = id))
     title(id)
     readline("Press [enter] to continue") 
   } 

The palette function can also be used to change the color palette. See Setting up Color Palettes in R

palette("ggplot2")
palette(palette()[-1]) # Remove 'black'
   # OR palette(palette.colors(palette = "ggplot2")[-1] )
with(iris, plot(Sepal.Length, Petal.Length, col = Species, pch=16))

cc <- palette()
palette(c(cc,"purple","brown")) # Add two colors
R> colors() |> length() # [1] 657
R> colors(distinct = T) |> length() # [1] 502

evoPalette

Evolve new colour palettes in R with evoPalette

rtist

rtist: Use the palettes of famous artists in your own visualizations.

SVG

Embed svg in html

svglite

svglite is better R's svg(). It was used by ggsave(). svglite 1.2.0, R Graphics Cookbook.

pdf -> svg

Using Inkscape. See this post.

svg -> png

SVG to PNG using the gyro package

read.table

clipboard

source("clipboard")
read.table("clipboard")

inline text

mydf <- read.table(header=T, text='
 cond yval
    A 2
    B 2.5
    C 1.6
')

http(s) connection

temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt", 
                           ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)

read only specific columns

Use 'colClasses' option in read.table, read.delim, .... For example, the following example reads only the 3rd column of the text file and also changes its data type from a data frame to a vector. Note that we have include double quotes around NULL.

x <- read.table("var_annot.vcf", colClasses = c(rep("NULL", 2), "character", rep("NULL", 7)), 
                skip=62, header=T, stringsAsFactors = FALSE)[, 1]
# 
system.time(x <- read.delim("Methylation450k.txt", 
                colClasses = c("character", "numeric", rep("NULL", 188)), stringsAsFactors = FALSE))

To know the number of columns, we might want to read the first row first.

library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()

Another method is to use pipe(), cut or awk. See ways to read only selected columns from a file into R

check.names = FALSE in read.table()

gx <- read.table(file, header = T, row.names =1)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH."        "IGHV1.69"    "IGKV1.5"     "IGKV2.24"    "KRTAP13.2"  
# [7] "KRTAP19.1"   "KRTAP2.4"    "KRTAP5.9"    "KRTAP6.3"    "Kua.UEV"  

gx <- read.table(file, header = T, row.names =1, check.names = FALSE)
colnames(gx) %>% grep("[^[:alnum:] ]", ., value = TRUE)
# [1] "hCG_1642354" "IGH@"        "IGHV1-69"    "IGKV1-5"     "IGKV2-24"    "KRTAP13-2"  
# [7] "KRTAP19-1"   "KRTAP2-4"    "KRTAP5-9"    "KRTAP6-3"    "Kua-UEV"  

setNames()

Change the colnames. See an example from tidymodels

Testing for valid variable names

Testing for valid variable names

make.names(): Make syntactically valid names out of character vectors

  • make.names()
  • A valid variable name consists of letters, numbers and the dot or underline characters. The variable name starts with a letter or the dot not followed by a number. See R variables.
make.names("abc-d") # [1] "abc.d"

Serialization

If we want to pass an R object to C (use recv() function), we can use writeBin() to output the stream size and then use serialize() function to output the stream to a file. See the post on R mailing list.

> a <- list(1,2,3)
> a_serial <- serialize(a, NULL)
> a_length <- length(a_serial)
> a_length
[1] 70
> writeBin(as.integer(a_length), connection, endian="big")
> serialize(a, connection)

In C++ process, I receive one int variable first to get the length, and then read <length> bytes from the connection.

socketConnection

See ?socketconnection.

Simple example

from the socketConnection's manual.

Open one R session

con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
writeLines(LETTERS, con1)
close(con1)

Open another R session (client)

con2 <- socketConnection(Sys.info()["nodename"], port = 22131)
# as non-blocking, may need to loop for input
readLines(con2)
while(isIncomplete(con2)) {
   Sys.sleep(1)
   z <- readLines(con2)
   if(length(z)) print(z)
}
close(con2)

Use nc in client

The client does not have to be the R. We can use telnet, nc, etc. See the post here. For example, on the client machine, we can issue

nc localhost 22131   [ENTER]

Then the client will wait and show anything written from the server machine. The connection from nc will be terminated once close(con1) is given.

If I use the command

nc -v -w 2 localhost -z 22130-22135

then the connection will be established for a short time which means the cursor on the server machine will be returned. If we issue the above nc command again on the client machine it will show the connection to the port 22131 is refused. PS. "-w" switch denotes the number of seconds of the timeout for connects and final net reads.

Some post I don't have a chance to read. http://digitheadslabnotebook.blogspot.com/2010/09/how-to-send-http-put-request-from-r.html

Use curl command in client

On the server,

con1 <- socketConnection(port = 8080, server = TRUE)

On the client,

curl --trace-ascii debugdump.txt http://localhost:8080/

Then go to the server,

while(nchar(x <- readLines(con1, 1)) > 0) cat(x, "\n")

close(con1) # return cursor in the client machine

Use telnet command in client

On the server,

con1 <- socketConnection(port = 8080, server = TRUE)

On the client,

sudo apt-get install telnet
telnet localhost 8080
abcdefg
hijklmn
qestst

Go to the server,

readLines(con1, 1)
readLines(con1, 1)
readLines(con1, 1)
close(con1) # return cursor in the client machine

Some tutorial about using telnet on http request. And this is a summary of using telnet.

Subsetting

Subset assignment of R Language Definition and Manipulation of functions.

The result of the command x[3:5] <- 13:15 is as if the following had been executed

`*tmp*` <- x
x <- "[<-"(`*tmp*`, 3:5, value=13:15)
rm(`*tmp*`)

Avoid Coercing Indices To Doubles

1 or 1L

Careful on NA value

See the example below. base::subset() or dplyr::filter() can remove NA subsets.

R> mydf = data.frame(a=1:3, b=c(NA,5,6))
R> mydf[mydf$b >5, ]
    a  b
NA NA NA
3   3  6
R> mydf[which(mydf$b >5), ]
  a b
3 3 6
R> mydf %>% dplyr::filter(b > 5)
  a b
1 3 6
R> subset(mydf, b>5)
  a b
3 3 6

Implicit looping

set.seed(1)
i <- sample(c(TRUE, FALSE), size=10, replace = TRUE)
# [1]  TRUE FALSE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE FALSE FALSE
sum(i)        # [1] 6
x <- 1:10
length(x[i])  # [1] 6
x[i[1:3]]     # [1]  1  3  4  6  7  9 10
length(x[i[1:3]]) # [1] 7

modelling

update()

Extract all variable names in lm(), glm(), ...

all.vars(formula(Model)[-2])

as.formula(): use a string in formula in lm(), glm(), ...

? as.formula
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
outcome <- "mpg"
variables <- c("cyl", "disp", "hp", "carb")

# Method 1. The 'Call' portion of the model is reported as “formula = f” 
# our modeling effort, 
# fully parameterized!
f <- as.formula(
  paste(outcome, 
        paste(variables, collapse = " + "), 
        sep = " ~ "))
print(f)
# mpg ~ cyl + disp + hp + carb

model <- lm(f, data = mtcars)
print(model)

# Call:
#   lm(formula = f, data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl         disp           hp         carb  
#     34.021595    -1.048523    -0.026906     0.009349    -0.926863  

# Method 2. eval() + bquote() + ".()"
format(terms(model))  #  or model$terms
# [1] "mpg ~ cyl + disp + hp + carb"

# The new line of code
model <- eval(bquote(   lm(.(f), data = mtcars)   ))

print(model)
# Call:
#   lm(formula = mpg ~ cyl + disp + hp + carb, data = mtcars)
# 
# Coefficients:
#   (Intercept)          cyl         disp           hp         carb  
#     34.021595    -1.048523    -0.026906     0.009349    -0.926863  

# Note if we skip ".()" operator
> eval(bquote(   lm(f, data = mtcars)   ))

Call:
lm(formula = f, data = mtcars)

Coefficients:
(Intercept)          cyl         disp           hp         carb  
  34.021595    -1.048523    -0.026906     0.009349    -0.926863 

reformulate

Simplifying Model Formulas with the R Function ‘reformulate()’

I() function

I() means isolates. See What does the capital letter "I" in R linear regression formula mean?, In R formulas, why do I have to use the I() function on power terms, like y ~ I(x^3)

Aggregating results from linear model

https://stats.stackexchange.com/a/6862

Replacement function "fun(x) <- a"

What are Replacement Functions in R?

R> xx <- c(1,3,66, 99)
R> "cutoff<-" <- function(x, value){
     x[x > value] <- Inf
     x
 }
R> cutoff(xx) <- 65 # xx & 65 are both input
R> xx
[1]   1   3 Inf Inf

R> "cutoff<-"(x = xx, value = 65)
[1]   1   3 Inf Inf

The statement fun(x) <- a and R will read x <- "fun<-"(x,a)

S3 and S4 methods and signature

Debug an S4 function

  • showMethods('FUNCTION')
  • getMethod('FUNCTION', 'SIGNATURE')
  • debug(, signature)
> args(debug)
function (fun, text = "", condition = NULL, signature = NULL) 

> library(genefilter) # Bioconductor
> showMethods("nsFilter")
Function: nsFilter (package genefilter)
eset="ExpressionSet"
> debug(nsFilter, signature="ExpressionSet")

library(DESeq2)
showMethods("normalizationFactors") # show the object class
                                    # "DESeqDataSet" in this case.
getMethod(`normalizationFactors`, "DESeqDataSet") # get the source code

See the source code of normalizationFactors<- (setReplaceMethod() is used) and the source code of estimateSizeFactors(). We can see how avgTxLength was used in estimateNormFactors().

Another example

library(GSVA)
args(gsva) # function (expr, gset.idx.list, ...)

showMethods("gsva")
# Function: gsva (package GSVA)
# expr="ExpressionSet", gset.idx.list="GeneSetCollection"
# expr="ExpressionSet", gset.idx.list="list"
# expr="matrix", gset.idx.list="GeneSetCollection"
# expr="matrix", gset.idx.list="list"
# expr="SummarizedExperiment", gset.idx.list="GeneSetCollection"
# expr="SummarizedExperiment", gset.idx.list="list"

debug(gsva, signature = c(expr="matrix", gset.idx.list="list"))
# OR
# debug(gsva, signature = c("matrix", "list"))
gsva(y, geneSets, method="ssgsea", kcdf="Gaussian")
Browse[3]> debug(.gsva)
# return(ssgsea(expr, gset.idx.list, alpha = tau, parallel.sz = parallel.sz, 
#      normalization = ssgsea.norm, verbose = verbose, 
#      BPPARAM = BPPARAM))

isdebugged("gsva")
# [1] TRUE
undebug(gsva)
library(IRanges)
ir <- IRanges(start=c(10, 20, 30), width=5)
ir

class(ir)
## [1] "IRanges"
## attr(,"package")
## [1] "IRanges"

getClassDef(class(ir))
## Class "IRanges" [package "IRanges"]
## 
## Slots:
##                                                                       
## Name:            start           width           NAMES     elementType
## Class:         integer         integer characterORNULL       character
##                                       
## Name:  elementMetadata        metadata
## Class: DataTableORNULL            list
## 
## Extends: 
## Class "Ranges", directly
## Class "IntegerList", by class "Ranges", distance 2
## Class "RangesORmissing", by class "Ranges", distance 2
## Class "AtomicList", by class "Ranges", distance 3
## Class "List", by class "Ranges", distance 4
## Class "Vector", by class "Ranges", distance 5
## Class "Annotated", by class "Ranges", distance 6
## 
## Known Subclasses: "NormalIRanges"

Check if a function is an S4 method

isS4(foo)

How to access the slots of an S4 object

  • @ will let you access the slots of an S4 object.
  • Note that often the best way to do this is to not access the slot directly but rather through an accessor function (e.g. coefs() rather than digging out the coefficients with $ or @). However, often such functions do not exist so you have to access the slots directly. This will mean that your code breaks if the internal implementation changes, however.
  • R - S4 Classes and Methods Hansen. getClass() or getClassDef().

setReplaceMethod()

See what methods work on an object

see what methods work on an object, e.g. a GRanges object:

methods(class="GRanges")

Or if you have an object, x:

methods(class=class(x))

View S3 function definition: double colon '::' and triple colon ':::' operators and getAnywhere()

?":::"

  • pkg::name returns the value of the exported variable name in namespace pkg
  • pkg:::name returns the value of the internal variable name
base::"+"
stats:::coef.default

predict.ppr
# Error: object 'predict.ppr' not found
stats::predict.ppr
# Error: 'predict.ppr' is not an exported object from 'namespace:stats'
stats:::predict.ppr  # OR  
getS3method("predict", "ppr")

getS3method("t", "test")

methods() + getAnywhere() functions

Read the source code (include Fortran/C, S3 and S4 methods)

S3 method is overwritten

For example, the select() method from dplyr is overwritten by grpreg package.

An easy solution is to load grpreg before loading dplyr.

mcols() and DataFrame() from Bioc S4Vectors package

  • mcols: Get or set the metadata columns.
  • colData: SummarizedExperiment instances from GenomicRanges
  • DataFrame: The DataFrame class extends the DataTable virtual class and supports the storage of any type of object (with length and [ methods) as columns.

For example, in Shrinkage of logarithmic fold changes vignette of the DESeq2paper package

> mcols(ddsNoPrior[genes, ])
DataFrame with 2 rows and 21 columns
   baseMean   baseVar   allZero dispGeneEst    dispFit dispersion  dispIter dispOutlier   dispMAP
  <numeric> <numeric> <logical>   <numeric>  <numeric>  <numeric> <numeric>   <logical> <numeric>
1  163.5750  8904.607     FALSE  0.06263141 0.03862798  0.0577712         7       FALSE 0.0577712
2  175.3883 59643.515     FALSE  2.25306109 0.03807917  2.2530611        12        TRUE 1.6011440
  Intercept strain_DBA.2J_vs_C57BL.6J SE_Intercept SE_strain_DBA.2J_vs_C57BL.6J WaldStatistic_Intercept
  <numeric>                 <numeric>    <numeric>                    <numeric>               <numeric>
1  6.210188                  1.735829    0.1229354                    0.1636645               50.515872
2  6.234880                  1.823173    0.6870629                    0.9481865                9.074686
  WaldStatistic_strain_DBA.2J_vs_C57BL.6J WaldPvalue_Intercept WaldPvalue_strain_DBA.2J_vs_C57BL.6J
                                <numeric>            <numeric>                            <numeric>
1                                10.60602         0.000000e+00                         2.793908e-26
2                                 1.92280         1.140054e-19                         5.450522e-02
   betaConv  betaIter  deviance  maxCooks
  <logical> <numeric> <numeric> <numeric>
1      TRUE         3  210.4045 0.2648753
2      TRUE         9  243.7455 0.3248949

Pipe

Packages take advantage of pipes

  • rstatix: Pipe-Friendly Framework for Basic Statistical Tests

findInterval()

Related functions are cuts() and split(). See also

Assign operator

  • Earlier versions of R used underscore (_) as an assignment operator.
  • Assignments with the = Operator
  • In R 1.8.0 (2003), the assign operator has been removed. See NEWS.
  • In R 1.9.0 (2004), "_" is allowed in valid names. See NEWS.
R162.png

Operator precedence

The ':' operator has higher precedence than '-' so 0:N-1 evaluates to (0:N)-1, not 0:(N-1) like you probably wanted.

order(), rank() and sort()

If we want to find the indices of the first 25 genes with the smallest p-values, we can use order(pval)[1:25].

> x = sample(10)
> x
 [1]  4  3 10  7  5  8  6  1  9  2
> order(x)
 [1]  8 10  2  1  5  7  4  6  9  3
> rank(x)
 [1]  4  3 10  7  5  8  6  1  9  2
> rank(10*x)
 [1]  4  3 10  7  5  8  6  1  9  2

> x[order(x)]
 [1]  1  2  3  4  5  6  7  8  9 10
> sort(x)
 [1]  1  2  3  4  5  6  7  8  9 10

relate order() and rank()

  • Order to rank: rank() = order(order())
    set.seed(1)
    x <- rnorm(5)
    order(x)
    # [1] 3 1 2 5 4
    rank(x)
    # [1] 2 3 1 5 4
    order(order(x))
    # [1] 2 3 1 5 4
    all(rank(x) == order(order(x)))
    # TRUE
  • Order to Rank method 2: rank(order()) = 1:n
    ord <- order(x)
    ranks <- integer(length(x))
    ranks[ord] <- seq_along(x)
    ranks
    # [1] 2 3 1 5 4
  • Rank to Order:
    ranks <- rank(x)
    ord <- order(ranks)
    ord
    # [1] 3 1 2 5 4

OS-dependent results on sorting string vector

Gene symbol case.

# mac: 
order(c("DC-UbP", "DC2")) # c(1,2)

# linux: 
order(c("DC-UbP", "DC2")) # c(2,1)

Affymetric id case.

# mac:
order(c("202800_at", "2028_s_at")) # [1] 2 1
sort(c("202800_at", "2028_s_at")) # [1] "2028_s_at" "202800_at"

# linux
order(c("202800_at", "2028_s_at")) # [1] 1 2
sort(c("202800_at", "2028_s_at")) # [1] "202800_at" "2028_s_at"

It does not matter if we include factor() on the character vector.

The difference is related to locale. See

# both mac and linux
stringr::str_order(c("202800_at", "2028_s_at")) # [1] 2 1
stringr::str_order(c("DC-UbP", "DC2")) # [1] 1 2

# Or setting the locale to "C"
Sys.setlocale("LC_ALL", "C"); sort(c("DC-UbP", "DC2"))
# Or
Sys.setlocale("LC_COLLATE", "C"); sort(c("DC-UbP", "DC2"))
# But not
Sys.setlocale("LC_ALL", "en_US.UTF-8"); sort(c("DC-UbP", "DC2"))

unique()

It seems it does not sort. ?unique.

# mac & linux
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"

do.call

do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it.

The do.call() function in R: Unlocking Efficiency and Flexibility

Below are some examples from the help.

  • Usage
do.call(what, args, quote = FALSE, envir = parent.frame())
# what: either a function or a non-empty character string naming the function to be called.
# args: a list of arguments to the function call. The names attribute of args gives the argument names.
# quote: a logical value indicating whether to quote the arguments.
# envir: an environment within which to evaluate the call. This will be most useful
#        if what is a character string and the arguments are symbols or quoted expressions.
  • do.call() is similar to lapply() but not the same. It seems do.call() can make a simple function vectorized.
> do.call("complex", list(imag = 1:3))
[1] 0+1i 0+2i 0+3i
> lapply(list(imag = 1:3), complex)
$imag
[1] 0+0i
> complex(imag=1:3)
[1] 0+1i 0+2i 0+3i
> do.call(function(x) x+1, list(1:3))
[1] 2 3 4
  • Applying do.call with Multiple Arguments
> do.call("sum", list(c(1,2,3,NA), na.rm = TRUE))
[1] 6
> do.call("sum", list(c(1,2,3,NA) ))
[1] NA
> tmp <- expand.grid(letters[1:2], 1:3, c("+", "-"))
> length(tmp)
[1] 3
> tmp[1:4,]
  Var1 Var2 Var3
1    a    1    +
2    b    1    +
3    a    2    +
4    b    2    +
> c(tmp, sep = "")
$Var1
 [1] a b a b a b a b a b a b
Levels: a b

$Var2
 [1] 1 1 2 2 3 3 1 1 2 2 3 3

$Var3
 [1] + + + + + + - - - - - -
Levels: + -

$sep
[1] ""
> do.call("paste", c(tmp, sep = ""))
 [1] "a1+" "b1+" "a2+" "b2+" "a3+" "b3+" "a1-" "b1-" "a2-" "b2-" "a3-"
[12] "b3-"
  • environment and quote arguments.
> A <- 2
> f <- function(x) print(x^2)
> env <- new.env()
> assign("A", 10, envir = env)
> assign("f", f, envir = env)
> f <- function(x) print(x)
> f(A)   
[1] 2
> do.call("f", list(A))
[1] 2
> do.call("f", list(A), envir = env)  
[1] 4
> do.call(f, list(A), envir = env)   
[1] 2                       # Why?

> eval(call("f", A))                      
[1] 2
> eval(call("f", quote(A)))               
[1] 2
> eval(call("f", A), envir = env)         
[1] 4
> eval(call("f", quote(A)), envir = env)  
[1] 100
> foo <- function(a=1, b=2, ...) { 
         list(arg=do.call(c, as.list(match.call())[-1])) 
  }
> foo()
$arg
NULL
> foo(a=1)
$arg
a 
1 
> foo(a=1, b=2, c=3)
$arg
a b c 
1 2 3 
  • do.call() + switch(). See an example from Seurat::NormalizeData.
do.call(
   what = switch(
     EXPR = margin,
     '1' = 'rbind',
     '2' = 'cbind',
     stop("'margin' must be 1 or 2")
   ),
   args = normalized.data
)
switch('a', 'a' = rnorm(3), 'b'=rnorm(4)) # switch returns a value
do.call(switch('a', 'a' = 'rnorm', 'b'='rexp'), args=list(n=4)) # switch returns a function
  • The function we want to call is a string that may change: glmnet
# Suppose we want to call cv.glmnet or cv.coxnet or cv.lognet or cv.elnet .... depending on the case
fun = paste("cv", subclass, sep = ".")
cvstuff = do.call(fun, list(predmat,y,type.measure,weights,foldid,grouped))

expand.grid, mapply, vapply

A faster way to generate combinations for mapply and vapply

do.call vs mapply

  • do.call() is doing what mapply() does but do.call() uses a list instead of multiple arguments. So do.call() more close to base::Map() function.
> mapply(paste, tmp[1], tmp[2], tmp[3], sep = "")
      Var1 
 [1,] "a1+"
 [2,] "b1+"
 [3,] "a2+"
 [4,] "b2+"
 [5,] "a3+"
 [6,] "b3+"
 [7,] "a1-"
 [8,] "b1-"
 [9,] "a2-"
[10,] "b2-"
[11,] "a3-"
[12,] "b3-"
# It does not work if we do not explicitly specify the arguments in mapply()
> mapply(paste, tmp, sep = "")
      Var1 Var2 Var3
 [1,] "a"  "1"  "+" 
 [2,] "b"  "1"  "+" 
 [3,] "a"  "2"  "+" 
 [4,] "b"  "2"  "+" 
 [5,] "a"  "3"  "+" 
 [6,] "b"  "3"  "+" 
 [7,] "a"  "1"  "-" 
 [8,] "b"  "1"  "-" 
 [9,] "a"  "2"  "-" 
[10,] "b"  "2"  "-" 
[11,] "a"  "3"  "-" 
[12,] "b"  "3"  "-" 
set.seed(1)
mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1))
# [1] 1.326108 9.885284
set.seed(1)
x <- replicate(1000, mapply(rweibull, 1, c(1, 10), MoreArgs=list(n=1)))
dim(x) # [1]  2 1000
rowMeans(x)
# [1]  1.032209 10.104131
set.seed(1); Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10))
# [1] 1.326108 9.885284
set.seed(1); x <- replicate(1000, Vectorize(rweibull)(n=1, shape=1, scale=c(1, 10)))

do.call vs lapply

What's the difference between lapply and do.call? It seems to me the best usage is combining both functions: do.call(..., lapply())

  • lapply returns a list of the same length as X, each element of which is the result of applying FUN to the corresponding element of X.
  • do.call constructs and executes a function call from a name or a function and a list of arguments to be passed to it. It is widely used, for example, to assemble lists into simpler structures (often with rbind or cbind).
  • Map applies a function to the corresponding elements of given vectors... Map is a simple wrapper to mapply which does not attempt to simplify the result, similar to Common Lisp's mapcar (with arguments being recycled, however). Future versions may allow some control of the result type.
> lapply(iris, class) # same as Map(class, iris)
$Sepal.Length
[1] "numeric"

$Sepal.Width
[1] "numeric"

$Petal.Length
[1] "numeric"

$Petal.Width
[1] "numeric"

$Species
[1] "factor"

> x <- lapply(iris, class)
> do.call(c, x)
Sepal.Length  Sepal.Width Petal.Length  Petal.Width      Species 
   "numeric"    "numeric"    "numeric"    "numeric"     "factor" 

https://stackoverflow.com/a/10801902

  • lapply applies a function over a list. So there will be several function calls.
  • do.call calls a function with a list of arguments (... argument) such as c() or rbind()/cbind() or sum or order or "[" or paste. So there is only one function call.
> X <- list(1:3,4:6,7:9)
> lapply(X,mean)
1
[1] 2

2
[1] 5

3
[1] 8
> do.call(sum, X)
[1] 45
> sum(c(1,2,3), c(4,5,6), c(7,8,9))
[1] 45
> do.call(mean, X) # Error
> do.call(rbind,X)
     [,1] [,2] [,3]
[1,]    1    2    3
[2,]    4    5    6
[3,]    7    8    9
> lapply(X,rbind)
1
     [,1] [,2] [,3]
[1,]    1    2    3

2
     [,1] [,2] [,3]
[1,]    4    5    6

3
     [,1] [,2] [,3]
[1,]    7    8    9
> mapply(mean, X, trim=c(0,0.5,0.1))
[1] 2 5 8
> mapply(mean, X) 
[1] 2 5 8

Below is a good example to show the difference of lapply() and do.call() - Generating Random Strings.

> set.seed(1)
> x <- replicate(2, sample(LETTERS, 4), FALSE)
> x
1
[1] "Y" "D" "G" "A"

2
[1] "B" "W" "K" "N"

> lapply(x, paste0)
1
[1] "Y" "D" "G" "A"

2
[1] "B" "W" "K" "N"

> lapply(x, paste0, collapse= "")
1
[1] "YDGA"

2
[1] "BWKN"

> do.call(paste0, x)
[1] "YB" "DW" "GK" "AN"

do.call + rbind + lapply

Lots of examples. See for example this one for creating a data frame from a vector.

x <- readLines(textConnection("---CLUSTER 1 ---
 3
 4
 5
 6
 ---CLUSTER 2 ---
 9
 10
 8
 11"))

 # create a list of where the 'clusters' are
 clust <- c(grep("CLUSTER", x), length(x) + 1L)

 # get size of each cluster
 clustSize <- diff(clust) - 1L

 # get cluster number
 clustNum <- gsub("[^0-9]+", "", x[grep("CLUSTER", x)])

 result <- do.call(rbind, lapply(seq(length(clustNum)), function(.cl){
     cbind(Object = x[seq(clust[.cl] + 1L, length = clustSize[.cl])]
         , Cluster = .cl
         )
     }))

 result

     Object Cluster
[1,] "3"    "1"
[2,] "4"    "1"
[3,] "5"    "1"
[4,] "6"    "1"
[5,] "9"    "2"
[6,] "10"   "2"
[7,] "8"    "2"
[8,] "11"   "2"

A 2nd example is to sort a data frame by using do.call(order, list()).

Another example is to reproduce aggregate(). aggregate() = do.call() + by().

attach(mtcars)
do.call(rbind, by(mtcars, list(cyl, vs), colMeans))
# the above approach give the same result as the following
# except it does not have an extra Group.x columns
aggregate(mtcars, list(cyl, vs), FUN=mean)

Run examples

When we call help(FUN), it shows the document in the browser. The browser will show

example(FUN, package = "XXX") was run in the console
To view output in the browser, the knitr package must be installed

How to get examples from help file, example()

Code examples in the R package manuals:

# How to run all examples from a man page
example(within)

# How to check your examples?
devtools::run_examples() 
testthat::test_examples()

See this post. Method 1:

example(acf, give.lines=TRUE)

Method 2:

Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)

"[" and "[[" with the sapply() function

Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.

sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)

is the same as

sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])

Dealing with dates

  • Simple examples
    dates <- c("January 15, 2023", "December 31, 1999")
    date_objects <- as.Date(dates, format = "%B %d, %Y") # format is for the input
    # [1] "2023-01-15" "1999-12-31"
  • Find difference
    # Convert the dates to Date objects
    date1 <- as.Date("6/29/21", format="%m/%d/%y")
    date2 <- as.Date("11/9/21", format="%m/%d/%y")
    
    # Calculate the difference in days
    diff_days <- as.numeric(difftime(date2, date1, units="days")) # 133
    # In months
    diff_days / (365.25/12)  # 4.36961   
    
    # OR using the lubridate package
    library(lubridate)
    # Convert the dates to Date objects
    date1 <- mdy("6/29/21")
    date2 <- mdy("11/9/21")
    interval(date1, date2) %/% months(1)
  • http://cran.r-project.org/web/packages/lubridate/vignettes/lubridate.html
    d1 = date()
    class(d1) # "character"
    d2 = Sys.Date()
    class(d2) # "Date"
    
    format(d2, "%a %b %d")
    
    library(lubridate); ymd("20140108") # "2014-01-08 UTC"
    mdy("08/04/2013") # "2013-08-04 UTC"
    dmy("03-04-2013") # "2013-04-03 UTC"
    ymd_hms("2011-08-03 10:15:03") # "2011-08-03 10:15:03 UTC"
    ymd_hms("2011-08-03 10:15:03", tz="Pacific/Auckland") 
    # "2011-08-03 10:15:03 NZST"
    ?Sys.timezone
    x = dmy(c("1jan2013", "2jan2013", "31mar2013", "30jul2013"))
    wday(x[1]) # 3
    wday(x[1], label=TRUE) # Tues
  • http://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/
  • http://rpubs.com/seandavi/GEOMetadbSurvey2014
  • We want our dates and times as class "Date" or the class "POSIXct", "POSIXlt". For more information type ?POSIXlt.
  • anytime package
  • weeks to Christmas difftime(as.Date(“2019-12-25”), Sys.Date(), units =“weeks”)
  • A Comprehensive Introduction to Handling Date & Time in R 2020
  • Working with Dates and Times Pt 1
    • Three major functions: as.Date(), as.POSIXct(), and as.POSIXlt().
    • POSIXct is a class in R that represents date-time data. The ct stands for “calendar time” and it represents the (signed) number of seconds since the beginning of 1970 as a numeric vector1. It stores date time as integer.
    • POSIXlt is a class in R that represents date-time data. It stands for “local time” and is a list with components as integer vectors, which can represent a vector of broken-down times. It stores date time as list:sec, min, hour, mday, mon, year, wday, yday, isdst, zone, gmtoff.
  • R lubridate: How To Efficiently Work With Dates and Times in R 2023

Nonstandard/non-standard evaluation, deparse/substitute and scoping

f <- function(x) {
  substitute(x)
}
f(1:10)
# 1:10
class(f(1:10)) # or mode()
# [1] "call"
g <- function(x) deparse(substitute(x))
g(1:10)
# [1] "1:10"
class(g(1:10)) # or mode()
# [1] "character"
  • quote(expr) - similar to substitute() but do nothing?? noquote - print character strings without quotes
mode(quote(1:10))
# [1] "call"
  • eval(expr, envir), evalq(expr, envir) - eval evaluates its first argument in the current scope before passing it to the evaluator: evalq avoids this.
sample_df <- data.frame(a = 1:5, b = 5:1, c = c(5, 3, 1, 4, 1))

subset1 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x)
  x[r, ]
}
x <- 4
condition <- 4
subset1(sample_df, a== 4) # same as subset(sample_df, a >= 4)
subset1(sample_df, a== x) # WRONG!
subset1(sample_df, a == condition) # ERROR

subset2 <- function(x, condition) {
  condition_call <- substitute(condition)
  r <- eval(condition_call, x, parent.frame())
  x[r, ]
}
subset2(sample_df, a == 4) # same as subset(sample_df, a >= 4)
subset2(sample_df, a == x) # 👌 
subset2(sample_df, a == condition) # 👍
  • deparse(expr) - turns unevaluated expressions into character strings. For example,
> deparse(args(lm))
[1] "function (formula, data, subset, weights, na.action, method = \"qr\", " 
[2] "    model = TRUE, x = FALSE, y = FALSE, qr = TRUE, singular.ok = TRUE, "
[3] "    contrasts = NULL, offset, ...) "                                    
[4] "NULL"     

> deparse(args(lm), width=20)
[1] "function (formula, data, "        "    subset, weights, "           
[3] "    na.action, method = \"qr\", " "    model = TRUE, x = FALSE, "   
[5] "    y = FALSE, qr = TRUE, "       "    singular.ok = TRUE, "        
[7] "    contrasts = NULL, "           "    offset, ...) "               
[9] "NULL"

Following is another example. Assume we have a bunch of functions (f1, f2, ...; each function implements a different algorithm) with same input arguments format (eg a1, a2). We like to run these function on the same data (to compare their performance).

f1 <- function(x) x+1; f2 <- function(x) x+2; f3 <- function(x) x+3

f1(1:3)
f2(1:3)
f3(1:3)

# Or
myfun <- function(f, a) {
    eval(parse(text = f))(a)
}
myfun("f1", 1:3)
myfun("f2", 1:3)
myfun("f3", 1:3)

# Or with lapply
method <- c("f1", "f2", "f3")
res <- lapply(method, function(M) {
                    Mres <- eval(parse(text = M))(1:3)
                    return(Mres)
})
names(res) <- method

library() accept both quoted and unquoted strings

How can library() accept both quoted and unquoted strings. The key lines are

  if (!character.only) 
     package <- as.character(substitute(package))

Lexical scoping

The ‘…’ argument

Functions

Function argument

Argument matching from R Language Definition manual.

Argument matching is augmented by the functions

Access to the partial matching algorithm used by R is via pmatch.

Check function arguments

Checking the inputs of your R functions: match.arg() , stopifnot()

stopifnot(): function argument sanity check

  • stopifnot(). stopifnot is a quick way to check multiple conditions on the input. so for instance. The code stops when either of the three conditions are not satisfied. However, it doesn't produce pretty error messages.
    stopifnot(condition1, condition2, ...)
    
  • Mining R 4.0.0 Changelog for Nuggets of Gold

Lazy evaluation in R functions arguments

R function arguments are lazy — they’re only evaluated if they’re actually used.

  • Example 1. By default, R function arguments are lazy.
f <- function(x) {
  999
}
f(stop("This is an error!"))
#> [1] 999
  • Example 2. If you want to ensure that an argument is evaluated you can use force().
add <- function(x) {
  force(x)
  function(y) x + y
}
adders2 <- lapply(1:10, add)
adders2[[1]](10)
#> [1] 11
adders2[[10]](10)
#> [1] 20
  • Example 3. Default arguments are evaluated inside the function.
f <- function(x = ls()) {
  a <- 1
  x
}

# ls() evaluated inside f:
f()
# [1] "a" "x"

# ls() evaluated in global environment:
f(ls())
# [1] "add"    "adders" "f" 
  • Example 4. Laziness is useful in if statements — the second statement below will be evaluated only if the first is true.
x <- NULL
if (!is.null(x) && x > 0) {

}

Use of functions as arguments

Just Quickly: The unexpected use of functions as arguments

body()

Remove top axis title base plot

Return functions in R

anonymous function

In R, the main difference between a lambda function (also known as an anonymous function) and a regular function is that a lambda function is defined without a name, while a regular function is defined with a name.

  • See Tidyverse page
  • But defining functions to use them only once is kind of overkill. That's why you can use so-called anonymous functions in R. For example, lapply(list(1,2,3), function(x) { x * x })
  • you can use lambda functions with many other functions in R that take a function as an argument. Some examples include sapply, apply, vapply, mapply, Map, Reduce, Filter, and Find. These functions all work in a similar way to lapply by applying a function to elements of a list or vector.
    Reduce(function(x, y) x*y, list(1, 2, 3, 4)) # 24
    
  • purrr anonymous function
  • The new pipe and anonymous function syntax in R 4.1.0
  • Functional programming from Advanced R
  • What are anonymous functions in R.
    > (function(x) x * x)(3)
    [1] 9
    > (\(x) x * x)(3)
    [1] 9

Backtick sign, infix/prefix/postfix operators

The backtick sign ` (not the single quote) refers to functions or variables that have otherwise reserved or illegal names; e.g. '&&', '+', '(', 'for', 'if', etc. See some examples in Advanced R and What do backticks do in R?.

iris %>%  `[[`("Species")

infix operator.

1 + 2    # infix
+ 1 2    # prefix
1 2 +    # postfix

Use with functions like sapply, e.g. sapply(1:5, `+`, 3) .

Error handling and exceptions, tryCatch(), stop(), warning() and message()

  • http://adv-r.had.co.nz/Exceptions-Debugging.html
  • Catch Me If You Can: Exception Handling in R
  • Temporarily disable warning messages
    # Method1: 
    suppressWarnings(expr)
    
    # Method 2:
    <pre>
    defaultW <- getOption("warn") 
    options(warn = -1) 
    [YOUR CODE] 
    options(warn = defaultW)
    
  • try() allows execution to continue even after an error has occurred. You can suppress the message with try(..., silent = TRUE).
    out <- try({
      a <- 1
      b <- "x"
      a + b
    })
    
    elements <- list(1:10, c(-1, 10), c(T, F), letters)
    results <- lapply(elements, log)
    is.error <- function(x) inherits(x, "try-error")
    succeeded <- !sapply(results, is.error)
    
  • tryCatch(): With tryCatch() you map conditions to handlers (like switch()), named functions that are called with the condition as an input. Note that try() is a simplified version of tryCatch().
    tryCatch(expr, ..., finally)
    
    show_condition <- function(code) {
      tryCatch(code,
        error = function(c) "error",
        warning = function(c) "warning",
        message = function(c) "message"
      )
    }
    show_condition(stop("!"))
    #> [1] "error"
    show_condition(warning("?!"))
    #> [1] "warning"
    show_condition(message("?"))
    #> [1] "message"
    show_condition(10)
    #> [1] 10
    

    Below is another snippet from available.packages() function,

    z <- tryCatch(download.file(....), error = identity)
    if (!inherits(z, "error")) STATEMENTS
    
  • The return class from tryCatch() may not be fixed.
    result <- tryCatch({
      # Code that might generate an error or warning
      log(99)
    }, warning = function(w) {
      # Code to handle warnings
      print(paste("Warning:", w))
    }, error = function(e) {
      # Code to handle errors
      print(paste("Error:", e))
    }, finally = {
      # Code to always run, regardless of whether an error or warning occurred
      print("Finished")
    })   
    # character type. But if we remove 'finally', it will be numeric.
    
  • Capture message, warnings and errors from a R function

suppressMessages()

suppressMessages(expression)

List data type

Create an empty list

out <- vector("list", length=3L) # OR out <- list()
for(j in 1:3) out[[j]] <- myfun(j)

outlist <- as.list(seq(nfolds))

Nested list of data frames

An array can only hold data of a single type. read.csv() returns a data frame, which can contain both numerical and character data.

res <- vector("list", 3) 
names(res) <- paste0("m", 1:3)
for (i in seq_along(res)) {
  res[[i]] <- vector("list", 2)  # second-level list with 2 elements
  names(res[[i]]) <- c("fc", "pre")
}

res[["m1"]][["fc"]] <- read.csv()

head(res$m1$fc) # Same as res[["m1"]][["fc"]]

Using $ in R on a List

How to Use Dollar Sign ($) Operator in R

Calling a function given a list of arguments

> args <- list(c(1:10, NA, NA), na.rm = TRUE)
> do.call(mean, args)
[1] 5.5
> mean(c(1:10, NA, NA), na.rm = TRUE)
[1] 5.5

Descend recursively through lists

x[[c(5,3)]] is the same as x[[5]][[3]]. See ?Extract.

Avoid if-else or switch

?plot.stepfun.

y0 <- c(1,2,4,3)
sfun0  <- stepfun(1:3, y0, f = 0)
sfun.2 <- stepfun(1:3, y0, f = .2)
sfun1  <- stepfun(1:3, y0, right = TRUE)

tt <- seq(0, 3, by = 0.1)
op <- par(mfrow = c(2,2))
plot(sfun0); plot(sfun0, xval = tt, add = TRUE, col.hor = "bisque")
plot(sfun.2);plot(sfun.2, xval = tt, add = TRUE, col = "orange") # all colors
plot(sfun1);lines(sfun1, xval = tt, col.hor = "coral")
##-- This is  revealing :
plot(sfun0, verticals = FALSE,
     main = "stepfun(x, y0, f=f)  for f = 0, .2, 1")

for(i in 1:3)
  lines(list(sfun0, sfun.2, stepfun(1:3, y0, f = 1))[[i]], col = i)
legend(2.5, 1.9, paste("f =", c(0, 0.2, 1)), col = 1:3, lty = 1, y.intersp = 1)

par(op)

File:StepfunExample.svg

Open a new Window device

X11() or dev.new()

par()

?par

text size (cex) and font size on main, lab & axis

Examples (default is 1 for each of them):

  • cex.main=0.9
  • cex.sub
  • cex.lab=0.8, font.lab=2 (x/y axis labels)
  • cex.axis=0.8, font.axis=2 (axis/tick text/labels)
  • col.axis="grey50"

An quick example to increase font size (cex.lab, cex.axis, cex.main) and line width (lwd) in a line plot and cex & lwd in the legend.

plot(x=x$mids, y=x$density, type="l", 
     xlab="p-value", ylab="Density", lwd=2, 
     cex.lab=1.5, cex.axis=1.5, 
     cex.main=1.5, main = "")
lines(y$mids, y$density, lty=2, pwd=2)
lines(z$mids, z$density, lty=3, pwd=2)
legend('topright',legend = c('Method A','Method B','Method C'),
       lty=c(2,1,3), lwd=c(2,2,2), cex = 1.5, xjust = 0.5, yjust = 0.5)

ggplot2 case (default font size is 11 points):

  • plot.title
  • plot.subtitle
  • axis.title.x, axis.title.y: (x/y axis labels)
  • axis.text.x & axis.text.y: (axis/tick text/labels)
ggplot(df, aes(x, y)) +
  geom_point() +
  labs(title = "Title", subtitle = "Subtitle", x = "X-axis", y = "Y-axis") +
  theme(plot.title = element_text(size = 20),
        plot.subtitle = element_text(size = 15),
        axis.title.x = element_text(size = 15),
        axis.title.y = element_text(size = 15),
        axis.text.x = element_text(size = 10),
        axis.text.y = element_text(size = 10))

Default font

layout

reset the settings

op <- par(mfrow=c(2,1), mar = c(5,7,4,2) + 0.1) 
....
par(op) # mfrow=c(1,1), mar = c(5,4,4,2) + .1

mtext (margin text) vs title

mgp (axis tick label locations or axis title)

  1. The margin line (in ‘mex’ units) for the axis title, axis labels and axis line. Note that ‘mgp[1]’ affects the axis ‘title’ whereas ‘mgp[2:3]’ affect tick mark labels. The default is ‘c(3, 1, 0)’. If we like to make the axis labels closer to an axis, we can use mgp=c(1.5, .5, 0) for example.
    • the default is c(3,1,0) which specify the margin line for the axis title, axis labels and axis line.
    • the axis title is drawn in the fourth line of the margin starting from the plot region, the axis labels are drawn in the second line and the axis line itself is the first line.
  2. Setting graph margins in R using the par() function and lots of cow milk
  3. Move Axis Label Closer to Plot in Base R (2 Examples)
  4. http://rfunction.com/archives/1302 mgp – A numeric vector of length 3, which sets the axis label locations relative to the edge of the inner plot window. The first value represents the location the labels/axis title (i.e. xlab and ylab in plot), the second the tick-mark labels, and third the tick marks. The default is c(3, 1, 0).

move axis title closer to axis

title(ylab="Within-cluster variance", line=0, 
      cex.lab=1.2, family="Calibri Light")

pch and point shapes

File:R pch.png

See here.

  • Full circle: pch=16
  • Display all possibilities: ggpubr::show_point_shapes()

lty (line type)

File:R lty.png

Line types in R: Ultimate Guide For R Baseplot and ggplot

See here.

ggpubr::show_line_types()

las (label style)

0: The default, parallel to the axis

1: Always horizontal boxplot(y~x, las=1)

2: Perpendicular to the axis

3: Always vertical

oma (outer margin), xpd, common title for two plots, 3 types of regions, multi-panel plots

no.readonly

R语言里par(no.readonly=TURE)括号里面这个参数什么意思?, R-par()

Non-standard fonts in postscript and pdf graphics

https://cran.r-project.org/doc/Rnews/Rnews_2006-2.pdf#page=41


NULL, NA, NaN, Inf

https://tomaztsql.wordpress.com/2018/07/04/r-null-values-null-na-nan-inf/

save()/load() vs saveRDS()/readRDS() vs dput()/dget() vs dump()/source()

  1. saveRDS() can only save one R object while save() does not have this constraint.
  2. saveRDS() doesn’t save the both the object and its name it just saves a representation of the object. As a result, the saved object can be loaded into a named object within R that is different from the name it had when originally serialized. See this post.
x <- 5
saveRDS(x, "myfile.rds")
x2 <- readRDS("myfile.rds")
identical(mod, mod2, ignore.environment = TRUE)

dput: Writes an ASCII text representation of an R object. The object name is not written (unlike dump).

$ data(pbc, package = "survival")
$ names(pbc)
$ dput(names(pbc))
c("id", "time", "status", "trt", "age", "sex", "ascites", "hepato", 
"spiders", "edema", "bili", "chol", "albumin", "copper", "alk.phos", 
"ast", "trig", "platelet", "protime", "stage")

> iris2 <- iris[1:2, ]
> dput(iris2)
structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5, 
3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L, 
1L), .Label = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")

User 'verbose = TRUE' in load()

When we use load(), it is helpful to add 'verbose =TRUE' to see what objects get loaded.

What are RDS files anyways

Archive Existing RDS Files

==, all.equal(), identical()

  • ==: exact match
  • all.equal: compare R objects x and y testing ‘near equality’
  • identical: The safe and reliable way to test two objects for being exactly equal.
x <- 1.0; y <- 0.99999999999
all.equal(x, y)
# [1] TRUE
identical(x, y)
# [1] FALSE

Be careful about using "==" to return an index of matches in the case of data with missing values.

R> c(1,2,NA)[c(1,2,NA) == 1]
[1]  1 NA
R> c(1,2,NA)[which(c(1,2,NA) == 1)]
[1] 1

See also the testhat package.

I found a case when I compare two objects where 1 is generated in Linux and the other is generated in macOS that identical() gives FALSE but all.equal() returns TRUE. The difference has a magnitude only e-17.

waldo

diffobj: Compare/Diff R Objects

https://cran.r-project.org/web/packages/diffobj/index.html

testthat

tinytest

tinytest: Lightweight but Feature Complete Unit Testing Framework

ttdo adds support of the 'diffobj' package for 'diff'-style comparison of R objects.

Numerical Pitfall

Numerical pitfalls in computing variance

.1 - .3/3
## [1] 0.00000000000000001388

Sys.getpid()

This can be used to monitor R process memory usage or stop the R process. See this post.

Sys.getenv() & make the script more portable

Replace all the secrets from the script and replace them with Sys.getenv("secretname"). You can save the secrets in an .Renviron file next to the script in the same project.

$ for v in 1 2; do MY=$v Rscript -e "Sys.getenv('MY')"; done
[1] "1"
[1] "2"
$ echo $MY
2

How to write R codes

  • Code smells and feels from R Consortium
    • write simple conditions,
    • handle class properly,
    • return and exit early,
    • polymorphism,
    • switch() [e.g., switch(var, value1=out1, value2=out2, value3=out3). Several examples in glmnet ]
    • case_when(),
    • %||%.
  • 5 Tips for Writing Clean R Code – Leave Your Code Reviewer Commentless
    • Comments
    • Strings
    • Loops
    • Code Sharing
    • Good Programming Practices

How to debug an R code

Debug R

Locale bug (grep did not handle UTF-8 properly PR#16264)

https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=16264

Path length in dir.create() (PR#17206)

https://bugs.r-project.org/bugzilla3/show_bug.cgi?id=17206 (Windows only)

install.package() error, R_LIBS_USER is empty in R 3.4.1 & .libPaths()

R_LIBS_USER=${R_LIBS_USER-'~/R/x86_64-pc-linux-gnu-library/3.4'}
R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4"

On Mac & R 3.4.0 (it's fine)

> Sys.getenv("R_LIBS_USER")
[1] "~/Library/R/3.4/library"
> .libPaths()
[1] "/Library/Frameworks/R.framework/Versions/3.4/Resources/library"

On Linux & R 3.3.1 (ARM)

> Sys.getenv("R_LIBS_USER")
[1] "~/R/armv7l-unknown-linux-gnueabihf-library/3.3"
> .libPaths()
[1] "/home/$USER/R/armv7l-unknown-linux-gnueabihf-library/3.3"
[2] "/usr/local/lib/R/library"

On Linux & R 3.4.1 (*Problematic*)

> Sys.getenv("R_LIBS_USER")
[1] ""
> .libPaths()
[1] "/usr/local/lib/R/site-library" "/usr/lib/R/site-library"
[3] "/usr/lib/R/library"

I need to specify the lib parameter when I use the install.packages command.

> install.packages("devtools", "~/R/x86_64-pc-linux-gnu-library/3.4")
> library(devtools)
Error in library(devtools) : there is no package called 'devtools'

# Specify lib.loc parameter will not help with the dependency package
> library(devtools, lib.loc = "~/R/x86_64-pc-linux-gnu-library/3.4")
Error: package or namespace load failed for 'devtools':
 .onLoad failed in loadNamespace() for 'devtools', details:
  call: loadNamespace(name)
  error: there is no package called 'withr'

# A solution is to redefine .libPaths
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works

A better solution is to specify R_LIBS_USER in ~/.Renviron file or ~/.bash_profile; see ?Startup.

Using external data from within another package

https://logfc.wordpress.com/2017/03/02/using-external-data-from-within-another-package/

How to run R scripts from the command line

https://support.rstudio.com/hc/en-us/articles/218012917-How-to-run-R-scripts-from-the-command-line

How to exit a sourced R script

Decimal point & decimal comma

Countries using Arabic numerals with decimal comma (Austria, Belgium, Brazil France, Germany, Netherlands, Norway, South Africa, Spain, Sweden, ...) https://en.wikipedia.org/wiki/Decimal_mark

setting seed locally (not globally) in R

https://stackoverflow.com/questions/14324096/setting-seed-locally-not-globally-in-r

R's internal C API

https://github.com/hadley/r-internals

cleancall package for C resource cleanup

Resource Cleanup in C and the R API

Random number generator

#include <R.h>

void myunif(){
  GetRNGstate();
  double u = unif_rand();
  PutRNGstate();
  Rprintf("%f\n",u);
}
$ R CMD SHLIB r_rand.c
$ R
R> dyn.load("r_rand.so")
R> set.seed(1)
R> .C("myunif")
0.265509
list()
R> .C("myunif")
0.372124
list()
R> set.seed(1)
R> .C("myunif")
0.265509
list()

Test For Randomness

Different results in Mac and Linux

Random numbers: multivariate normal

Why MASS::mvrnorm() gives different result on Mac and Linux/Windows?

The reason could be the covariance matrix decomposition - and that may be due to the LAPACK/BLAS libraries. See

rle() running length encoding

citation()

citation()
citation("MASS")
toBibtex(citation())

Notes on Citing R and R Packages with examples.

R not responding request to interrupt stop process

R not responding request to interrupt stop process. R is executing (for example) a C / C++ library call that doesn't provide R an opportunity to check for interrupts. It seems to match with the case I'm running (dist() function).

Monitor memory usage

  • x <- rnorm(2^27) will create an object of the size 1GB (2^27*8/2^20=1024 MB).
  • Windows: memory.size(max=TRUE)
  • Linux
    • RStudio: htop -p PID where PID is the process ID of /usr/lib/rstudio/bin/rsession, not /usr/lib/rstudio/bin/rstudio. This is obtained by running x <- rnorm(2*1e8). The object size can be obtained through print(object.size(x), units = "auto"). Note that 1e8*8/2^20 = 762.9395.
    • R: htop -p PID where PID is the process ID of /usr/lib/R/bin/exec/R. Alternatively, use htop -p `pgrep -f /usr/lib/R/bin/exec/R`
    • To find the peak memory usage grep VmPeak /proc/$PID/status
  • mem_used() function from pryr package. It is not correct or useful if I use it to check the value compared to the memory returned by jobload in biowulf. So I cannot use it to see the memory used in running mclapply().
  • peakRAM: Monitor the Total and Peak RAM Used by an Expression or Function
  • Error: protect () : protection stack overflow and ?Memory

References:

Monitor Data

Monitoring Data in R with the lumberjack Package

Pushover

Monitoring Website SSL/TLS Certificate Expiration Times with R, {openssl}, {pushoverr}, and {DT}

pushoverr

Resource

Books

  • Efficient R programming by Colin Gillespie and Robin Lovelace. It works to re-create the html version of the book if we follow their simple instruction in the Appendix. Note that pdf version has advantages of expected output (mathematical notations, tables) over the epub version.
    # R 3.4.1
    .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
    setwd("/tmp/efficientR/")
    bookdown::render_book("index.Rmd", output_format = "bookdown::pdf_book")
    # generated pdf file is located _book/_main.pdf
    
    bookdown::render_book("index.Rmd", output_format = "bookdown::epub_book")
    # generated epub file is located _book/_main.epub.
    # This cannot be done in RStudio ("parse_dt" not resolved from current namespace (lubridate))
    # but it is OK to run in an R terminal
    

Videos

Webinar

useR!

R consortium

https://www.youtube.com/channel/UC_R5smHVXRYGhZYDJsnXTwg/featured

Blogs, Tips, Socials, Communities

Bug Tracking System

https://bugs.r-project.org/bugzilla3/ and Search existing bug reports. Remember to select 'All' in the Status drop-down list.

Use sessionInfo().

License

Some Notes on GNU Licenses in R Packages

Why Dash uses the mit license (and not a copyleft gpl license)

Interview questions

  • Does R store matrices in column-major order or row-major order?
    • Matrices are stored in column-major order, which means that elements are arranged and accessed by columns. This is in contrast to languages like Python, where matrices (or arrays) are typically stored in row-major order.
  • Explain the difference between == and === in R. Provide an example to illustrate their use.
    • The == operator is used for testing equality of values in R. It returns TRUE if the values on the left and right sides are equal, otherwise FALSE. The === operator does not exist in base R.
  • What is the purpose of the apply() function in R? How does it differ from the for loop?
    • The apply() function in R is used to apply a function over the margins of an array or matrix. It is often used as an alternative to loops for applying a function to each row or column of a matrix.
  • Describe the concept of factors in R. How are they used in data manipulation and analysis?
    • Factors in R are used to represent categorical data. They are an essential data type for statistical modeling and analysis. Factors store both the unique values that occur in a dataset and the corresponding integer codes used to represent those values.
  • What is the significance
of the attach() and detach() functions in R? When should they be used?
    • A: The attach() function is used to add a data frame to the search path in R, making it easier to access variables within the data frame. The detach() function is used to remove a data frame from the search path, which can help avoid naming conflicts and reduce memory usage.
  • Explain the concept of vectorization in R. How does it impact the performance of R code?
    • Vectorization in R refers to the ability to apply operations to entire vectors or arrays at once, without needing to write explicit loops. This can significantly improve the performance of R code, as it allows operations to be performed in a more efficient, vectorized manner by taking advantage of R's underlying C code.
  • Describe the difference between data.frame and matrix in R. When would you use one over the other?
    • A data.frame in R is a two-dimensional structure that can store different types of data (e.g., numeric, character, factor) in its columns. It is similar to a table in a database.
    • A matrix in R is also a two-dimensional structure, but it can only store elements of the same data type. It is more like a mathematical matrix.
    • You would use a data.frame when you have heterogeneous data (i.e., different types of data) and need to work with it as a dataset. You would use a matrix when you have homogeneous data (i.e., the same type of data) and need to perform matrix operations.
  • What are the benefits of using the dplyr package in R for data manipulation? Provide an example of how you would use dplyr to filter a data frame.
    • The dplyr package provides a set of functions that make it easier to manipulate data frames in R.
    • It uses a syntax that is easy to read and understand, making complex data manipulations more intuitive.
    • To filter a data frame using dplyr, you can use the filter() function. For example, filter(df, column_name == value) would filter df to include only rows where column_name is equal to value.