R: Difference between revisions

From 太極
Jump to navigation Jump to search
Line 1: Line 1:
== Install Rtools for Windows users ==
= Install and upgrade R =
See http://goo.gl/gYh6C for a step-by-step instruction with screenshot.
[[Install_R|Here]]


My preferred way is not to check the option of setting PATH environment. But I manually add the followings to the PATH environment (based on Rtools v3.0)
== New release ==
* 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).  
 
== [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.
 
== rstudio.cloud ==
 
== [https://www.rdocumentation.org/ RDocumentation] ==
The interactive engine is based on [https://github.com/datacamp/datacamp-light DataCamp Light]
 
For example, [https://www.rdocumentation.org/packages/dplyr/versions/0.5.0/topics/tbl_df tbl_df] function from dplyr package.
 
The website [https://cdn.datacamp.com/dcl/standalone-example.html DataCamp] allows to run ''library()'' on the Script window. After that, we can use the packages on ''R Console''.
 
[http://documents.datacamp.com/default_r_packages.txt 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|R web applications]]
 
= Creating local repository for CRAN and Bioconductor =
[[R_repository|R repository]]
 
= Parallel Computing =
See [[R_parallel|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 =
* [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
 
== bigmemory, biganalytics, bigtabulate ==
 
== ff, ffbase ==
* 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
 
== biglm ==
 
== data.table ==
See [[Tidyverse#data.table|data.table]].
 
== disk.frame ==
[https://www.brodrigues.co/blog/2019-10-05-parallel_maxlik/ Split-apply-combine for Maximum Likelihood Estimation of a linear model]
 
== Apache arrow ==
* https://arrow.apache.org/docs/r/
* [https://www.infoworld.com/article/3637038/the-best-open-source-software-of-2021.html#slide17 The best open source software of 2021]
 
= Reproducible Research =
* http://cran.r-project.org/web/views/ReproducibleResearch.html
* [[Reproducible|Reproducible]]
 
== Reproducible Environments ==
https://rviews.rstudio.com/2019/04/22/reproducible-environments/
 
== 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]
 
== 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!
 
= 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]
 
== Rcpp ==
http://cran.r-project.org/web/packages/Rcpp/index.html. See more [[Rcpp|here]].
 
== RInside : embed R in C++ code ==
* http://dirk.eddelbuettel.com/code/rinside.html
* http://dirk.eddelbuettel.com/papers/rfinance2010_rcpp_rinside_tutorial_handout.pdf
 
=== 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 [http://www.youtube.com/watch?v=UQ8yKQcPTg0 Youtube]. I have tested this '''qtdensity''' example successfully using Qt 4.8.5.
# Follow the instruction [[#cairoDevice|cairoDevice]] to install required libraries for cairoDevice package and then cairoDevice itself.
# 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]]
 
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
<pre>
<pre>
c:\Rtools\bin;
cd ~/R/x86_64-pc-linux-gnu-library/3.0/RInside/examples/wt
c:\Rtools\gcc-4.6.3\bin;
make
C:\Program Files\R\R-2.15.2\bin\i386;
sudo ./wtdensity --docroot . --http-address localhost --http-port 8080
</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]).


We can make our life easy by creating a file <Rcommand.bat> with the content (also useful if you have C:\cygwin\bin in your PATH although cygwin setup will not do it automatically for you.)
=== Windows 7 ===
 
To make RInside works on Windows OS, try the following
PS. I put <Rcommand.bat> under C:\Program Files\R folder. I create a shortcut called 'Rcmd' on desktop. I enter '''C:\Windows\System32\cmd.exe /K "Rcommand.bat"''' in the ''Target'' entry and
# 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''.
'''"C:\Program Files\R"''' in ''Start in'' entry.
# 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>
@echo off
@echo off
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\Rtools\bin;c:\Rtools\gcc-4.6.3\bin;%PATH%
set PATH=C:\Program Files\R\R-2.15.2\bin\i386;%PATH%
set PATH=C:\R\R-3.0.1\bin\i386;%PATH%
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_LIBS=`Rscript -e "Rcpp:::LdFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set PKG_CPPFLAGS=`Rscript -e "Rcpp:::CxxFlags()"`
set R_HOME=C:\R\R-3.0.1
echo Setting environment for using R
echo Setting environment for using R
cmd
cmd
</pre>
</pre>
So we can open the Command Prompt anywhere and run <Rcommand.bat> to get all environment variables ready! On Windows Vista, 7 and 8, we need to run it as administrator. OR we can change the security of the property so the current user can have an executive right.
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>


=== Compile and install an R package ===
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/[email protected]/msg04377.html
So the Qt and Wt web tool applications on Windows may or may not be possible.
 
== GUI ==
=== Qt and R ===
* 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://qtinterfaces.r-forge.r-project.org
 
== tkrplot ==
On Ubuntu, we need to install tk packages, such as by
<pre>
<pre>
cd C:\Documents and Settings\brb
sudo apt-get install tk-dev
wget http://www.bioconductor.org/packages/2.11/bioc/src/contrib/affxparser_1.30.2.tar.gz
C:\progra~1\r\r-2.15.2\bin\R CMD INSTALL --build affxparser_1.30.2.tar.gz
</pre>
</pre>


Helpful - check Chapter 6 of [http://cran.r-project.org/doc/manuals/r-release/R-admin.html R Installation and Administration]
== reticulate - Interface to 'Python' ==
[[Python#R_and_Python:_reticulate_package|Python -> reticulate]]
 
== Hadoop (eg ~100 terabytes) ==
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
 
* 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


=== Check/Upload to CRAN ===
=== [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.


http://win-builder.r-project.org/
=== 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


== Install R using binary package ==
== [http://cran.r-project.org/web/packages/XML/index.html XML] ==
=== Ubuntu/Debian ===
On Ubuntu, we need to install libxml2-dev before we can install XML package.
<pre>
<pre>
sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys E084DAB9
gksudo gedit /etc/apt/sources.list
# deb http://cran.fhcrc.org/bin/linux/ubuntu precise/
sudo apt-get update
sudo apt-get update
sudo apt-get install r-base
sudo apt-get install libxml2-dev
</pre>
</pre>


=== Redhat el6 ===
On CentOS,
It should be pretty easy to install via the EPEL:  http://fedoraproject.org/wiki/EPEL
<pre>
yum -y install libxml2 libxml2-devel
</pre>


Just follow the instructions to enable the EPEL and then from the CLI as root:
=== 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)


yum install R
# Read and parse HTML file
doc.html = htmlTreeParse('http://apiolaza.net/babel.html', useInternal = TRUE)


or via sudo:
# 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))


sudo yum install R
# Replace all by spaces
doc.text = gsub('\n', ' ', doc.text)


== Install R from source (ix86, x86_64 and arm platforms, Linux system) ==
# Join all the elements of the character vector into a single
# character string, separated by spaces
doc.text = paste(doc.text, collapse = ' ')
</pre>


=== Debian system (focus on arm architecture with notes from x86 system) ===
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.
==== Simplest configuration ====
{{Pre}}
On my debian system in Pogoplug (armv5) OR Raspberry Pi (armv6), I can compile R. See R's [http://cran.r-project.org/doc/manuals/R-admin.html#Installing-R-under-Unix_002dalikes admin manual]. If I don't need x11, I just need to install 2 required packages.
> 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"


* install gfortran: '''apt-get install gfortran'''
> # try picard
* install readline library: '''apt-get install libreadline5-dev''' (pogoplug), '''apt-get install libreadline6-dev''' (raspberry pi)
> 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"
</pre>
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").


Note: if I need x11, I should install
=== xmlview ===
* libx11 and libx11-devel, libXt, libXt-devel (for fedora)
* http://rud.is/b/2016/01/13/cobble-xpath-interactively-with-the-xmlview-package/
* '''libx11-dev''' (for debian) or '''xorg-dev''' (for pogoplug/raspberry pi)
and optional
* '''texinfo''' (to fix 'WARNING: you cannot build info or HTML versions of the R manuals')


Note that it is also safe to install required tools via (in ubuntu)
== RCurl ==
<pre>sudo apt-get install r-base-dev </pre>
On Ubuntu, we need to install the packages (the first one is for XML package that RCurl suggests)
See [[#Install r-base and r-base-dev]]
{{Pre}}
Or even better with
# Test on Ubuntu 14.04
<pre>sudo apt-get build-dep r-base</pre>
sudo apt-get install libxml2-dev
See [[#Install all dependencies for building R]]
sudo apt-get install libcurl4-openssl-dev
Since with the first approach, running ./configure still complains cannot x11 header/libs still missing. The second approach will pull in dependence like jdk, tcl, tex and more. The apt-get build-dep gave a more complete list than apt-get install r-base-dev for some reasons.
</pre>


[Arm architecture]I also run '''apt-get install readline-common'''. I don't know if this is necessary.
=== Scrape google scholar results ===
Since I don't need x11, I use the option in configure command. After running
https://github.com/tonybreyal/Blog-Reference-Functions/blob/master/R/googleScholarXScraper/googleScholarXScraper.R
 
No google ID is required
 
Seems not work
<pre>
<pre>
wget http://cran.r-project.org/src/base/R-2/R-2.15.2.tar.gz
Error in data.frame(footer = xpathLVApply(doc, xpath.base, "/font/span[@class='gs_fl']",  :
tar xzvf R-2.15.2.tar.gz
  arguments imply differing number of rows: 2, 0
cd R-2.15.2
./configure --with-x=no --enable-R-shlib
</pre>
</pre>
I got
 
=== [https://cran.r-project.org/web/packages/devtools/index.html devtools] ===
'''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
 
# 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
</pre>
 
[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.
 
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/hadley/httr httr] ===
httr imports curl, jsonlite, mime, openssl and R6 packages.
 
When I tried to install httr package, I got an error and some message:
<pre>
<pre>
R is now configured for armv5tel-unknown-linux-gnueabi
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).
 
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.


  Source directory:         .
[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)
  Installation directory:   /usr/local


  C compiler:               gcc -std=gnu99  -g -O2
=== [http://cran.r-project.org/web/packages/curl/ curl] ===
  Fortran 77 compiler:      gfortran  -g -O2
curl is independent of RCurl package.


  C++ compiler:             g++  -g -O2
* http://cran.r-project.org/web/packages/curl/vignettes/intro.html
  Fortran 90/95 compiler:    gfortran -g -O2
* https://www.opencpu.org/posts/curl-release-0-8/
   Obj-C compiler:
 
{{Pre}}
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>


  Interfaces supported:
=== [http://ropensci.org/packages/index.html rOpenSci] packages ===
  External libraries:        readline
'''rOpenSci''' contains packages that allow access to data repositories through the R statistical programming environment
  Additional capabilities:  NLS
  Options enabled:          shared R library, shared BLAS, R profiling


  Recommended packages:      yes
== [https://cran.r-project.org/web/packages/remotes/index.html 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).


configure: WARNING: you cannot build info or HTML versions of the R manuals
Example:
configure: WARNING: you cannot build PDF versions of the R manuals
{{Pre}}
configure: WARNING: you cannot build PDF versions of vignettes and help pages
# https://github.com/henrikbengtsson/matrixstats
configure: WARNING: I could not determine a browser
remotes::install_github('HenrikBengtsson/matrixStats@develop')
configure: WARNING: I could not determine a PDF viewer
</pre>
</pre>
PS 1. On my raspberry pi machine, it shows '''R is now configured for armv6l-unknown-linux-gnueabihf'''.


PS 2. On my x86 system, it shows
== DirichletMultinomial ==
On Ubuntu, we do
<pre>
<pre>
R is now configured for x86_64-unknown-linux-gnu
sudo apt-get install libgsl0-dev
</pre>


  Source directory:         .
== Create GUI ==
  Installation directory:    /usr/local
=== [http://cran.r-project.org/web/packages/gWidgets/index.html gWidgets] ===


  C compiler:               gcc -std=gnu99  -g -O2
== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ==
  Fortran 77 compiler:      gfortran  -g -O2
[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]


  C++ compiler:              g++  -g -O2
== json ==
  Fortran 90/95 compiler:    gfortran -g -O2
[[R_web#json|R web -> json]]
  Obj-C compiler:


  Interfaces supported:      X11, tcltk
== Map ==
  External libraries:       readline, lzma
=== [https://rstudio.github.io/leaflet/ leaflet] ===
  Additional capabilities:  PNG, JPEG, TIFF, NLS, cairo
* rstudio.github.io/leaflet/#installation-and-use
  Options enabled:           shared R library, shared BLAS, R profiling, Java
* https://metvurst.wordpress.com/2015/07/24/mapview-basic-interactive-viewing-of-spatial-data-in-r-6/


  Recommended packages:     yes
=== choroplethr ===
</pre>
* http://blog.revolutionanalytics.com/2014/01/easy-data-maps-with-r-the-choroplethr-package-.html
* http://www.arilamstein.com/blog/2015/06/25/learn-to-map-census-data-in-r/
* http://www.arilamstein.com/blog/2015/09/10/user-question-how-to-add-a-state-border-to-a-zip-code-map/


[arm] However, '''make''' gave errors for recommanded packages like KernSmooth, MASS, boot, class, cluster, codetools, foreign, lattice, mgcv, nlme, nnet, rpart, spatial, and survival. The error stems from
=== ggplot2 ===
'''gcc: SHLIB_LIBADD: No such file or directory'''. Note that I can get this error message even I try '''install.packages("MASS", type="source")'''. A suggested fix is [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=679180 here]; adding '''perl = TRUE''' in sub() call for two lines in '''src/library/tools/R/install.R''' file. However, I got another error '''shared object 'MASS.so' not found'''. See also http://ftp.debian.org/debian/pool/main/r/r-base/.
[https://randomjohn.github.io/r-maps-with-census-data/ How to make maps with Census data in R]


== [http://cran.r-project.org/web/packages/googleVis/index.html googleVis] ==
See an example from [[R#RJSONIO|RJSONIO]] above.


<pre>
== [https://cran.r-project.org/web/packages/googleAuthR/index.html googleAuthR] ==
make[1]: Entering directory `/mnt/usb/R-2.15.2/src/library/Recommended'
Create R functions that interact with OAuth2 Google APIs easily, with auto-refresh and Shiny compatibility.
make[2]: Entering directory `/mnt/usb/R-2.15.2/src/library/Recommended'
begin installing recommended package MASS
* installing *source* package 'MASS' ...
** libs
make[3]: Entering directory `/tmp/Rtmp4caBfg/R.INSTALL1d1244924c77/MASS/src'
gcc -std=gnu99 -I/mnt/usb/R-2.15.2/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c MASS.c -o MASS.o
gcc -std=gnu99 -I/mnt/usb/R-2.15.2/include -DNDEBUG  -I/usr/local/include    -fpic  -g -O2  -c lqs.c -o lqs.o
gcc -std=gnu99 -shared -L/usr/local/lib -o MASSSHLIB_EXT MASS.o lqs.o SHLIB_LIBADD -L/mnt/usb/R-2.15.2/lib -lR
gcc: SHLIB_LIBADD: No such file or directory
make[3]: *** [MASSSHLIB_EXT] Error 1
make[3]: Leaving directory `/tmp/Rtmp4caBfg/R.INSTALL1d1244924c77/MASS/src'
ERROR: compilation failed for package 'MASS'
* removing '/mnt/usb/R-2.15.2/library/MASS'
make[2]: *** [MASS.ts] Error 1
make[2]: Leaving directory `/mnt/usb/R-2.15.2/src/library/Recommended'
make[1]: *** [recommended-packages] Error 2
make[1]: Leaving directory `/mnt/usb/R-2.15.2/src/library/Recommended'
make: *** [stamp-recommended] Error 2
root@debian:/mnt/usb/R-2.15.2#
root@debian:/mnt/usb/R-2.15.2# bin/R


R version 2.15.2 (2012-10-26) -- "Trick or Treat"
== gtrendsR - Google Trends ==
Copyright (C) 2012 The R Foundation for Statistical Computing
* [http://blog.revolutionanalytics.com/2015/12/download-and-plot-google-trends-data-with-r.html Download and plot Google Trends data with R]
ISBN 3-900051-07-0
* [https://datascienceplus.com/analyzing-google-trends-data-in-r/ Analyzing Google Trends Data in R]
Platform: armv5tel-unknown-linux-gnueabi (32-bit)
* [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]


R is free software and comes with ABSOLUTELY NO WARRANTY.
== quantmod ==
You are welcome to redistribute it under certain conditions.
[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.
Type 'license()' or 'licence()' for distribution details.


R is a collaborative project with many contributors.
# Initial data downloading
Type 'contributors()' for more information and
# Update existing data
'citation()' on how to cite R or R packages in publications.
# Create a batch file


Type 'demo()' for some demos, 'help()' for on-line help, or
== [http://cran.r-project.org/web/packages/caret/index.html caret] ==
'help.start()' for an HTML browser interface to help.
* http://topepo.github.io/caret/index.html & https://github.com/topepo/caret/
Type 'q()' to quit R.
* https://www.r-project.org/conferences/useR-2013/Tutorials/kuhn/user_caret_2up.pdf
* https://github.com/cran/caret source code mirrored on github
* Cheatsheet https://www.rstudio.com/resources/cheatsheets/
* [https://daviddalpiaz.github.io/r4sl/the-caret-package.html Chapter 21 of "R for Statistical Learning"]


> library(MASS)
== Tool for connecting Excel with R ==
Error in library(MASS) : there is no package called 'MASS'
* https://bert-toolkit.com/
> library()
* [http://www.thertrader.com/2016/11/30/bert-a-newcomer-in-the-r-excel-connection/ BERT: a newcomer in the R Excel connection]
Packages in library '/mnt/usb/R-2.15.2/library':
* http://blog.revolutionanalytics.com/2018/08/how-to-use-r-with-excel.html


base                    The R Base Package
== write.table ==
compiler                The R Compiler Package
=== Output a named vector ===
datasets                The R Datasets Package
<pre>
grDevices              The R Graphics Devices and Support for Colours
vec <- c(a = 1, b = 2, c = 3)
                        and Fonts
write.csv(vec, file = "my_file.csv", quote = F)
graphics                The R Graphics Package
x = read.csv("my_file.csv", row.names = 1)
grid                    The Grid Graphics Package
vec2 <- x[, 1]
methods                Formal Methods and Classes
names(vec2) <- rownames(x)
parallel                Support for Parallel computation in R
all.equal(vec, vec2)
splines                Regression Spline Functions and Classes
 
stats                  The R Stats Package
# one liner: row names of a 'matrix' become the names of a vector
stats4                  Statistical Functions using S4 Classes
vec3 <- as.matrix(read.csv('my_file.csv', row.names = 1))[, 1]
tcltk                  Tcl/Tk Interface
all.equal(vec, vec3)
tools                  Tools for Package Development
utils                  The R Utils Package
> Sys.info()["machine"]
  machine
"armv5tel"
> gc()
        used (Mb) gc trigger (Mb) max used (Mb)
Ncells 170369  4.6    350000  9.4  350000  9.4
Vcells 163228  1.3    905753  7.0  784148  6.0
</pre>
</pre>
See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=679180
PS 3. The complete log of building R from source is in here [[File:Build_R_log.txt‎]]


==== Full configuration ====
=== Avoid leading empty column to header ===
[https://stackoverflow.com/a/2478624 write.table writes unwanted leading empty column to header when has rownames]
<pre>
<pre>
  Interfaces supported:      X11, tcltk
write.table(a, 'a.txt', col.names=NA)
  External libraries:        readline
  Additional capabilities:  PNG, JPEG, TIFF, NLS, cairo
  Options enabled:          shared R library, shared BLAS, R profiling, Java
</pre>
</pre>


=== Install r-base and r-base-dev ===
=== Add blank field AND column names in write.table ===
In fact, if we want to take a short cut, it seems OK to run the following statement to install r-base and all required components for building r-base and extra packages. Notice that for readline package, it installs 'libreadline6-dev' instead of 'libreadline5-dev' as I just did.
* '''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
 
=== read.delim(, row.names=1) and write.table(, row.names=TRUE) ===
[https://www.statology.org/read-delim-in-r/ How to Use read.delim Function in R]


Note that I did not touch /etc/apt/sources.list file so I don't know what version of R will be installed by this method.
Case 1: no row.names
<pre>
<pre>
root@debian:/mnt/usb/R-2.15.2# apt-get install r-base-dev
write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=FALSE)
Reading package lists... Done
my_df <- read.delim('my_data.txt')  # the rownames will be 1, 2, 3, ...
Building dependency tree
Reading state information... Done
The following package was automatically installed and is no longer required:
  libreadline5
Use 'apt-get autoremove' to remove them.
The following extra packages will be installed:
  defoma dpatch file fontconfig libblas-dev libblas3gf libbz2-dev libcairo2 libdatrie1
  libjpeg62-dev liblapack-dev liblapack3gf libnewt0.52 libpango1.0-0 libpango1.0-common
  libpaper-utils libpaper1 libpcre3-dev libpcrecpp0 libpixman-1-0 libpng12-dev libreadline-dev
  libreadline6-dev libthai-data libthai0 libxcb-render-util0 libxcb-render0 r-base-core unzip
  whiptail xdg-utils zip zlib1g-dev
Suggested packages:
  defoma-doc psfontmgr x-ttcidfont-conf dfontmgr curl ttf-japanese-gothic ttf-japanese-mincho
  ttf-thryomanes ttf-baekmuk ttf-arphic-gbsn00lp ttf-arphic-bsmi00lp ttf-arphic-gkai00mp
  ttf-arphic-bkai00mp ess r-doc-info r-doc-pdf r-mathlib r-base-html cdbs debhelper gvfs-bin
Recommended packages:
  libfont-freetype-perl fakeroot patchutils libfribidi0 r-recommended r-doc-html iceweasel
  www-browser x11-utils x11-xserver-utils shared-mime-info
The following packages will be REMOVED:
  libreadline5-dev
The following NEW packages will be installed:
  defoma dpatch file fontconfig libblas-dev libblas3gf libbz2-dev libcairo2 libdatrie1
  libjpeg62-dev liblapack-dev liblapack3gf libnewt0.52 libpango1.0-0 libpango1.0-common
  libpaper-utils libpaper1 libpcre3-dev libpcrecpp0 libpixman-1-0 libpng12-dev libreadline-dev
  libreadline6-dev libthai-data libthai0 libxcb-render-util0 libxcb-render0 r-base-core
  r-base-dev unzip whiptail xdg-utils zip zlib1g-dev
0 upgraded, 34 newly installed, 1 to remove and 0 not upgraded.
Need to get 22.9 MB of archives.
After this operation, 58.2 MB of additional disk space will be used.
Do you want to continue [Y/n]? n
</pre>
</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.
=== Install all dependencies for building R ===
This is a comprehensive list. This list is even larger than r-base-dev.
<pre>
<pre>
root@debian:/mnt/usb/R-2.15.2# apt-get build-dep r-base
write.table(df, 'my_data.txt', quote=FALSE, sep='\t', row.names=TRUE)
Reading package lists... Done
my_df <- read.delim('my_data.txt')  # it will automatically assign the rownames
Building dependency tree
Reading state information... Done
The following packages will be REMOVED:
  libreadline5-dev
The following NEW packages will be installed:
  bison ca-certificates ca-certificates-java debhelper defoma ed file fontconfig gettext
  gettext-base html2text intltool-debian java-common libaccess-bridge-java
  libaccess-bridge-java-jni libasound2 libasyncns0 libatk1.0-0 libaudit0 libavahi-client3
  libavahi-common-data libavahi-common3 libblas-dev libblas3gf libbz2-dev libcairo2
  libcairo2-dev libcroco3 libcups2 libdatrie1 libdbus-1-3 libexpat1-dev libflac8
  libfontconfig1-dev libfontenc1 libfreetype6-dev libgif4 libglib2.0-dev libgtk2.0-0
  libgtk2.0-common libice-dev libjpeg62-dev libkpathsea5 liblapack-dev liblapack3gf libnewt0.52
  libnspr4-0d libnss3-1d libogg0 libopenjpeg2 libpango1.0-0 libpango1.0-common libpango1.0-dev
  libpcre3-dev libpcrecpp0 libpixman-1-0 libpixman-1-dev libpng12-dev libpoppler5 libpulse0
  libreadline-dev libreadline6-dev libsm-dev libsndfile1 libthai-data libthai0 libtiff4-dev
  libtiffxx0c2 libunistring0 libvorbis0a libvorbisenc2 libxaw7 libxcb-render-util0
  libxcb-render-util0-dev libxcb-render0 libxcb-render0-dev libxcomposite1 libxcursor1
  libxdamage1 libxext-dev libxfixes3 libxfont1 libxft-dev libxi6 libxinerama1 libxkbfile1
  libxmu6 libxmuu1 libxpm4 libxrandr2 libxrender-dev libxss-dev libxt-dev libxtst6 luatex m4
  openjdk-6-jdk openjdk-6-jre openjdk-6-jre-headless openjdk-6-jre-lib openssl pkg-config
  po-debconf preview-latex-style shared-mime-info tcl8.5-dev tex-common texi2html texinfo
  texlive-base texlive-binaries texlive-common texlive-doc-base texlive-extra-utils
  texlive-fonts-recommended texlive-generic-recommended texlive-latex-base texlive-latex-extra
  texlive-latex-recommended texlive-pictures tk8.5-dev tzdata-java whiptail x11-xkb-utils
  x11proto-render-dev x11proto-scrnsaver-dev x11proto-xext-dev xauth xdg-utils xfonts-base
  xfonts-encodings xfonts-utils xkb-data xserver-common xvfb zlib1g-dev
0 upgraded, 136 newly installed, 1 to remove and 0 not upgraded.
Need to get 139 MB of archives.
After this operation, 410 MB of additional disk space will be used.
Do you want to continue [Y/n]?
</pre>
</pre>


== Web Applications ==
== Read/Write Excel files package ==
* 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.
* [http://cran.r-project.org/web/packages/xlsx/index.html xlsx]: depends on Java
** [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.
** It can't be used to open .xls or .xlm files.
** When I try the package to read an xlsx file, I got a warning: No data found on worksheet. 6/28/2018
** [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. 
** It is part of tidyverse package. The [https://readxl.tidyverse.org/index.html readxl] website provides several articles for more examples.
** [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.'''
** For writing to Excel formats, use writexl or openxlsx package.
:<syntaxhighlight lang='rsplus'>
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")
</syntaxhighlight>
* [https://ropensci.org/blog/technotes/2017/09/08/writexl-release writexl]: zero dependency xlsx writer for R
:<syntaxhighlight lang='rsplus'>
library(writexl)
mylst <- list(sheet1name = df1, sheet2name = df2)
write_xlsx(mylst, "output.xlsx")
</syntaxhighlight>
 
For the Chromosome column, integer values becomes strings (but converted to double, so 5 becomes 5.000000) or NA (empty on sheets).
{{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>


=== Create HTML5 web and slides ===
The hidden worksheets become visible (Not sure what are those first rows mean in the output).
http://www.gastonsanchez.com/depot/knitr-slides. The HTML5 slides work on my IE 8 too.
{{Pre}}
> 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"     
</pre>


HTML5 slides examples
The Chinese character works too.
* http://yihui.name/slides/knitr-slides.html
{{Pre}}
* http://yihui.name/slides/2012-knitr-RStudio.html
> read_excel("~/Downloads/testChinese.xlsx", 1)
* http://yihui.name/slides/2011-r-dev-lessons.html#slide1
  中文 B C
* http://inundata.org/R_talks/BARUG/#intro
1    a b c
2    1 2 3
</pre>


Software requirement
To read all worksheets we need a convenient function
* Rstudio
{{Pre}}
* knitr, XML, RCurl
read_excel_allsheets <- function(filename) {
* [http://johnmacfarlane.net/pandoc/ pandoc package] This is a command line tool. I am testing it on Windows 7.
    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>


Slide #22 gives an instruction to create
=== [https://cran.r-project.org/web/packages/readr/ readr] ===
* regular html file by using RStudio -> Knit HTML button
* HTML5 slides by using pandoc from command line.


Files:
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.
* Rcmd source: [https://github.com/yihui/knitr-examples/blob/master/009-slides.Rmd 009-slides.Rmd] Note that IE 8 was not supported by github. For IE 9, be sure to turn off "Compatibility View".
* markdown output: 009-slides.md
* HTML output: 009-slides.html


We can create Rcmd source in Rstudio by File -> New -> R Markdown.
[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.


There are 4 ways to produce slides with pandoc
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.
* S5
* 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.
* DZSlides
* readr::read_delim(Filename, "\t") will miss the last column. The dim is 6100 x 21.
* Slidy
* '''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!)
* Slideous


Use the markdown file (md) and convert it with pandoc
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.
<pre>
pandoc -s -S -i -t dzslides --mathjax html5_slides.md -o html5_slides.html
</pre>


If we are comfortable with HTML and CSS code, open the html file (generated by pandoc) and modify the CSS style at will.
Note that '''data.table::fread()''' can read a selection of the columns.


=== Markdown language ===
=== Speed comparison ===
[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.


According to [http://en.wikipedia.org/wiki/Markdown wikipedia]:
== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ==
See [[Ggplot2|ggplot2]]


''Markdown is a lightweight markup language, originally created by John Gruber with substantial contributions from Aaron Swartz, allowing people “to write using an easy-to-read, easy-to-write plain text format, then convert it to structurally valid XHTML (or HTML)”.
== Data Manipulation & Tidyverse ==
''
See [[Tidyverse|Tidyverse]].


* Markup is a general term for content formatting - such as HTML - but markdown is a library that generates HTML markup.
== Data Science ==
See [[Data_science|Data science]] page


* [http://stackoverflow.com/editing-help Nice summary from stackoverflow.com] and more complete list from [https://github.com/adam-p/markdown-here/wiki/Markdown-Cheatsheet github].
== 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)


* An example https://gist.github.com/jeromyanglim/2716336
== Plot, image ==
=== [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.


* [http://daringfireball.net/projects/markdown/basics basics] and [http://daringfireball.net/projects/markdown/syntax syntax]
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].


* Convert mediawiki to markdown using online conversion tool from [http://johnmacfarlane.net/pandoc/try/ pandoc].
=== png and resolution ===
It seems people use '''res=300''' as a definition of high resolution.  


* [http://support.mashery.com/docs/customizing_your_portal/Markdown_Cheat_Sheet Cheat sheet].
<ul>
<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.
<li>[https://stackoverflow.com/a/51194014 Saving high resolution plot in png].
<pre>
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.


* [http://dillinger.io/ Cloud-enabled HTML5 markdown editor]
# 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"
</pre>
<li>Chapter 14.5 [https://r-graphics.org/recipe-output-bitmap 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.
<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>


* [http://www.crypti.cc/markdown-here/livedemo.html live demo]
=== PowerPoint ===
<ul>
<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'''.
<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()


* [http://www.rstudio.com/ide/docs/r_markdown R markdown file] and use it in [http://www.rstudio.com/ide/docs/authoring/using_markdown RStudio]. Customizing Chunk Options can be found in [http://yihui.name/knitr/options knitr page] and [http://rpubs.com/gallery/options rpubs.com].
svg("svg7.svg", width=7, height=7) # default
plot(1:10, main="width=7, height=7")
dev.off()
</pre>
</ul>


=== [http://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol HTTP protocol] ===
=== magick ===
https://cran.r-project.org/web/packages/magick/


* http://en.wikipedia.org/wiki/File:Http_request_telnet_ubuntu.png
See an example [[:File:Progpreg.png|here]] I created.
* [http://en.wikipedia.org/wiki/Query_string Query string]
* 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.rstudio.com/shiny/ shiny] ===
=== [http://cran.r-project.org/web/packages/Cairo/index.html Cairo] ===
The following is what we see on a browser after we run an example from shiny package. See http://rstudio.github.com/shiny/tutorial/#hello-shiny. Note that the R session needs to be on; i.e. R command prompt will not be returned unless we press Ctrl+C or ESC.
See [[Heatmap#White_strips_.28artifacts.29|White strips problem]] in png() or tiff().


[[File:ShinyHello.png|100px]]
=== geDevices ===
[[File:Shinympg.png|100px]]
* [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.
[[File:ShinyReactivity.png|100px]]
* [https://www.jumpingrivers.com/blog/r-knitr-markdown-png-pdf-graphics/ Setting the Graphics Device in a RMarkdown Document]
[[File:ShinyTabsets.png|100px]]
[[File:ShinyUpload.png|100px]]


shiny depends on [http://cran.r-project.org/web/packages/websockets/index.html websockets], caTools, bitops, digest packages.
=== [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()).


Q & A:
For ubuntu OS, we need to install 2 libraries and 1 R package '''RGtk2'''.
* Q: If we run ''runExample('01_hello')'' in Rserve from an R client, we can continue our work in R client without losing the functionality of the GUI from shiny. Question: how do we kill the job?
* If I run the example "01_hello", the browser only shows the control but not graph on Firefox? A: Use Chrome or Opera as the default browser.
* If I run the example "01_hello" on RHEL the first time, it works fine. But if I click 'Ctrl + C' to stop it and run it again, I got a message
<pre>
<pre>
Warning in .SOCK_SERVE(port) : R-Websockets(tcpserv): bind() failed.
sudo apt-get install libgtk2.0-dev libcairo2-dev
Error in createContext(port, webpage, is.binary = is.binary) :
  Unable to bind socket on port 8100; is it realsy in use?
</pre>
</pre>
A simple solution is to close R and open it again.
* Q: Deployment on web. A: Not ready yet. Shiny server platform is still under beta testing. Shiny apps are hosted using the R websockets package which acts more like a tcp server than a web server, and that architecture just doesn't fit with rApache, or even apache for that matter.


* Q: How difficult to put the code in Gist:github? A: Just create an account. Do not even need to create a repository. Just go to http://gist.github.com and create a new gist. The new gist can be secret or public. A secret gist can not be edited again after it is created although it works fine when it was used in runGist() function.
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].
 
=== dpi requirement for publication ===
[http://www.cookbook-r.com/Graphs/Output_to_a_file/ For import into PDF-incapable programs (MS Office)]
 
=== sketcher: photo to sketch effects ===
https://htsuda.net/sketcher/
 
=== 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]


== [http://igraph.org/r/ igraph] ==
[https://shiring.github.io/genome/2016/12/14/homologous_genes_part2_post creating directed networks with igraph]


==== shiny server ====
== Identifying dependencies of R functions and scripts ==
See https://github.com/rstudio/shiny-server
https://stackoverflow.com/questions/8761857/identifying-dependencies-of-r-functions-and-scripts
{{Pre}}
library(mvbutils)
foodweb(where = "package:batr")


It works on my ubuntu server. To test, I need to run
foodweb( find.funs("package:batr"), prune="survRiskPredict", lwd=2)
<pre>
 
sudo shiny-server
foodweb( find.funs("package:batr"), prune="classPredict", lwd=2)
# maybe I need to add ampersand '&' sign to the end of the above command
</pre>
</pre>


==== Heatmap Example ====
== [http://cran.r-project.org/web/packages/iterators/ iterators] ==
http://taichi.selfip.net:3838/hello/
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


=== [http://rapache.net/ RApache] ===
Iterator can be combined to use with foreach package http://www.exegetic.biz/blog/2013/11/iterators-in-r/ has more elaboration.


=== [http://cran.r-project.org/web/packages/gWidgetsWWW/index.html gWidgetsWWW] ===
== 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>
hcl_palettes(plot = TRUE) # a quick overview
hcl_palettes(palette = "Dark 2", n=5, plot = T)
q4 <- qualitative_hcl(4, palette = "Dark 3")
</pre>
</ul>
* [https://statisticsglobe.com/create-color-range-between-two-colors-in-r Create color range between two colors in R] using colorRampPalette()
* [http://novyden.blogspot.com/2013/09/how-to-expand-color-palette-with-ggplot.html How to expand color palette with ggplot and RColorBrewer]
* 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>
</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]


* http://www.jstatsoft.org/v49/i10/paper
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.  
* [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] ===
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.


Since R 2.13, the internal web server was exposed.
[[:File:GgplotPalette.svg]]


[https://docs.google.com/present/view?id=0AUTe_sntp1JtZGdnbjVicTlfMzFuZDQ5dmJxNw Tutorual from useR2012] and [https://github.com/rstats/RookTutorial Jeffrey Horner]
=== [http://rpubs.com/gaston/colortools colortools] ===
Tools that allow users generate color schemes and palettes


Here is another [http://www.rinfinance.com/agenda/2011/JeffHorner.pdf one] from http://www.rinfinance.com.
=== [https://github.com/daattali/colourpicker colourpicker] ===
A Colour Picker Tool for Shiny and for Selecting Colours in Plots


Rook is also supported by [rApache too. See http://rapache.net/manual.html.
=== eyedroppeR ===
[http://gradientdescending.com/select-colours-from-an-image-in-r-with-eyedropper/ Select colours from an image in R with {eyedroppeR}]


Google group. https://groups.google.com/forum/?fromgroups#!forum/rrook
== [https://github.com/kevinushey/rex rex] ==
Friendly Regular Expressions


Advantage
== [http://cran.r-project.org/web/packages/formatR/index.html formatR] ==
* the web applications are created on desktop, whether it is Windows, Mac or Linux.
'''The best strategy to avoid failure is to put comments in complete lines or after complete R expressions.'''
* 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.  


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


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].
<pre>
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
* Comments appearing at the beginning of a line within a long complete statement. This will break tidy_source().
<pre>
<pre>
library(Rook)
cat("abcd",
s <- Rhttpd$new()
    # This is my comment
s$start(quiet=TRUE)
    "defg")
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.
will result in
<pre>
> 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"
  ^
</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>
cat("abcd"
    ,"defg"  # This is my comment
  ,"ghij")
</pre>
will become
<pre>
cat("abcd", "defg"  # This is my comment
, "ghij")
</pre>
Still bad!!
* Comments appearing at the end of a line within a long complete statement ''breaks'' tidy_source() function. For example,
<pre>
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)
</pre>
will result in
<pre>
> 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%
</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>
 
== styler ==
https://cran.r-project.org/web/packages/styler/index.html Pretty-prints R code without changing the user's formatting intent.


[[File:Rook.png|100px]]
== Download papers ==
[[File:Rook2.png|100px]]
=== [http://cran.r-project.org/web/packages/biorxivr/index.html biorxivr] ===
[[File:Rookapprnorm.png|100px]]
Search and Download Papers from the bioRxiv Preprint Server (biology)


We can add Rook '''application''' to the server; see ?Rhttpd.
=== [http://cran.r-project.org/web/packages/aRxiv/index.html aRxiv] ===
<pre>
Interface to the arXiv API
s$add(
    app=system.file('exampleApps/helloworld.R',package='Rook'),name='hello'
)
s$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()
=== [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/


#Server started on 127.0.0.1:10221
== [https://github.com/ColinFay/aside aside]: set it aside ==
#[1] RookTest http://127.0.0.1:10221/custom/RookTest
An RStudio addin to run long R commands aside your current session.
#[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
== Teaching ==
## Not run:  
* [https://cran.r-project.org/web/packages/smovie/vignettes/smovie-vignette.html smovie]: Some Movies to Illustrate Concepts in Statistics
s$stop()


## End(Not run)
== Organize R research project ==
s$remove(all=TRUE)
* [https://cran.r-project.org/web/views/ReproducibleResearch.html CRAN Task View: Reproducible Research]
rm(s)
* [https://ntguardian.wordpress.com/2019/02/04/organizing-r-research-projects-cpat-case-study/ Organizing R Research Projects: CPAT, A Case Study]
</pre>
* [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())'''.
For example, the interface and the source code of ''summary'' app are given below
** [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]


[[File:Rookappsummary.png|100px]]
=== How to save (and load) datasets in R (.RData vs .Rds file) ===
[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]


<nowiki>
=== Naming convention ===
app <- function(env) {
<ul>
    req <- Rook::Request$new(env)
<li>[https://stackoverflow.com/a/1946879 What is your preferred style for naming variables in R?]
    res <- Rook::Response$new()
* 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)
    res$write('Choose a CSV file:\n')
* Use of underscores: really annoying for ESS users
    res$write('<form method="POST" enctype="multipart/form-data">\n')
* '''camelCase''': Winner
    res$write('<input type="file" name="data">\n')
<li>However, the [https://stackoverflow.com/a/13413278 survey] said (no surprises perhaps) that
    res$write('<input type="submit" name="Upload">\n</form>\n<br>')
* '''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


    if (!is.null(req$POST())){
dataClinicalDesign
data <- req$POST()[['data']]
dataGeneExpression
res$write("<h3>Summary of Data</h3>");
dataAnnotation
res$write("<pre>")
</pre>
res$write(paste(capture.output(summary(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
<pre>
res$write("</pre>")
# Search all variables ending with .Data
res$write("<h3>First few lines (head())</h3>");
ls()[grep("\\.Data$", ls())]
res$write("<pre>")
# Search all variables starting with data_
res$write(paste(capture.output(head(read.csv(data$tempfile,stringsAsFactors=FALSE)),file=NULL),collapse='\n'))
ls()[grep("^data_", ls())]
res$write("</pre>")
</pre>
    }
</ul>
    res$finish()
}
</nowiki>


More example:
=== Efficient Data Management in R ===
* http://lamages.blogspot.com/2012/08/rook-rocks-example-with-googlevis.html
[https://www.mzes.uni-mannheim.de/socialsciencedatalab/article/efficient-data-r/ Efficient Data Management in R]. .Rprofile, renv package and dplyr package.
* [http://www.road2stat.com/cn/r/rook.html Self-organizing map]
* 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].


=== [http://www.stat.ucla.edu/~jeroen/stockplot Stockplot] ===
== Text to speech ==
[https://shirinsplayground.netlify.com/2018/06/googlelanguager/ Text-to-Speech with the googleLanguageR package]


=== [http://www.rforge.net/FastRWeb/ FastRWeb] ===
== Speech to text ==
https://github.com/ggerganov/whisper.cpp and an R package [https://github.com/bnosac/audio.whisper audio.whisper]


=== [http://sysbio.mrc-bsu.cam.ac.uk/Rwui/tutorial/Instructions.html Rwui] ===
== 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/CGIwithR/index.html CGHWithR] (removed from CRAN) ===
== logR ==
But it is still working with old version of R.
https://github.com/jangorecki/logR


== Creating local repository for CRAN and Bioconductor (focus on Windows binary packages only) ==
== Progress bar ==
=== How to set up a local repository ===
https://github.com/r-lib/progress#readme


* CRAN specific: http://cran.r-project.org/mirror-howto.html
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'.
* Bioconductor specific: http://www.bioconductor.org/about/mirrors/mirror-how-to/


General guide: http://cran.r-project.org/doc/manuals/R-admin.html#Setting-up-a-package-repository
== cron ==
* [https://github.com/bnosac/cronr cronR]
* [https://mathewanalytics.com/building-a-simple-pipeline-in-r/ Building a Simple Pipeline in R]


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:'''
== 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".


* "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.
== utils package ==
* '''"win.binary": located at bin/windows/contrib/x.y for R versions x.y.z and containing .zip files for Windows.'''
https://www.rdocumentation.org/packages/utils/versions/3.6.2
* "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.)
== tools package ==
* 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>)]


To add your repository to the list offered by setRepositories(), see the help file for that function.
= Different ways of using R =
[https://www.amazon.com/Extending-Chapman-Hall-John-Chambers/dp/1498775713 Extending R] by John M. Chambers (2016)


A repository can contain subdirectories, when the descriptions in the PACKAGES file of packages in subdirectories must include a line of the form
== 10 things R can do that might surprise you ==
https://simplystatistics.org/2019/03/13/10-things-r-can-do-that-might-surprise-you/


<nowiki>Path: path/to/subdirectory</nowiki>
== R call C/C++ ==
Mainly talks about .C() and .Call().


—once again write_PACKAGES is the simplest way to set this up.
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.


==== Space requirement if we want to mirror WHOLE repository ====
* [http://cran.r-project.org/doc/manuals/R-exts.html R-Extension manual] of course.
* Whole CRAN takes about 92GB (rsync -avn  cran.r-project.org::CRAN > ~/Downloads/cran).
* [http://r-pkgs.had.co.nz/src.html Compiled Code] chapter from 'R Packages' by Hadley Wickham
* 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.
* 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


On the other hand, we if only care about Windows binary part, the space requirement is largely reduced.
=== .Call ===
* CRAN: 2.7GB
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/CallExternal ?.Call]
* Bioconductor: 28GB.
* [http://mazamascience.com/WorkingWithData/?p=1099 Using R — .Call(“hello”)]
* http://adv-r.had.co.nz/C-interface.html
* [https://working-with-data.mazamascience.com/2021/07/16/using-r-callhello/ Using R – .Call(“hello”)]


==== Misc notes ====
Be sure to add the ''PACKAGE'' parameter to avoid an error like
* 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.
* 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>
<pre>
Warning: unable to access index for repository http://arraytools.no-ip.org/CRAN/src/contrib
cvfit <- cv.grpsurvOverlap(X, Surv(time, event), group,
Warning message:
                            cv.ind = cv.ind, seed=1, penalty = 'cMCP')
package ‘glmnet’ is not available (for R version 2.15.2)  
Error in .Call("standardize", X) :  
  "standardize" not resolved from current namespace (grpreg)
</pre>
</pre>
The error was given by available.packages() function.


To bypass the requirement of src directory, I can use
=== NAMESPACE file & useDynLib ===
<pre>
* https://cran.r-project.org/doc/manuals/r-release/R-exts.html#useDynLib
install.packages("glmnet", contriburl = contrib.url(getOption('repos'), "win.binary"))
* We don't need to include double quotes around the C/Fortran subroutines in .C() or .Fortran()
* 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().
* stats example: [https://github.com/wch/r-source/blob/trunk/src/library/stats/NAMESPACE NAMESPACE]
 
(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
{{Pre}}
library.dynam("libname", package, lib.loc)  
</pre>
</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.
=== library.dynam.unload() ===
* https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.dynam.html
* 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.
* digest package example [https://github.com/cran/digest/blob/master/R/zzz.R zzz.R]
 
=== gcc ===
[http://rorynolan.rbind.io/2019/06/30/strexgcc/ Coping with varying `gcc` versions and capabilities in R packages]
 
=== Primitive functions ===
[https://nathaneastwood.github.io/2020/02/01/primitive-functions-list/ Primitive Functions List]
 
== SEXP ==
Some examples from packages
 
* [https://www.bioconductor.org/packages/release/bioc/html/sva.html sva] package has one C code function
 
== R call Fortran ==
* [https://stat.ethz.ch/pipermail/r-devel/2015-March/070851.html R call Fortran 90]
* [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)
 
== Embedding R ==
 
* 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.
* [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
 
=== 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
<pre>R_HOME/bin/R CMD R_HOME/bin/exec/R</pre>
 
Note:
# '''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''.
 
More examples of embedding can be found in ''tests/Embedding'' directory. Read <index.html> for more information about these test examples.


==== To create CRAN repository ====
=== An example from Bioconductor workshop ===
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.
* What is covered in this section is different from [[R#Create_a_standalone_Rmath_library|Create and use a standalone Rmath library]].
* 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://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


Dry run (-n option). Pipe out the process to a text file for an examination.
Example:
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>
<pre>
rsync -avn cran.r-project.org::CRAN > crandryrun.txt
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
</pre>
</pre>
To mirror only partial repository, it is necessary to create directories before running rsync command.
 
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].
<pre>
<pre>
cd
export R_HOME=/home/brb/Downloads/R-3.0.2
mkdir -p ~/Rmirror/CRAN/bin/windows/contrib/2.15
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib
rsync -rtlzv --delete cran.r-project.org::CRAN/bin/windows/contrib/2.15/ ~/Rmirror/CRAN/bin/windows/contrib/2.15
./embed # No need to include R CMD in front.
(one line with space before ~/Rmirror)
</pre>


# src directory is very large (~27GB) since it contains source code for each R version.  
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].
# We just need the files PACKAGES and PACKAGES.gz in CRAN/src/contrib. So I comment out the following line.
 
# rsync -rtlzv --delete cran.r-project.org::CRAN/src/ ~/Rmirror/CRAN/src/
Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
mkdir -p ~/Rmirror/CRAN/src/contrib
 
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/PACKAGES ~/Rmirror/CRAN/src/contrib/
=== Create a Simple Socket Server in R ===
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/PACKAGES.gz ~/Rmirror/CRAN/src/contrib/
This example is coming from this [http://epub.ub.uni-muenchen.de/2085/1/tr012.pdf paper].
</pre>
 
And optionally
Create an R function
<pre>
<pre>
library(tools)
simpleServer <- function(port=6543)
write_PACKAGES("~/Rmirror/CRAN/bin/windows/contrib/2.15", type="win.binary")  
{
  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>
and if we want to get src directory
Then run simpleServer(). Open another terminal and try to communicate with the server
<pre>
<pre>
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/*.tar.gz ~/Rmirror/CRAN/src/contrib/
$ telnet localhost 6543
rsync -rtlzv --delete cran.r-project.org::CRAN/src/contrib/2.15.3 ~/Rmirror/CRAN/src/contrib/
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.
</pre>
</pre>


We can use '''du -h''' to check the folder size.  
=== [http://www.rforge.net/Rserve/doc.html Rserve] ===
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]].


For example (as of 1/7/2013),
See my [[Rserve]] page.
 
=== outsider ===
* [https://joss.theoj.org/papers/10.21105/joss.02038 outsider]: Install and run programs, outside of R, inside of R
* [https://github.com/stephenturner/om..bcftools Run bcftools with outsider in R]
 
=== (Commercial) [http://www.statconn.com/ StatconnDcom] ===
 
=== [http://rdotnet.codeplex.com/ R.NET] ===
 
=== [https://cran.r-project.org/web/packages/rJava/index.html rJava] ===
* [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
{{Pre}}
# 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
</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>
<pre>
$ du -k ~/Rmirror --max-depth=1 --exclude ".*" | sort -nr | cut -f2 | xargs -d '\n' du -sh
/usr/lib/jvm/java-8-oracle/jre/lib/amd64
30G /home/brb/Rmirror
/usr/lib/jvm/java-8-oracle/jre/lib/amd64/server
28G /home/brb/Rmirror/Bioc
2.7G /home/brb/Rmirror/CRAN
</pre>
</pre>
* And then run '''sudo ldconfig'''


==== To create Bioconductor repository ====
Now go back to R
Dry run
{{Pre}}
<pre>
install.packages("rJava")
rsync -avn bioconductor.org::2.11 > biocdryrun.txt
</pre>
</pre>
Then creates directories before running rsync.
Done!


If above does not work, a simple way is by (under Ubuntu) running
<pre>
<pre>
cd
sudo apt-get install r-cran-rjava
mkdir -p ~/Rmirror/Bioc
wget -N http://www.bioconductor.org/biocLite.R -P ~/Rmirror/Bioc
</pre>
</pre>
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.
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


Optionally, we can add the following in order to see the Bioconductor front page.
=== 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>
rsync -zrtlv  --delete bioconductor.org::2.11/BiocViews.html ~/Rmirror/Bioc/packages/2.11/
$ Rscript --help
rsync -zrtlv  --delete bioconductor.org::2.11/index.html ~/Rmirror/Bioc/packages/2.11/
Usage: /path/to/Rscript [--options] [-e expr [-e expr2 ...] | file] [args]
</pre>
</pre>


The software part (aka bioc directory) installation:
Example:
<pre>
<pre>
cd
args = commandArgs(trailingOnly=TRUE)
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/bin/windows
# test if there is at least one argument: if not, return an error
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/src
if (length(args)==0) {
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/bin/windows/ ~/Rmirror/Bioc/packages/2.11/bioc/bin/windows
  stop("At least one argument must be supplied (input file).n", call.=FALSE)
# Either rsync whole src directory or just essential files
} else if (length(args)==1) {
# rsync -zrtlv  --delete bioconductor.org::2.11/bioc/src/ ~/Rmirror/Bioc/packages/2.11/bioc/src
  # default output file
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/src/contrib/PACKAGES ~/Rmirror/Bioc/packages/2.11/bioc/src/contrib/
  args[2] = "out.txt"
rsync -zrtlv  --delete bioconductor.org::2.11/bioc/src/contrib/PACKAGES.gz ~/Rmirror/Bioc/packages/2.11/bioc/src/contrib/
}
# Optionally the html part
cat("args[1] = ", args[1], "\n")
mkdir -p ~/Rmirror/Bioc/packages/2.11/bioc/html
cat("args[2] = ", args[2], "\n")
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
</pre>
</pre>
and annotation (aka data directory) part:
<pre>
<pre>
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/annotation/bin/windows
Rscript --vanilla sillyScript.R iris.txt out.txt
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/annotation/src/contrib
# args[1] =  iris.txt
# one line for each of the following
# args[2] =  out.txt
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/
</pre>
</pre>
and experiment directory:
 
=== Rscript, #! Shebang and optparse package ===
<ul>
<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?]
<li>[https://blog.rmhogervorst.nl/blog/2020/04/14/where-does-the-output-of-rscript-go/ Where does the output of Rscript go?]
<li>Create a file <shebang.R>.
<pre>
<pre>
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15
#!/usr/bin/env Rscript
mkdir -p ~/Rmirror/Bioc/packages/2.11/data/experiment/src/contrib
print ("shebang works")
# 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/
</pre>
</pre>
and extra directory:
Then in the command line
<pre>
<pre>
mkdir -p ~/Rmirror/Bioc/packages/2.11/extra/bin/windows/contrib/2.15
chmod u+x shebang.R
mkdir -p ~/Rmirror/Bioc/packages/2.11/extra/src/contrib
./shebang.R
# 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/
</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>


=== To test local repository ===
=== [http://dirk.eddelbuettel.com/code/littler.html littler] ===
Provides hash-bang (#!) capability for R


==== Create soft links in Apache server ====
FAQs:
<pre>
* [http://stackoverflow.com/questions/3205302/difference-between-rscript-and-littler Difference between Rscript and littler]
su
* [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]
ln -s /home/brb/Rmirror/CRAN /var/www/html/CRAN
* [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?]
ln -s /home/brb/Rmirror/Bioc /var/www/html/Bioc
{{Pre}}
ls -l /var/www/html
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()"
</pre>
</pre>
The soft link mode should be 777.


==== To test CRAN ====
We can install littler using two ways.
Replace the host name arraytools.no-ip.org by IP address 10.133.2.111 if necessary.
* 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 [http://dirk.eddelbuettel.com/code/littler.examples.html 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 [[R#RInside|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
<pre>export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/brb/Downloads/R-3.0.2/lib </pre>


The real build process looks like (check <Makefile> for completeness)
<pre>
<pre>
r <- getOption("repos"); r["CRAN"] <- "http://arraytools.no-ip.org/CRAN"
g++ -I/home/brb/Downloads/R-3.0.2/include \
options(repos=r)
    -I/home/brb/Downloads/R-3.0.2/library/Rcpp/include \
install.packages("glmnet")
    -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>
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).


Hello World example of embedding R in C++.
<pre>
<pre>
r <- getOption("repos"); r["CRAN"] <- "http://cran.r-project.org"
#include <RInside.h>                    // for the embedded R via RInside
r <- c(r, BRB='http://arraytools.no-ip.org/CRAN')
 
#                        CRAN                            CRANextra                                  BRB
int main(int argc, char *argv[]) {
# "http://cran.r-project.org" "http://www.stats.ox.ac.uk/pub/RWin"  "http://arraytools.no-ip.org/CRAN"
 
options(repos=r)
    RInside R(argc, argv);             // create an embedded R instance
install.packages('ForImp')
 
    R["txt"] = "Hello, world!\n"; // assign a char* (string) to 'txt'
 
    R.parseEvalQ("cat(txt)");          // eval the init string, ignoring any returns
 
    exit(0);
}
</pre>
</pre>


Note by default, CRAN mirror is selected interactively.
The above can be compared to the Hello world example in Qt.
<pre>
<pre>
> getOption("repos")
#include <QApplication.h>
                                CRAN                            CRANextra
#include <QPushButton.h>
                            "@CRAN@" "http://www.stats.ox.ac.uk/pub/RWin"
 
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();
}
</pre>
</pre>


==== To test Bioconductor ====
=== [http://www.rfortran.org/ RFortran] ===
<pre>
RFortran is an open source project with the following aim:
# CRAN part:
 
r <- getOption("repos"); r["CRAN"] <- "http://arraytools.no-ip.org/CRAN"
''To provide an easy to use Fortran software library that enables Fortran programs to transfer data and commands to and from R.''
options(repos=r)
 
# Bioconductor part:
It works only on Windows platform with Microsoft Visual Studio installed:(
options("BioC_mirror" = "http://arraytools.no-ip.org/Bioc")
 
source("http://bioconductor.org/biocLite.R")
== Call R from other languages ==
# This source biocLite.R line can be placed either before or after the previous 2 lines
=== C ===
biocLite("aCGH")
[http://sebastian-mader.net/programming/using-r-from-c-c/ Using R from C/C++]
</pre>
 
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
 
=== JRI ===
http://www.rforge.net/JRI/
 
=== ryp2 ===
http://rpy.sourceforge.net/rpy2.html


If there is a connection problem, check folder attributes.
== Create a standalone Rmath library ==
<pre>
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].
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).
Here is my experience based on R 3.0.2 on Windows OS.


* 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.
=== Create a static library <libRmath.a> and a dynamic library <Rmath.dll> ===
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.
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.
<pre>
<pre>
options(install.packages.check.source = "no")
cd C:\R\R-3.0.2\src\nmath\standalone
make -f Makefile.win
</pre>
</pre>


* If we only mirror the essential directories, we can run biocLite() successfully. However, the R console will give some warning
=== Use Rmath library in our code ===
<pre>
<pre>
> biocLite("aCGH")
set CPLUS_INCLUDE_PATH=C:\R\R-3.0.2\src\include
BioC_mirror: http://arraytools.no-ip.org/Bioc
set LIBRARY_PATH=C:\R\R-3.0.2\src\nmath\standalone
Using Bioconductor version 2.11 (BiocInstaller 1.8.3), R version 2.15.
# It is not LD_LIBRARY_PATH in above.
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
# 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.


The downloaded binary packages are in
# Force to link against the static library <libRmath.a>
        C:\Users\limingc\AppData\Local\Temp\Rtmp8IGGyG\downloaded_packages
g++ RmathEx1.cpp -lRmath -lm -o RmathEx1.exe
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/data/experiment/bin/windows/contrib/2.15
# OR
Warning: unable to access index for repository http://arraytools.no-ip.org/Bioc/packages/2.11/extra/bin/windows/contrib/2.15
g++ RmathEx1.cpp -Wl,-Bstatic -lRmath -lm -o RmathEx1.exe
> library()
 
# Force to link against dynamic library <Rmath.dll>
g++ RmathEx1.cpp Rmath.dll -lm -o RmathEx1Dll.exe
</pre>
</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!
=== CRAN repository directory structure ===
The information below is specific to R 2.15.2. There are linux and macosx subdirecotries whenever there are windows subdirectory.
<pre>
<pre>
bin/winows/contrib/2.15
c:\R>RmathEx1
src/contrib
Enter a argument for the normal cdf:
  /contrib/2.15.2
1
  /contrib/Archive
Enter a argument for the chi-squared cdf:
web/checks
1
  /dcmeta
Prob(Z <= 1) = 0.841345
  /packages
Prob(Chi^2 <= 1)= 0.682689
  /views
</pre>
</pre>


A clickable map [http://taichi.selfip.net:81/RmirrorMap/Rmirror.html]
Below is the cpp program <RmathEx1.cpp>.
 
=== Bioconductor repository directory structure ===
The information below is specific to Bioc 2.11. There are linux and macosx subdirecotries whenever there are windows subdirectory.
<pre>
<pre>
bioc/bin/windows/contrib/2.15
//RmathEx1.cpp
    /html
#define MATHLIB_STANDALONE
    /install
#include <iostream>
    /license
#include "Rmath.h"
    /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 ===
using std::cout; using std::cin; using std::endl;
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]
int main()
# [http://taichi.selfip.net:81/Rsummary/bioc.html Bioc software]
{
# [http://taichi.selfip.net:81/Rsummary/annotation.html Bioc annotation]
  double x1, x2;
# [http://taichi.selfip.net:81/Rsummary/experiment.html Bioc experiment]
  cout << "Enter a argument for the normal cdf:" << endl;
  cin >> x1;
  cout << "Enter a argument for the chi-squared cdf:" << endl;
  cin >> x2;


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


=== snowfall package ===
== Calling R.dll directly ==
http://www.imbi.uni-freiburg.de/parallel/docs/Reisensburg2009_TutParallelComputing_Knaus_Porzelius.pdf
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.


== Cloud Computing ==
== Create HTML report ==
[http://www.bioconductor.org/packages/release/bioc/html/ReportingTools.html ReportingTools] (Jason Hackney) from Bioconductor. See [[Genome#ReportingTools|Genome->ReportingTools]].


=== Install R on Amazon EC2 ===
=== [http://cran.r-project.org/web/packages/htmlTable/index.html htmlTable] package ===
http://randyzwitch.com/r-amazon-ec2/
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.
 
* http://cran.r-project.org/web/packages/htmlTable/vignettes/general.html
* 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]


== Big Data Analysis ==
=== [https://cran.r-project.org/web/packages/formattable/index.html formattable] ===
http://blog.comsysto.com/2013/02/14/my-favorite-community-links/
* https://github.com/renkun-ken/formattable
* http://www.magesblog.com/2016/01/formatting-table-output-in-r.html
* [https://www.displayr.com/formattable/ Make Beautiful Tables with the Formattable Package]


== Useful R packages ==
=== [https://github.com/crubba/htmltab htmltab] package ===
This package is NOT used to CREATE html report but EXTRACT html table.


=== Large data (~100 terabytes) ===
=== [http://cran.r-project.org/web/packages/ztable/index.html ztable] package ===
See also [http://cran.r-project.org/web/views/HighPerformanceComputing.html HighPerformanceComputing]
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.


* RHadoop
== Create academic report ==
* Hive
[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.
* [http://cran.r-project.org/web/packages/mapReduce/ MapReduce]. Introduction by [http://www.linuxjournal.com/content/introduction-mapreduce-hadoop-linux Linux Journal].


=== [http://cran.r-project.org/web/packages/XML/index.html XML] ===
== Create pdf and epub files ==
On Ubuntu, we need to install libxml2-dev before we can install XML package.
{{Pre}}
<pre>
# Idea:
sudo apt-get update
#        knitr        pdflatex
sudo apt-get install libxml2-dev
#  rnw -------> tex ----------> pdf
library(knitr)
knit("example.rnw") # create example.tex file
</pre>
* A very simple example <002-minimal.Rnw> from [http://yihui.name/knitr/demo/minimal/ yihui.name] works fine on linux.
{{Pre}}
git clone https://github.com/yihui/knitr-examples.git
</pre>
</pre>
* <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!


=== [http://cran.r-project.org/web/packages/GenOrd/index.html GenOrd]: Generate ordinal and discrete variables with given correlation matrix and marginal distributions ===
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.
[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] ===
Or starts with markdown file. Download the example <001-minimal.Rmd> and remove the last line of getting png file from internet.
http://heuristically.wordpress.com/2013/05/20/geolocate-ip-addresses-in-r/
{{Pre}}
# Idea:
#        knitr        pandoc
#  rmd -------> md ----------> pdf


=== [http://cran.r-project.org/web/packages/RJSONIO/index.html RJSONIO] ===
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 !!
</pre>


=== [http://cran.r-project.org/web/packages/Rcpp/index.html Rcpp] ===
To create an epub file (not success yet on Windows OS, missing figures on Linux OS)
{{Pre}}
# Idea:
#        knitr        pandoc
#  rnw -------> tex ----------> markdown or epub


=== [http://cran.r-project.org/web/packages/caret/index.html caret] ===
library(knitr)
knit("DESeq2.Rnw") # create DESeq2.tex
system("pandoc  -f latex -t markdown -o DESeq2.md DESeq2.tex")
</pre>


=== xlsx package ===
Convert tex to epub
* http://tex.stackexchange.com/questions/156668/tex-to-epub-conversion


=== [http://cran.r-project.org/web/packages/ggplot2/index.html ggplot2] ===
=== [https://www.rdocumentation.org/packages/knitr/versions/1.20/topics/kable kable()] for tables ===
Create Tables In LaTeX, HTML, Markdown And ReStructuredText


=== stringr and plyr ===
* https://rmarkdown.rstudio.com/lesson-7.html
http://martinsbioblogg.wordpress.com/2013/03/24/using-r-reading-tables-that-need-a-little-cleaning/
* 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


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.
== Create Word report ==


=== [http://cran.r-project.org/web/packages/jpeg/index.html jpeg] ===
=== Using the power of Word ===
If we want to create the image on this wiki left hand side panel, we can use jpeg package to read an existing plot and then edit and save it.
[https://www.rforecology.com/post/exporting-tables-from-r-to-microsoft-word/ How to go from R to nice tables in Microsoft Word]
 
== Different ways of using R ==
 
=== Create HTML 5 web and slides ===
See [[#Create HTML5 web and slides|here]]
 
=== Create academic report ===
[http://cran.r-project.org/web/packages/reports/index.html reports] package and [https://github.com/trinker/reports github] repository. The youtube video gives an overview of the package.


=== Create Word report ===
=== knitr + pandoc ===
 
==== knitr + pandoc ====
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* http://www.r-statistics.com/2013/03/write-ms-word-document-using-r-with-as-little-overhead-as-possible/
* http://www.carlboettiger.info/2012/04/07/writing-reproducibly-in-the-open-with-knitr.html
* 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.
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.
Line 925: Line 1,504:
* Word docx: pandoc report.md -o report.docx
* Word docx: pandoc report.md -o report.docx


==== pander ====
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!
<pre>
knit("example.Rmd")
pandoc("example.md", format="epub")
</pre>
 
PS. If we don't remove the link, we will get an error message (pandoc 1.10.1 on Windows 7)
<pre>
> 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
</pre>
 
=== pander ===
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:
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:


Line 942: Line 1,537:
# http://rapporter.github.com/pander/#examples
# http://rapporter.github.com/pander/#examples


==== R2wd ====
=== 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.  
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>
Line 963: Line 1,558:
The solution is to launch 32-bit R instead of 64-bit R since statconnDCOM does not support 64-bit R.
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 ====
=== 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
The best rendering of advanced tables is done by converting from pdf to Word. See http://biostat.mc.vanderbilt.edu/wiki/Main/SweaveConvert


==== rtf ====
=== rtf ===
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.
Use [http://cran.r-project.org/web/packages/rtf/ rtf] package for Rich Text Format (RTF) Output.


==== xtable ====
=== [https://www.rdocumentation.org/packages/xtable/versions/1.8-2 xtable] ===
Package xtable will produce html output. 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.
Package xtable will produce html output.  
{{Pre}}
print(xtable(X), type="html")
</pre>
 
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 ===
<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, "")


=== Call R from C/C++ ===
# Function to add n line spaces
* Use eval() function. See R-Ext 8.1 and 8.2.
body_add_par_n <- function (doc, n) {
* http://stackoverflow.com/questions/7457635/calling-r-function-from-c
  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>
x = read_docx("myfile.docx")
content <- docx_summary(x) # a vector
grep("nlme", content$text, ignore.case = T, value = T)
</pre>
</ul>
 
== Powerpoint ==
<ul>
<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]
</li>
<li>[https://davidgohel.github.io/flextable/ flextable] (imports '''officer''')
</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>
 
== PDF manipulation ==
[https://github.com/pridiltal/staplr staplr]
 
== 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]
 
== COM client or server ==
 
=== Client ===
 
[http://www.omegahat.org/RDCOMClient/ RDCOMClient] where [http://cran.r-project.org/web/packages/excel.link/index.html excel.link] depends on it.
 
=== Server ===
[http://www.omegahat.org/RDCOMServer/ RDCOMServer]


=== Use R under proxy ===
== Use R under proxy ==
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy


=== What is the best place to save Rconsole on Windows platform ===
== RStudio ==
Put it in ''C:/Users/USERNAME/Documents'' folder so no matter how R was upgraded/downgraded, it always find my preference.
* [https://github.com/rstudio/rstudio Github]
* Installing RStudio (1.0.44) on Ubuntu will not install Java even the source code contains 37.5% Java??
* [https://www.rstudio.com/products/rstudio/download/preview/ Preview]
 
=== rstudio.cloud ===
https://rstudio.cloud/
 
=== Launch RStudio ===
[[Rstudio#Multiple_versions_of_R|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


=== Web scraping ===
=== package search ===
http://www.slideshare.net/schamber/web-data-from-r#btnNext
https://github.com/RhoInc/CRANsearcher


=== Launch Rstudio ===
=== Git ===
If multiple versions of R was detected, Rstudio can not be launched successfully. A java-like clock will be spinning without a stop. The trick is to click Ctrl key and click the Rstudio at the same time.
* (Video) [https://www.rstudio.com/resources/videos/happy-git-and-gihub-for-the-user-tutorial/ Happy Git and Gihub for the useR – Tutorial]
After done that, it will show up a selection of R to choose from.
* [https://owi.usgs.gov/blog/beyond-basic-git/ Beyond Basic R - Version Control with Git]


[[File:RStudio.jpg|100px]]
== 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]


=== List files using regular expression ===
== List files using regular expression ==
* Extension
* Extension
<pre>
<pre>
list.files(pattern = "\\.txt$")
list.files(pattern = "\\.txt$")
</pre>
</pre>
where the dot (.) is a metacharacter. It is used to refer to any character.
* Start with
* Start with
<pre>
<pre>
Line 1,001: Line 1,679:
</pre>
</pre>


=== Hidden tool: rsync in Rtools ===
Using '''Sys.glob()"' as
<pre>
> Sys.glob("~/Downloads/*.txt")
[1] "/home/brb/Downloads/ip.txt"      "/home/brb/Downloads/valgrind.txt"
</pre>
 
== Hidden tool: rsync in Rtools ==
<pre>
<pre>
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
c:\Rtools\bin>rsync -avz "/cygdrive/c/users/limingc/Downloads/a.exe" "/cygdrive/c/users/limingc/Documents/"
Line 1,012: Line 1,696:
c:\Rtools\bin>
c:\Rtools\bin>
</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
total size is 8036311  speedup is 1.95
c:\Rtools\bin>rm c:\users\limingc\Documents\binary\procexp.exe
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/"
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].
sending incremental file list
binary/
binary/procexp.exe


sent 1767277 bytes  received 35 bytes  3534624.00 bytes/sec
== Install rgdal package (geospatial Data) on ubuntu ==
total size is 8036311  speedup is 4.55
Terminal
 
{{Pre}}
c:\Rtools\bin>
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>
</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
=== Install rgdal package on ubuntu ===
<pre>
sudo apt-get install libgdal1-dev libproj-dev
R
R
> install.packages("rgdal")
{{Pre}}
install.packages("rgdal")
</pre>
</pre>


=== Embedding R ===
== Install sf package ==
First make sure before 'make' R, R is configured with
I got the following error even I have installed some libraries.
<pre>
<pre>
./configure --enable-R-shlib
checking GDAL version >= 2.0.1... no
configure: error: sf is not compatible with GDAL versions below 2.0.1
</pre>
</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


Reference http://bioconductor.org/help/course-materials/2012/Seattle-Oct-2012/AdvancedR.pdf
sudo apt update
 
sudo apt-cache policy libgdal-dev # Make sure a version >= 2.0 appears
<pre>
mli@PhenomIIx6:~/Downloads/R-2.15.2/library/AdvancedR/embedding$ export R_HOME=/home/mli/Downloads/R-2.15.2
mli@PhenomIIx6:~/Downloads/R-2.15.2/library/AdvancedR/embedding$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/mli/Downloads/R-2.15.2/lib
mli@PhenomIIx6:~/Downloads/R-2.15.2/library/AdvancedR/embedding$ g++ embed.c -I/home/mli/Downloads/R-2.15.2/include -L/home/mli/Downloads/R-2.15.2/lib -lR


mli@PhenomIIx6:~/Downloads/R-2.15.2/library/AdvancedR/embedding$ R CMD ./a.out
sudo apt install libgdal-dev # works on ubuntu 20.04 too
WARNING: ignoring environment value of R_HOME
                            # no need the previous lines
</pre>


R version 2.15.2 (2012-10-26) -- "Trick or Treat"
== Database ==
Copyright (C) 2012 The R Foundation for Statistical Computing
* https://cran.r-project.org/web/views/Databases.html
ISBN 3-900051-07-0
* [http://blog.revolutionanalytics.com/2017/08/a-modern-database-interface-for-r.html A modern database interface for R]
Platform: x86_64-pc-linux-gnu (64-bit)


R is free software and comes with ABSOLUTELY NO WARRANTY.
=== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ===
You are welcome to redistribute it under certain conditions.
* https://cran.r-project.org/web/packages/RSQLite/vignettes/RSQLite.html
Type 'license()' or 'licence()' for distribution details.
* https://github.com/rstats-db/RSQLite


  Natural language support but running in an English locale
'''Creating a new database''':
{{Pre}}
library(DBI)


R is a collaborative project with many contributors.
mydb <- dbConnect(RSQLite::SQLite(), "my-db.sqlite")
Type 'contributors()' for more information and
dbDisconnect(mydb)
'citation()' on how to cite R or R packages in publications.
unlink("my-db.sqlite")


Type 'demo()' for some demos, 'help()' for on-line help, or
# temporary database
'help.start()' for an HTML browser interface to help.
mydb <- dbConnect(RSQLite::SQLite(), "")
Type 'q()' to quit R.
dbDisconnect(mydb)
</pre>


'''Loading data''':
{{Pre}}
mydb <- dbConnect(RSQLite::SQLite(), "")
dbWriteTable(mydb, "mtcars", mtcars)
dbWriteTable(mydb, "iris", iris)


ns> require(stats); require(graphics)
dbListTables(mydb)


ns> ns(women$height, df = 5)
dbListFields(con, "mtcars")
                1            2          3          4            5
[1,] 0.000000e+00 0.000000e+00  0.00000000 0.00000000  0.0000000000
[2,] 7.592323e-03 0.000000e+00 -0.08670223 0.26010669 -0.1734044626
[3,] 6.073858e-02 0.000000e+00 -0.15030440 0.45091320 -0.3006088020
[4,] 2.047498e-01 6.073858e-05 -0.16778345 0.50335034 -0.3355668952
[5,] 4.334305e-01 1.311953e-02 -0.13244035 0.39732106 -0.2648807067
[6,] 6.256681e-01 8.084305e-02 -0.07399720 0.22199159 -0.1479943948
[7,] 6.477162e-01 2.468416e-01 -0.02616007 0.07993794 -0.0532919575
[8,] 4.791667e-01 4.791667e-01  0.01406302 0.02031093 -0.0135406187
[9,] 2.468416e-01 6.477162e-01  0.09733619 0.02286023 -0.0152401533
[10,] 8.084305e-02 6.256681e-01  0.27076826 0.06324188 -0.0405213106
[11,] 1.311953e-02 4.334305e-01  0.48059836 0.12526031 -0.0524087186
[12,] 6.073858e-05 2.047498e-01  0.59541597 0.19899261  0.0007809246
[13,] 0.000000e+00 6.073858e-02  0.50097182 0.27551020  0.1627793975
[14,] 0.000000e+00 7.592323e-03  0.22461127 0.35204082  0.4157555879
[15,] 0.000000e+00 0.000000e+00 -0.14285714 0.42857143  0.7142857143
attr(,"degree")
[1] 3
attr(,"knots")
20%  40%  60%  80%
60.8 63.6 66.4 69.2
attr(,"Boundary.knots")
[1] 58 72
attr(,"intercept")
[1] FALSE
attr(,"class")
[1] "ns"    "basis"  "matrix"


ns> summary(fm1 <- lm(weight ~ ns(height, df = 5), data = women))
dbReadTable(con, "mtcars")
</pre>


Call:
'''Queries''':
lm(formula = weight ~ ns(height, df = 5), data = women)
{{Pre}}
dbGetQuery(mydb, 'SELECT * FROM mtcars LIMIT 5')


Residuals:
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < 4.6')
    Min      1Q  Median      3Q      Max
-0.38333 -0.12585  0.07083  0.15401  0.30426


Coefficients:
dbGetQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" < :x', params = list(x = 4.6))
                    Estimate Std. Error t value Pr(>|t|)
(Intercept)        114.7447    0.2338  490.88  < 2e-16 ***
ns(height, df = 5)1  15.9474    0.3699  43.12 9.69e-12 ***
ns(height, df = 5)2  25.1695    0.4323  58.23 6.55e-13 ***
ns(height, df = 5)3  33.2582    0.3541  93.93 8.91e-15 ***
ns(height, df = 5)4  50.7894    0.6062  83.78 2.49e-14 ***
ns(height, df = 5)5  45.0363    0.2784  161.75  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1


Residual standard error: 0.2645 on 9 degrees of freedom
res <- dbSendQuery(con, "SELECT * FROM mtcars WHERE cyl = 4")
Multiple R-squared: 0.9998,     Adjusted R-squared: 0.9997
dbFetch(res)
F-statistic:  9609 on 5 and 9 DF,  p-value: < 2.2e-16
</pre>


'''Batched queries''':
{{Pre}}
dbClearResult(rs)
rs <- dbSendQuery(mydb, 'SELECT * FROM mtcars')
while (!dbHasCompleted(rs)) {
  df <- dbFetch(rs, n = 10)
  print(nrow(df))
}


ns> ## example of safe prediction
dbClearResult(rs)
ns> plot(women, xlab = "Height (in)", ylab = "Weight (lb)")
</pre>


ns> ht <- seq(57, 73, length.out = 200)
'''Multiple parameterised queries''':
 
{{Pre}}
ns> lines(ht, predict(fm1, data.frame(height=ht)))
rs <- dbSendQuery(mydb, 'SELECT * FROM iris WHERE "Sepal.Length" = :x')
 
dbBind(rs, param = list(x = seq(4, 4.4, by = 0.1)))
ns> ## Don't show:
nrow(dbFetch(rs))
ns> ## Consistency:
#> [1] 4
ns> x <- c(1:3,5:6)
dbClearResult(rs)
 
</pre>
ns> stopifnot(identical(ns(x), ns(x, df = 1)),
ns+          identical(ns(x, df=2), ns(x, df=2, knots=NULL)),# not true till 2.15.2
ns+          !is.null(kk <- attr(ns(x), "knots")),# not true till 1.5.1
ns+          length(kk) == 0)


ns> ## End Don't show
'''Statements''':
ns>
{{Pre}}
ns>
dbExecute(mydb, 'DELETE FROM iris WHERE "Sepal.Length" < 4')
ns>
#> [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)
</pre>
</pre>


The above result can be compared with running
=== [https://cran.r-project.org/web/packages/sqldf/ sqldf] ===
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]


mli@PhenomIIx6:~/Downloads/R-2.15.2/library/AdvancedR/embedding$ R
=== [https://cran.r-project.org/web/packages/RPostgreSQL/index.html RPostgreSQL] ===
WARNING: ignoring environment value of R_HOME


R version 2.15.2 (2012-10-26) -- "Trick or Treat"
=== [[MySQL#Use_through_R|RMySQL]] ===
Copyright (C) 2012 The R Foundation for Statistical Computing
* http://datascienceplus.com/bringing-the-powers-of-sql-into-r/
ISBN 3-900051-07-0
* See [[MySQL#Installation|here]] about the installation of the required package ('''libmysqlclient-dev''') in Ubuntu.
Platform: x86_64-pc-linux-gnu (64-bit)


R is free software and comes with ABSOLUTELY NO WARRANTY.
=== MongoDB ===
You are welcome to redistribute it under certain conditions.
* http://www.r-bloggers.com/r-and-mongodb/
Type 'license()' or 'licence()' for distribution details.
* http://watson.nci.nih.gov/~sdavis/blog/rmongodb-using-R-with-mongo/


  Natural language support but running in an English locale
=== odbc ===


R is a collaborative project with many contributors.
=== RODBC ===
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.


Type 'demo()' for some demos, 'help()' for on-line help, or
=== DBI ===
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.


> library(splines)
=== [https://cran.r-project.org/web/packages/dbplyr/index.html dbplyr] ===
> example("ns")
* To use databases with dplyr, you need to first install dbplyr
* https://db.rstudio.com/dplyr/
* Five commonly used backends: RMySQL, RPostgreSQ, RSQLite, ODBC, bigrquery.
* http://www.datacarpentry.org/R-ecology-lesson/05-r-and-databases.html


ns> require(stats); require(graphics)
'''Create a new SQLite database''':
{{Pre}}
surveys <- read.csv("data/surveys.csv")
plots <- read.csv("data/plots.csv")


ns> ns(women$height, df = 5)
my_db_file <- "portal-database.sqlite"
                1            2          3          4            5
my_db <- src_sqlite(my_db_file, create = TRUE)
[1,] 0.000000e+00 0.000000e+00  0.00000000 0.00000000  0.0000000000
[2,] 7.592323e-03 0.000000e+00 -0.08670223 0.26010669 -0.1734044626
[3,] 6.073858e-02 0.000000e+00 -0.15030440 0.45091320 -0.3006088020
[4,] 2.047498e-01 6.073858e-05 -0.16778345 0.50335034 -0.3355668952
[5,] 4.334305e-01 1.311953e-02 -0.13244035 0.39732106 -0.2648807067
[6,] 6.256681e-01 8.084305e-02 -0.07399720 0.22199159 -0.1479943948
[7,] 6.477162e-01 2.468416e-01 -0.02616007 0.07993794 -0.0532919575
[8,] 4.791667e-01 4.791667e-01  0.01406302 0.02031093 -0.0135406187
[9,] 2.468416e-01 6.477162e-01  0.09733619 0.02286023 -0.0152401533
[10,] 8.084305e-02 6.256681e-01  0.27076826 0.06324188 -0.0405213106
[11,] 1.311953e-02 4.334305e-01  0.48059836 0.12526031 -0.0524087186
[12,] 6.073858e-05 2.047498e-01  0.59541597 0.19899261  0.0007809246
[13,] 0.000000e+00 6.073858e-02  0.50097182 0.27551020  0.1627793975
[14,] 0.000000e+00 7.592323e-03  0.22461127 0.35204082  0.4157555879
[15,] 0.000000e+00 0.000000e+00 -0.14285714 0.42857143  0.7142857143
attr(,"degree")
[1] 3
attr(,"knots")
20%  40%  60%  80%
60.8 63.6 66.4 69.2
attr(,"Boundary.knots")
[1] 58 72
attr(,"intercept")
[1] FALSE
attr(,"class")
[1] "ns"    "basis"  "matrix"


ns> summary(fm1 <- lm(weight ~ ns(height, df = 5), data = women))
copy_to(my_db, surveys)
copy_to(my_db, plots)
my_db
</pre>


Call:
'''Connect to a database''':
lm(formula = weight ~ ns(height, df = 5), data = women)
{{Pre}}
download.file(url = "https://ndownloader.figshare.com/files/2292171",
              destfile = "portal_mammals.sqlite", mode = "wb")


Residuals:
library(dbplyr)
    Min      1Q  Median      3Q      Max
library(dplyr)
-0.38333 -0.12585  0.07083  0.15401  0.30426
mammals <- src_sqlite("portal_mammals.sqlite")
</pre>


Coefficients:
'''Querying the database with the SQL syntax''':
                    Estimate Std. Error t value Pr(>|t|)
{{Pre}}
(Intercept)        114.7447    0.2338  490.88  < 2e-16 ***
tbl(mammals, sql("SELECT year, species_id, plot_id FROM surveys"))
ns(height, df = 5)1  15.9474    0.3699  43.12 9.69e-12 ***
</pre>
ns(height, df = 5)2  25.1695    0.4323  58.23 6.55e-13 ***
ns(height, df = 5)3  33.2582    0.3541  93.93 8.91e-15 ***
ns(height, df = 5)4  50.7894    0.6062  83.78 2.49e-14 ***
ns(height, df = 5)5  45.0363    0.2784  161.75  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1


Residual standard error: 0.2645 on 9 degrees of freedom
'''Querying the database with the dplyr syntax''':
Multiple R-squared: 0.9998,    Adjusted R-squared: 0.9997
{{Pre}}
F-statistic:  9609 on 5 and 9 DF, p-value: < 2.2e-16
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>


ns> ## example of safe prediction
'''Simple database queries''':
ns> plot(women, xlab = "Height (in)", ylab = "Weight (lb)")
{{Pre}}
surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight)
</pre>


ns> ht <- seq(57, 73, length.out = 200)
'''Laziness''' (instruct R to stop being lazy):
{{Pre}}
data_subset <- surveys %>%
  filter(weight < 5) %>%
  select(species_id, sex, weight) %>%
  collect()
</pre>


ns> lines(ht, predict(fm1, data.frame(height=ht)))
'''Complex database queries''':
{{Pre}}
plots <- tbl(mammals, "plots")
plots # # The plot_id column features in the plots table


ns> ## Don't show:
surveys # The plot_id column also features in the surveys table
ns> ## Consistency:
ns> x <- c(1:3,5:6)


ns> stopifnot(identical(ns(x), ns(x, df = 1)),
# Join databases method 1
ns+          identical(ns(x, df=2), ns(x, df=2, knots=NULL)),# not true till 2.15.2
plots %>%
ns+          !is.null(kk <- attr(ns(x), "knots")),# not true till 1.5.1
  filter(plot_id == 1) %>%
ns+          length(kk) == 0)
  inner_join(surveys) %>%
 
  collect()
ns> ## End Don't show
ns>
ns>
ns>
</pre>
</pre>


Note that if I follow the instruction to put embed.c at the end of g++ command, I will get an error.
=== NoSQL ===
<pre>
[https://ropensci.org/technotes/2018/01/25/nodbi/ nodbi: the NoSQL Database Connector]
mli@PhenomIIx6:~/Downloads/R-2.15.2/library/AdvancedR/embedding$ g++ -I/home/mli/Downloads/R-2.15.2/include -L/home/mli/Downloads/R-2.15.2/lib -lR embed.c
/tmp/cc7Vum5j.o: In function `main':
embed.c:(.text+0x1c): undefined reference to `Rf_initEmbeddedR'
embed.c:(.text+0x2b): undefined reference to `Rf_endEmbeddedR'
/tmp/cc7Vum5j.o: In function `doSplinesExample()':
embed.c:(.text+0x45): undefined reference to `Rf_mkString'
embed.c:(.text+0x52): undefined reference to `Rf_install'
embed.c:(.text+0x5d): undefined reference to `Rf_lang2'
embed.c:(.text+0x6d): undefined reference to `Rf_protect'
embed.c:(.text+0x74): undefined reference to `R_GlobalEnv'
embed.c:(.text+0x87): undefined reference to `R_tryEval'
embed.c:(.text+0x91): undefined reference to `Rf_unprotect'
embed.c:(.text+0x9b): undefined reference to `Rf_ScalarLogical'
embed.c:(.text+0xa8): undefined reference to `Rf_install'
embed.c:(.text+0xb3): undefined reference to `Rf_lang2'
embed.c:(.text+0xc3): undefined reference to `Rf_protect'
embed.c:(.text+0xcd): undefined reference to `Rf_install'
embed.c:(.text+0xdc): undefined reference to `CDR'
embed.c:(.text+0xe7): undefined reference to `SET_TAG'
embed.c:(.text+0xee): undefined reference to `R_GlobalEnv'
embed.c:(.text+0x102): undefined reference to `R_tryEval'
embed.c:(.text+0x10c): undefined reference to `Rf_unprotect'
embed.c:(.text+0x116): undefined reference to `Rf_mkString'
embed.c:(.text+0x123): undefined reference to `Rf_install'
embed.c:(.text+0x12e): undefined reference to `Rf_lang2'
embed.c:(.text+0x13e): undefined reference to `Rf_protect'
embed.c:(.text+0x145): undefined reference to `R_GlobalEnv'
embed.c:(.text+0x158): undefined reference to `R_tryEval'
embed.c:(.text+0x162): undefined reference to `Rf_unprotect'
collect2: ld returned 1 exit status
</pre>


=== Set up Emacs on Windows ===
== Github ==
Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
<pre>
(setq-default inferior-R-program-name
              "c:/program files/r/r-2.15.2/bin/i386/rterm.exe")
</pre>


=== Database ===
=== R source  ===
==== [[MySQL#Use_through_R|RMySQL]] ====
https://github.com/wch/r-source/  Daily update, interesting, should be visited every day. Clicking '''1000+ commits''' to look at daily changes.


==== [http://cran.r-project.org/web/packages/RSQLite/index.html RSQLite] ====
If we are interested in a certain branch (say 3.2), look for R-3-2-branch.
[http://blog.sobbayi.com/sqlite-vs-mysql-how-to-decide-which-to-use/ Not suitable] for client/server architecture. The limit is quite large; see [http://sqlite.org/limits.html here].


=== Github ===
=== R packages (only) source (metacran) ===
* https://github.com/cran/ by [https://github.com/gaborcsardi Gábor Csárdi], the author of '''[http://igraph.org/ igraph]''' software.


==== R source ====
=== Bioconductor packages source ===
* https://github.com/wch/r-source/ (daily update, interesting, should be visited every day)
<strike>[https://stat.ethz.ch/pipermail/bioc-devel/2015-June/007675.html Announcement], https://github.com/Bioconductor-mirror </strike>
* https://github.com/SurajGupta/r-source  (update for each R release)


==== github ====
=== Send local repository to Github in R by using reports package ===
https://github.com/languages/R
http://www.youtube.com/watch?v=WdOI_-aZV0Y


==== My collection ====
=== My collection ===
* https://github.com/arraytools
* https://github.com/arraytools
* https://gist.github.com/4383351 heatmap using leukemia data
* https://gist.github.com/4383351 heatmap using leukemia data
Line 1,343: Line 1,926:
* https://gist.github.com/4484270 biocLite  
* https://gist.github.com/4484270 biocLite  


==== How to download ====
=== How to download ===


Clone ~ Download.  
Clone ~ Download.  
Line 1,372: Line 1,955:
</pre>
</pre>


== Tricks ==
=== Jekyll ===
[http://statistics.rainandrhino.org/2015/12/15/jekyll-r-blogger-knitr-hyde.html An Easy Start with Jekyll, for R-Bloggers]
 
== Connect R with Arduino ==
* https://zhuhao.org/post/connect-arduino-chips-with-r/
* http://lamages.blogspot.com/2012/10/connecting-real-world-to-r-with-arduino.html
* http://jean-robert.github.io/2012/11/11/thermometer-R-using-Arduino-Java.html
* http://bio7.org/?p=2049
* http://www.rforge.net/Arduino/svn.html
 
== Android App ==
* [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]
 
== Common plots tips ==
=== Create an empty plot ===
'''plot.new()'''   
 
=== Overlay plots ===
[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>
 
=== 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 [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'''.
* 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>
 
=== Grouped boxplots ===
* [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)
 
=== [https://www.samruston.co.uk/ Weather Time Line] ===
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].
 
=== 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>
 
[[:File:Ggplot2bar.svg]]
 
=== 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]
 
Use text().
 
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].
 
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.
 
=== Grouped barplots ===
* https://www.r-graph-gallery.com/barplot/, https://www.r-graph-gallery.com/48-grouped-barplot-with-ggplot2/ (simpliest, no error bars)
{{Pre}}
library(ggplot2)
# 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)
 
=== Math expression ===
* [https://www.rdocumentation.org/packages/grDevices/versions/3.5.0/topics/plotmath ?plotmath]
* https://stackoverflow.com/questions/4973898/combining-paste-and-expression-functions-in-plot-labels
* Some cases
** 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")
 
# 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")
</pre>
 
=== Impose a line to a scatter plot ===
* abline + lsfit # least squares
{{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>
 
=== 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]


=== Editor ===
=== 3D scatterplot ===
* [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'''.
* [[R_web#plotly|R web > plotly]]
 
=== Rotating x axis labels for barplot ===
https://stackoverflow.com/questions/10286473/rotating-x-axis-labels-in-r-for-barplot
{{Pre}}
barplot(mytable,main="Car makes",ylab="Freqency",xlab="make",las=2)
</pre>
 
=== 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
{{Pre}}
plot(1:10, rnorm(10), ylim=c(0,10), yaxs="i")
</pre>
 
=== Different colors of axis labels in barplot ===
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]
 
Method 1: Append labels for the 2nd, 3rd, ... color gradually because 'col.axis' argument cannot accept more than one color.
{{Pre}}
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")
</pre>
 
Method 2: text() which can accept multiple colors in 'col' parameter but we need to find out the (x, y) by ourselves.
{{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)
</pre>
 
=== Use text() to draw labels on X/Y-axis including rotation ===
* 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]
* [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).
** par("usr") is determined *after* a plot has been created
** [http://sphaerula.com/legacy/R/placingTextInPlots.html Example of using the "usr" parameter]
* https://datascienceplus.com/building-barplots-with-error-bars/
{{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>
* 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
 
=== Include labels on the top axis/margin: axis() and mtext() ===
<pre>
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
</pre>
See also [[#15_Questions_All_R_Users_Have_About_Plots| 15_Questions_All_R_Users_Have_About_Plots]]
 
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].
 
=== Legend tips ===
[https://r-coder.com/add-legend-r/ Add legend to a plot in R]
 
[https://stackoverflow.com/a/36842578 Increase/decrease legend font size] '''cex''' & [[Ggplot2#Legend_size|ggplot2]] package case.
{{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>
 
'''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>
legend("bottomright", inset=.05, )
</pre>
 
'''legend without a box'''
<pre>
legend(, bty = "n")
</pre>
 
'''Add a legend title'''
<pre>
legend(, title = "")
</pre>
 
[https://stackoverflow.com/a/60971923 Add a common legend to multiple plots]. Use the layout function.
 
=== Superimpose a density plot or any curves ===
Use '''lines()'''.
 
Example 1
{{Pre}}
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)
</pre>
 
Example 2
{{Pre}}
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
</pre>
 
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 ===
[https://rcrastinate.rbind.io/post/using-custom-scales-with-the-scales-package/ Using custom scales with the 'scales' package]
 
== Time series ==
* [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]
 
=== Time series stock price plot ===
* http://blog.revolutionanalytics.com/2015/08/plotting-time-series-in-r.html (ggplot2, xts, [https://rstudio.github.io/dygraphs/ dygraphs])
* [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
 
{{Pre}}
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))
</pre>
 
=== tidyquant: Getting stock data ===
[http://varianceexplained.org/r/stock-changes/ The 'largest stock profit or loss' puzzle: efficient computation in R]
 
=== Timeline plot ===
* https://stackoverflow.com/questions/20695311/chronological-timeline-with-points-in-time-and-format-date
* [https://github.com/shosaco/vistime vistime] - Pretty Timelines in R
 
=== Clockify ===
[https://datawookie.dev/blog/2021/09/clockify-time-tracking-from-r/ Clockify]
 
== 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.
 
== 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]
 
== Text mining ==
* [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].
 
== World map ==
[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)
 
== Diagram/flowchart/Directed acyclic diagrams (DAGs) ==
* [https://finnstats.com/index.php/2021/06/29/transition-plot-in-r-change-in-time-visualization/ Transition plot in R-change in time visualization]
 
=== [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]
 
=== [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
 
=== Gmisc ===
[http://gforge.se/2020/08/easy-flowchart/ Easiest flowcharts eveR?]
 
=== Concept Maps ===
[https://github.com/rstudio/concept-maps/ concept-maps] where the diagrams are generated from https://app.diagrams.net/.
 
=== 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]
 
== Venn Diagram ==
[[Venn_diagram|Venn diagram]]
 
== hexbin plot ==
* [https://datasciencetut.com/how-to-create-a-hexbin-chart-in-r/ How to create a hexbin chart in R]
* [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.
 
== Bump chart/Metro map ==
https://dominikkoch.github.io/Bump-Chart/
 
== Amazing/special plots ==
See [[Amazing_plot|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 ==
* [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.
 
== Amazon Alexa ==
* http://blagrants.blogspot.com/2016/02/theres-party-at-alexas-place.html
 
== 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 ==
[http://blog.revolutionanalytics.com/2017/07/secret-package.html Securely store API keys in R scripts with the "secret" package]
 
== Credentials and secrets ==
[https://datascienceplus.com/how-to-manage-credentials-and-secrets-safely-in-r/ How to manage credentials and secrets safely in R]
 
== 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]
 
=== getPass ===
[https://cran.r-project.org/web/packages/getPass/README.html getPass]
 
== Vision and image recognition ==
* 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
 
== Creating a Dataset from an Image ==
[https://ivelasq.rbind.io/blog/reticulate-data-recreation/ Creating a Dataset from an Image in R Markdown using reticulate]
 
== Turn pictures into coloring pages ==
https://gist.github.com/jeroen/53a5f721cf81de2acba82ea47d0b19d0
 
== Numerical optimization ==
[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]
 
* [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]
* [http://stat.ethz.ch/R-manual/R-patched/library/stats/html/optimize.html optimize]: One Dimensional Optimization
* [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
* [https://blogs.rstudio.com/ai/posts/2021-04-22-torch-for-optimization/ torch for optimization]. L-BFGS optimizer.
 
== 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
 
= R packages =
[[R_packages|R packages]]
 
= Tricks =
 
== Getting help ==
* 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/
 
== Better Coder/coding, best practices ==
* 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
 
== [https://en.wikipedia.org/wiki/Scientific_notation#E-notation E-notation] ==
6.022E23 (or 6.022e23) is equivalent to 6.022×10^23
 
== Getting user's home directory ==
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"
 
# 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"
</pre>
 
== 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.
 
== Distinguish Windows and Linux/Mac, R.Version() ==
identical(.Platform$OS.type, "unix") returns TRUE on Mac and Linux.
 
* [https://www.r-bloggers.com/identifying-the-os-from-r/ Identifying the OS from R]
* [https://stackoverflow.com/questions/4747715/how-to-check-the-os-within-r How to check the OS within R]
<pre>
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)
}
</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>
 
== 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]
 
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
<pre>
R_LIBS_SITE=F:/R/library
</pre>
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].
 
=== 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 [https://gist.github.com/arraytools/ed16a486e19702ae94bde4212ad59ecb 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() ==
<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>
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")
</pre>
<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'''
 
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>
 
'''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.
<pre>
if (interactive()) {
  .Last <- function() try(savehistory(file.path(Sys.getenv("HOME"), ".Rhistory")))
}
</pre>
 
== Disable "Save workspace image?" prompt when exit R? ==
[https://stackoverflow.com/a/4996252 How to disable "Save workspace image?" prompt in R?]
 
== 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.
 
== getRversion() ==
<pre>
getRversion()
[1] ‘4.3.0’
</pre>
 
== 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.
 
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
</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"
 
> length(system('tasklist /FI "IMAGENAME eq Rgui.exe" ', intern = TRUE))-3
</pre>
 
== Editor ==
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs
http://en.wikipedia.org/wiki/R_(programming_language)#Editors_and_IDEs


* [http://www.rstudio.com/ Rstudio] - editor/R terminal/R graphics/file browser/package manager.
<ul>
* [http://www.geany.org/ geany] - I like the feature that it shows defined functions on the side panel even for R code.
<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).
* Edit the file ''C:\Program Files\GNU Emacs 23.2\site-lisp\site-start.el'' with something like
<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].
* [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
* 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


=== Create R package from R code with roxyPackage ===
== GUI for Data Analysis ==
http://lamages.blogspot.com/2013/03/create-r-package-from-single-r-file.html
[https://www.r-bloggers.com/2023/06/update-to-data-science-software-popularity/ Update to Data Science Software Popularity] 6/7/2023
 
=== 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]
 
=== 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 ===
* https://www.jamovi.org/
* [http://r4stats.com/2019/01/09/updated-review-jamovi/ Updated Review: jamovi User Interface to R]
 
== Scope ==
See
* [http://cran.r-project.org/doc/manuals/R-intro.html#Assignment-within-functions Assignments within functions] in the '''An Introduction to R''' manual.
 
=== source() ===
* [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()''')
 
{{Pre}}
## 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()
</pre>
 
'''Note that any ordinary assignments done within the function are local and temporary and are lost after exit from the function.'''
 
Example 1.
<pre>
> 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
</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.
 
Other resource: [http://adv-r.had.co.nz/Functions.html Advanced R] by Hadley Wickham.
 
Example 3. [https://stackoverflow.com/questions/1169534/writing-functions-in-r-keeping-scoping-in-mind Writing functions in R, keeping scoping in mind]
 
=== New environment ===
* 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!
 
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>
 
Just look at the contents of rda file without saving to anywhere (?load)
<pre>
local({
  load("myfile.rda")
  ls()
})
</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.
{{Pre}}
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")
</pre>
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>
myEnv <- new.env()   
source("some_other_script.R", local=myEnv)
attach(myEnv, name="sourced_scripts")
search()
ls(2)
ls(myEnv)
with(myEnv, print(x))
</pre>
 
=== 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 [https://www.rdocumentation.org/packages/utils/versions/3.6.1/topics/str ?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 [[Tidyverse#broom|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
{{Pre}}
> 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"
</pre>
 
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 ==
* [http://datascienceplus.com/strategies-to-speedup-r-code/ Strategies to speedup R code] from DataScience+
 
=== Profiler ===
* [https://www.rstudio.com/resources/videos/understand-code-performance-with-the-profiler/ Understand Code Performance with the profiler] (Video)
* [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]
 
== && vs & ==
See https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/Logic.


=== llply() from plyr package ===
* 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 &]
 
<pre>
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
</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))
 
if (!is.null(exprTest) && any(is.na(exprTest))) { ... }
</pre>
 
== for-loop, control flow ==
* [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]
 
== Vectorization ==
* [https://en.wikipedia.org/wiki/Vectorization_%28mathematics%29 Vectorization (Mathematics)] from wikipedia
* [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://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]
 
=== lapply vs for loop ===
* [https://stackoverflow.com/a/42440872 lapply vs for loop - Performance R]
* https://code-examples.net/en/q/286e03a
* [https://johanndejong.wordpress.com/2016/07/07/r-are-apply-loops-faster-than-for-loops/ R: are *apply loops faster than for loops?]
 
=== [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/split split()] and sapply() ===
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?]
* 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://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
 
# bigmemory vignette
planeindices <- split(1:nrow(x), x[,'TailNum'])
planeStart <- sapply(planeindices,
                    function(i) birthmonth(x[i, c('Year','Month'),
                                            drop=FALSE]))
</pre>
* Split rows of a data frame/matrix; e.g. rows represents genes. The data frame/matrix is split directly.
{{Pre}}
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
</pre>
* 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>
* 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.
* 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()].
{{Pre}}
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
</pre>
 
=== strsplit and sapply ===
{{Pre}}
> 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"
</pre>
 
=== Mean of duplicated columns: rowMeans; compute Means by each row ===
<ul>
<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 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
 
# 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)))
</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
 
apply(x, 1, mean, na.rm=T)
# [1] 31 27 28 29 30 31 32 33 34 35
</pre>
</li>
<li>[https://cran.r-project.org/web/packages/matrixStats/index.html matrixStats]: Functions that Apply to Rows and Columns of Matrices (and to Vectors)
</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>
 
=== Mean of duplicated rows: colMeans and rowsum ===
<ul>
<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'''.
{{Pre}}
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
</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]]
 
</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.''
{{Pre}}
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
</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)
 
# 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
</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
 
[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.
 
Some short examples:
* [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?]
 
=== Apply vs for loop ===
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`?]
 
=== Progress bar ===
[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?]
 
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://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]
 
=== simplify option in sapply() ===
<pre>
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>
 
=== 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>
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)
</pre>
</li>
<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.
<pre>
z <- mapply(function(u, v) { max(u, v) },
            u = x[, 1], v = x[, 2])
</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, ...)
 
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(xs[[i]], ws[[i]])
})
</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, ...)
 
> 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
</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]
 
=== sapply & vapply ===
* [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.
 
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].
 
=== 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].
 
=== replicate ===
https://www.datacamp.com/community/tutorials/tutorial-on-loops-in-r
{{Pre}}
> 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
</pre>
 
See [[#parallel_package|parSapply()]] for a parallel version of replicate().
 
=== 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
 
[[2]]
[1] 2 2 2
 
[[3]]
[1] 3 3
 
[[4]]
[1] 4
</pre>
* [http://biolitika.si/vectorizing-functions-in-r-is-easy.html Vectorizing functions in R is easy]
{{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
 
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
</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.
 
[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.
 
# plyr has a common syntax -- easier to remember
# plyr requires less code since it takes care of the input and output format
# plyr can easily be run in parallel -- faster
 
Tutorials
* [http://dplyr.tidyverse.org/articles/dplyr.html Introduction to dplyr] from http://dplyr.tidyverse.org/.
* A video of [http://cran.r-project.org/web/packages/dplyr/index.html dplyr] package can be found on [http://vimeo.com/103872918 vimeo].
* [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.
 
Examples of using dplyr:
* [http://wiekvoet.blogspot.com/2015/03/medicines-under-evaluation.html Medicines under evaluation]
* [http://rpubs.com/seandavi/GEOMetadbSurvey2014 CBI GEO Metadata Survey]
* [http://datascienceplus.com/r-for-publication-by-page-piccinini-lesson-3-logistic-regression/ Logistic Regression] by Page Piccinini. mutate(), inner_join() and %>%.
* [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] ===
'''Tibbles''' are data frames, but slightly tweaked to work better in the '''tidyverse'''.
 
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
 
Tibbles [https://cran.r-project.org/web/packages/tibble/vignettes/tibble.html Vignette]
 
{{Pre}}
> data(pew, package = "efficient")
> dim(pew)
[1] 18 10
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"    "tbl"        "data.frame"
 
> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
# A tibble: 162 x 3
                                                      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
</pre>
 
To show all rows of a tibble object, use the '''print()''' method.
<pre>
print(tbObj, n= Inf)
 
tbObj %>% print(n= nrow(.))
</pre>
 
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 [https://stackoverflow.com/questions/21618423/extract-a-dplyr-tbl-column-as-a-vector extract a column from a tibble object], use '''[[''' or '''$''' or dplyr::pull(). [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ Select Data Frame Columns in R].
{{Pre}}
TibbleObject$VarName
# OR
TibbleObject[["VarName"]]
# OR
pull(TibbleObject, VarName) # won't be a tibble object anymore
 
dplyr::select(TibbleObject, -c(VarName1, VarName2)) # still a tibble object
# OR
dplyr::select(TibbleObject, 2:5) #
</pre>
 
'''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]
<pre>
my_data <- as_tibble(iris)
class(my_data)
</pre>
 
To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)
 
=== 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.
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>
Line 1,396: Line 3,471:
returns a list of 49 GOs.
returns a list of 49 GOs.


=== mclapply() from paralle package is a mult-core version of lapply() ===
=== ddply() ===
Note that Windows OS can not take advantage of it.
http://lamages.blogspot.com/2012/06/transforming-subsets-of-data-in-r-with.html
 
=== ldply() ===
[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]
 
=== Performance/speed comparison ===
[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]
 
== 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
{{Pre}}
get_seed <- function() {
  sample.int(.Machine$integer.max, 1)
}
</pre>
Note: .Machine$integer.max = 2147483647 = 2^31 - 1.


Another choice for Windows OS is to use parLapply() function in parallel package.
=== 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 [https://stat.ethz.ch/R-manual/R-patched/library/base/html/Random.html ?Random].  
<pre>
<pre>
ncores <- as.integer( Sys.getenv('NUMBER_OF_PROCESSORS') )
set.seed(as.numeric(Sys.time()))
cl <- makeCluster(getOption("cl.cores", ncores))
 
LLID2GOIDs2 <- parLapply(cl, rLLID, function(x) {
set.seed(as.numeric(Sys.Date())) # same seed for each day
                                    library(org.Hs.eg.db); get("org.Hs.egGO")[[x]]}  
</pre>
                        )
 
stopCluster(cl)
=== .Machine and the largest integer, double ===
See [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/.Machine ?.Machine].  
{{Pre}}
                          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
</pre>
</pre>
It does work. Cut the computing time from 100 sec to 29 sec on 4 cores.


=== regular expression ===
=== NA when overflow ===
* ?regexpr in R
<pre>
* http://biostat.mc.vanderbilt.edu/wiki/pub/Main/SvetlanaEdenRFiles/regExprTalk.pdf
tmp <- 156287L
* http://www.johndcook.com/r_language_regex.html
tmp*tmp
* http://en.wikibooks.org/wiki/R_Programming/Text_Processing#Regular_Expressions
# [1] NA
* http://www.endmemo.com/program/R/grep.php
# Warning message:
* http://ucfagls.wordpress.com/2012/08/15/processing-sample-labels-using-regular-expressions-in-r/
# In tmp * tmp : NAs produced by integer overflow
* http://www.dummies.com/how-to/content/how-to-use-regular-expressions-in-r.html
.Machine$integer.max
* http://www.r-bloggers.com/example-8-27-using-regular-expressions-to-read-data-with-variable-number-of-words-in-a-field/
# [1] 2147483647
* http://www.r-bloggers.com/using-regular-expressions-in-r-case-study-in-cleaning-a-bibtex-database/
</pre>
* http://cbio.ensmp.fr/~thocking/papers/2011-08-16-directlabels-and-regular-expressions-for-useR-2011/2011-useR-named-capture-regexp.pdf
* http://stackoverflow.com/questions/5214677/r-find-the-last-dot-in-a-string
* http://stackoverflow.com/questions/10294284/remove-all-special-characters-from-a-string-in-r


Not specific to R
== How to select a seed for simulation or randomization ==
* http://www.autohotkey.com/docs/misc/RegEx-QuickRef.htm
* [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]
* http://opencompany.org/download/regex-cheatsheet.pdf
* [https://www.makeuseof.com/tag/lesson-gamers-rng/ What Is RNG? A Lesson for Gamers ]
 
== set.seed() allow alphanumeric seeds ==
https://stackoverflow.com/a/10913336
 
== set.seed(), for loop and saving random seeds ==
<ul>
<li>[https://www.jottr.org/2020/09/21/detect-when-the-random-number-generator-was-used/ Detect When the Random Number Generator Was Used]
<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!
 
.Random.seed <- seeds[[23]]  # restore
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].
 
== 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
 
# 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
</pre>
 
=== Getting different results with set.seed() in RStudio ===
[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().''
 
=== dplyr::sample_n() ===
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.
 
== Regular Expression ==
See [[Regular_expression|here]].
 
== Read rrd file ==
* https://en.wikipedia.org/wiki/RRDtool
* http://oss.oetiker.ch/rrdtool/
* https://github.com/pldimitrov/Rrd
* http://plamendimitrov.net/blog/2014/08/09/r-package-for-working-with-rrd-files/
 
== 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.
<ul>
<li>Database connections
<pre>
library(RSQLite)
sqlite_get_query <- function(db, sql) {
  conn <- dbConnect(RSQLite::SQLite(), db)
  on.exit(dbDisconnect(conn), add = TRUE)
  dbGetQuery(conn, sql)
}
</pre>
<li>File connections
<pre>
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>


Example
== file, connection ==
* grep("\\.zip$", pkgs) or grep("\\.tar.gz$", pkgs)
* [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)
* read() and write()
* 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)


=== Clipboard ===
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>
<pre>
con <- gzfile(FileName, "r") # Or gzfile(FileName, open = 'r')
x <- read.delim(con)
close(x)
</pre>
=== withr package ===
https://cran.r-project.org/web/packages/withr/index.html . Reverse suggested by [https://cran.r-project.org/web/packages/languageserver/index.html languageserver].
== Clipboard (?connections), textConnection(), pipe() ==
<ul>
<li>On Windows, we can use readClipboard() and writeClipboard().
{{Pre}}
source("clipboard")
source("clipboard")
read.table("clipboard")
read.table("clipboard")
</pre></li>
<li>Clipboard -> R. Reading/writing clipboard on macOS. Use [https://www.rdocumentation.org/packages/base/versions/3.5.0/topics/textConnection textConnection()] function:
{{Pre}}
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'
</pre>
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().
</li>
<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
* pbpaste: takes data from the clipboard buffer and writes it to the standard output
{{Pre}}
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)
</pre>
<li>
<li>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.
</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>
=== clipr ===
[https://cran.rstudio.com/web/packages/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 ==
* [https://www.gastonsanchez.com/r4strings/ Handling Strings with R](ebook) by Gaston Sanchez.
* [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.'''
* [http://theautomatic.net/2019/05/17/four-ways-to-reverse-a-string-in-r/ Four ways to reverse a string in R]
* [https://statisticaloddsandends.wordpress.com/2022/05/05/a-short-note-on-the-startswith-function/ A short note on the startsWith function]
=== format(): padding with zero ===
<pre>
ngenes <- 10
genenames <- paste0("bm", gsub(" ", "0", format(1:ngenes))); genenames
#  [1] "bm01" "bm02" "bm03" "bm04" "bm05" "bm06" "bm07" "bm08" "bm09" "bm10"
</pre>
=== noquote() ===
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/noquote noqute] Print character strings without quotes.
=== stringr package ===
* https://stringr.tidyverse.org/index.html
* [https://stringr.tidyverse.org/articles/from-base.html Vignette compares stringr functions to their base R equivalents]
* 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].
=== glue package ===
<ul>
<li>[https://cran.r-project.org/web/packages/glue/index.html glue]. Useful in a loop and some function like ggtitle() or ggsave().
<pre>
library(glue)
name <- "Fred"
glue('My name is {name}.')  # My name is Fred.
</pre>
</li>
<li>[https://en.wikipedia.org/wiki/String_interpolation String interpolation] </li>
</ul>
=== 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>
=== number of characters limit ===
[https://twitter.com/eddelbuettel/status/1438326822635180036 It's a limit on a (single) input line in the REPL]
== 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
[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)
== 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 [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.
<pre>
url <- paste0("ftp://ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/All/",
              "GCF_000001405.13.assembly.txt")
f1 <- tempfile()
download.file(url, f1)
</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.
The following R command will show the exact svn revision for the R you are currently using.
<pre>
R.Version()$"svn rev"
</pre>
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 [https://github.com/wch/r-source/commits/trunk/src/library/utils/R/windows/sysutils.R R> src> library> utils > R > windows > sysutils.R].
'''R up to 3.2.2'''
<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).
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'''
<pre>
setInternet2 <- function(use = TRUE) {
    if(!is.na(use)) stop("use != NA is defunct")
    NA
}
</pre>
</pre>


=== read/download/source a file from internet ===
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.
* Simple text file
 
== 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 [https://stat.ethz.ch/R-manual/R-devel/library/base/html/is.finite.html ?is.finite].
 
[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]
 
== replace() function ==
* [https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/replace replace](vector, index, values)
* https://stackoverflow.com/a/11811147
 
== 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()
<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.
 
== read/download/source a file from internet ==
=== Simple text file http ===
<pre>
<pre>
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
retail <- read.csv("http://robjhyndman.com/data/ausretail.csv",header=FALSE)
</pre>
</pre>


* Zip file
=== Zip, RData, gz file and url() function ===
<pre>
x <- read.delim(gzfile("filename.txt.gz"), nrows=10)
</pre>
<pre>
<pre>
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
Line 1,449: Line 3,925:
close(con)
close(con)
</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.
Another example is [https://stackoverflow.com/a/9548672 Read gzipped csv directly from a url in R]
<pre>
con <- gzcon(url(paste("http://dumps.wikimedia.org/other/articlefeedback/",
                      "aa_combined-20110321.csv.gz", sep="")))
txt <- readLines(con)
dat <- read.csv(textConnection(txt))
</pre>
Another example of using url() is
<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].
'''Dropbox''' is easy and works for load(), wget, ...
[https://stackoverflow.com/a/46875562 R download .RData] or [https://stackoverflow.com/a/56670130 Directly loading .RData from github] from Github.


* Google drive file based on https
=== zip function ===
This will include 'hallmarkFiles' root folder in the files inside zip.
<pre>
<pre>
zip(zipfile = 'myFile.zip',
    files = dir('hallmarkFiles', full.names = TRUE))
# Verify/view the files. 'list = TRUE' won't extract
unzip('testZip.zip', list = TRUE)
</pre>
=== [http://cran.r-project.org/web/packages/downloader/index.html 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 [http://www.omegahat.org/RCurl/FAQ.html RCurl] package ===
{{Pre}}
require(RCurl)
require(RCurl)
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
myCsv <- getURL("https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0AkuuKBh0jM2TdGppUFFxcEdoUklCQlJhM2kweGpoUUE&single=true&gid=0&output=csv")
Line 1,457: Line 3,966:
</pre>
</pre>


* Github files https
=== Google sheet file using [https://github.com/jennybc/googlesheets googlesheets] package ===
[http://www.opiniomics.org/reading-data-from-google-sheets-into-r/ Reading data from google sheets into R]
 
=== 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>
x = getURL("https://gist.github.com/arraytools/6671098/raw/c4cb0ca6fe78054da8dbe253a05f7046270d5693/GeneIDs.txt",
            ssl.verifypeer = FALSE)
read.table(text=x)
</pre>
* [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.
 
=== skimr: A frictionless, pipeable approach to dealing with summary statistics ===
[https://ropensci.org/blog/2017/07/11/skimr/ skimr for useful and tidy summary statistics]


http://support.rstudio.org/help/kb/faq/configuring-r-to-use-an-http-proxy
=== modelsummary ===
[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


=== broom ===
[[Tidyverse#broom|Tidyverse->broom]]


=== Create publication tables using '''tables''' package ===
=== Create publication tables using '''tables''' package ===
See p13 for example in http://www.ianwatson.com.au/stata/tabout_tutorial.pdf
See p13 for example at [http://www.ianwatson.com.au/stata/tabout_tutorial.pdf#page=13 here]


R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,  
R's [http://cran.r-project.org/web/packages/tables/index.html tables] packages is the best solution. For example,  
<pre>
{{Pre}}
> library(tables)
> library(tables)
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
> tabular( (Species + 1) ~ (n=1) + Format(digits=2)*
Line 1,506: Line 4,036:
</pre>
</pre>


See also a collection of R packages related to reproducible research in http://cran.r-project.org/web/views/ReproducibleResearch.html
=== fgsea example ===
[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]
 
=== (archived) ClinReport: Statistical Reporting in Clinical Trials ===
https://cran.r-project.org/web/packages/ClinReport/index.html


=== Create flat tables in R console using ftable() ===
== Append figures to PDF files ==
[https://stackoverflow.com/a/13274272 How to append a plot to an existing pdf file]. Hint: use the recordPlot() function.
 
== Save base graphics as pseudo-objects ==
[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.
<pre>
<pre>
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
</pre>
== Extracting tables from PDFs ==
<ul>
<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'''.
</li>
<li>
[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
pdf_data(pdf_file)[[1]]  # data frame/tibble format
</pre>
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.
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]
</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>
</li>
<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.
<li>I found it is easier to use copy the column (it works) from PDF and paste them to Excel </li>
</ul>
== Print tables ==
=== addmargins() ===
* [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?]
=== tableone ===
* https://cran.r-project.org/web/packages/tableone/
* [https://datascienceplus.com/table-1-and-the-characteristics-of-study-population/ Table 1 and the Characteristics of Study Population]
* [https://www.jianshu.com/p/e76f2b708d45 如何快速绘制论文的表1(基本特征三线表)?]
* See Table 1 from [https://boiled-data.github.io/ClassificationDiabetes.html Tidymodels Machine Learning: Diabetes Classification]
=== Some examples ===
Cox models
* [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]
=== finalfit package ===
[https://finalfit.org/index.html summary_factorlist()] from the finalfit package.
=== table1 ===
* https://cran.r-project.org/web/packages/table1/
* [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.
=== gtsummary ===
* [https://education.rstudio.com/blog/2020/07/gtsummary/ Presentation-Ready Summary Tables with gtsummary]
* [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.
=== gt ===
[https://www.r-bloggers.com/2024/02/introduction-to-clinical-tables-with-the-gt-package/ Introduction to Clinical Tables with the {gt} Package]
=== 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) ===
[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.
<pre>
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
</pre>
=== stats::xtabs() ===
=== stats::ftable() ===
{{Pre}}
> ftable(Titanic, row.vars = 1:3)
> ftable(Titanic, row.vars = 1:3)
                   Survived  No Yes
                   Survived  No Yes
Line 1,600: Line 4,240:
</pre>
</pre>


=== Handling length 2^31 and more in R 3.0.0 ===
== 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).
 
== Handling length 2^31 and more in R 3.0.0 ==


From R News for 3.0.0 release:
From R News for 3.0.0 release:
Line 1,632: Line 4,287:
# 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.
# 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.
# 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].
# 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 ===
== NA in index ==
* Question: what is seq(1, 3)[c(1, 2, NA)]?
* Question: what is seq(1, 3)[c(1, 2, NA)]?


Line 1,661: Line 4,317:
Don't just use x != "" OR !is.na(x).
Don't just use x != "" OR !is.na(x).


=== Serialization ===
=== Some functions ===
* X %>% [https://tidyr.tidyverse.org/reference/drop_na.html tidyr::drop_na()]
* '''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)]
 
== Constant and 'L' ==
Add 'L' after a constant. For example,
{{Pre}}
for(i in 1L:n) { }
 
if (max.lines > 0L) { }
 
label <- paste0(n-i+1L, ": ")
 
n <- length(x);  if(n == 0L) { }
</pre>
 
== 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>
* https://dplyr.tidyverse.org/reference/case_when.html
* [https://rpubs.com/DaveRosenman/ifelsealternative Using dplyr’s mutate and case_when functions as alternative for if else statement]
* [http://www.datasciencemadesimple.com/case-statement-r-using-case_when-dplyr/ Case when in R using case_when() Dplyr – case_when in R]
* [https://predictivehacks.com/how-to-convert-continuous-variables-into-categorical-by-creating-bins/ How To Convert Continuous Variables Into Categorical By Creating Bins]
<ul>
<li>[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/cut ?cut]
{{Pre}}
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
</pre>
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]'''.
<pre>
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()
</pre>
<li>[https://www.spsanderson.com/steveondata/posts/2024-03-20/index.html A Guide to Using the cut() Function in R]
<li>[https://youtu.be/7oyiPBjLAWY?t=2480 tibble object]
{{Pre}}
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>
 
=== How to change one of the level to NA ===
https://stackoverflow.com/a/25354985. Note that the factor level is removed.
<pre>
x <- factor(c("a", "b", "c", "NotPerformed"))
levels(x)[levels(x) == 'NotPerformed'] <- NA
</pre>
 
[https://webbedfeet.netlify.app/post/creating-missing-values-in-factors/ Creating missing values in factors]
 
=== Concatenating two factor vectors ===
Not trivial. [https://stackoverflow.com/a/5068939 How to concatenate factors, without them being converted to integer level?].
<pre>
unlist(list(f1, f2))
# unlist(list(factor(letters[1:5]), factor(letters[5:2])))
</pre>
 
=== droplevels() ===
[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.
 
=== factor(x , levels = ...) vs levels(x) <-  ===
<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.
 
{| class="wikitable"
|-
| [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
|}
 
<syntaxhighlight lang='rsplus'>
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
</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 ***
 
# 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 ***
</syntaxhighlight>
 
=== stats::relevel() ===
[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.
 
=== reorder(), levels() and boxplot() ===
<ul>
<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)
<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.
<pre>
# 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
</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
 
ordered(levels=c("a", "b", "c"))
# Error in factor(x, ..., ordered = TRUE) :
#  argument "x" is missing, with no default
</pre>
 
== 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/
 
=== 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, [https://developer.r-project.org/Blog/public/2020/02/16/stringsasfactors/ 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.
<pre>
> 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
</pre>
 
=== Create unique rownames: make.unique() ===
<pre>
groupCodes <- c(rep("Cont",5), rep("Tre1",5), rep("Tre2",5))
rownames(mydf) <- make.unique(groupCodes)
</pre>
 
=== data.frame() will change rownames ===
<pre>
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"
</pre>
 
=== Print a data frame without rownames ===
<pre>
# Method 1.
rownames(df1) <- NULL
 
# Method 2.
print(df1, row.names = FALSE)
</pre>
 
=== 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)
 
# Method 2:
bob[] <- lapply(bob, as.character)
</pre>
 
[https://stackoverflow.com/a/2853231 To replace only factor columns]:
<pre>
# 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
</pre>
 
=== Sort Or Order A Data Frame ===
[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)
 
=== data.frame to vector ===
<pre>
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
</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.
 
=== 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 [http://adv-r.had.co.nz/Data-structures.html#data-frames Advanced R -> Data structures] chapter.
 
=== 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.
 
=== merge ===
* [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].
* [https://www.geeksforgeeks.org/merge-dataframes-by-row-names-in-r/ Merge DataFrames by Row Names in R]
* [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]
* [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]
 
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
<pre>
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>
 
=== 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>
X <- data.frame(x=1:2, y=3:4)
</pre>
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.
* [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.
 
<syntaxhighlight lang='r'>
# 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
</syntaxhighlight>
 
* 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.
<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>
 
=== matrix vs data.frame ===
Case 1: colnames() is safer than names() if the object could be a data frame or a matrix.
<pre>
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
</pre>
 
Case 2:
{{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
 
ip2 <- as.data.frame(installed.packages()[,c(1,3:4)], stringsAsFactors = FALSE) # matrix -> data.frame
unique(ip2$Priority)    # OK
</pre>
 
The length of a matrix and a data frame is different.
{{Pre}}
> 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
> 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.
 
=== How to Remove Duplicates ===
[https://www.r-bloggers.com/2021/08/how-to-remove-duplicates-in-r-with-example/ How to Remove Duplicates in R with Example]
 
=== Convert a matrix (not data frame) of characters to numeric ===
[https://stackoverflow.com/a/20791975 Just change the mode of the object]
{{Pre}}
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
</pre>
 
=== Convert Data Frame Row to Vector ===
as.numeric() or '''c()'''
 
=== Convert characters to integers ===
mode(x) <- "integer"
 
=== Non-Standard Evaluation ===
[https://thomasadventure.blog/posts/understanding-nse-part1/ 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 [https://www.datanovia.com/en/lessons/select-data-frame-columns-in-r/ 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.
<pre>
class(USArrests)  # "data.frame"
USArrests$"Assault"
</pre>
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
 
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>
data.table(USArrests)[1:2, list(Murder, Assault)]
</pre>
 
=== 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]
 
=== 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) )
 
# 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))]
</pre>
 
=== 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]
 
=== Danger of selecting rows from a data frame ===
<pre>
> 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
</pre>
 
=== 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]
 
=== Create an empty data.frame ===
https://stackoverflow.com/questions/10689055/create-an-empty-data-frame
<pre>
# 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)
</pre>
 
=== Objects from subsetting a row in a data frame vs matrix ===
* [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>
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>
 
'trees' data from the 'datasets' package
<pre>
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
</pre>
</li>
</ul>
 
=== 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>
# 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>
 
=== tibble and data.table ===
* [[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]
 
== matrix ==
 
=== Define and subset a matrix ===
* [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]'''.
 
<pre>
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
</pre>
 
=== 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 ===
[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.
 
=== 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.
 
{{Pre}}
> 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
</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).
 
== format.pval/print p-values/format p values ==
[https://rdrr.io/r/base/format.pval.html format.pval()]. By default it will show 5 significant digits (getOption("digits")-2).
{{Pre}}
> format.pval(c(stats::runif(5), pi^-100, NA))
[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"
 
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
</pre>
 
== Customize R: options() ==
 
=== Change the default R repository, my .Rprofile ===
[[Rstudio#Change_repository|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")'''
{{Pre}}
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"))
}
</pre>
 
=== 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
<pre>
options(browser='seamonkey')
</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.
 
For one-time only purpose, we can use the ''browser'' option in help.start() function:
{{Pre}}
> 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
 
=== Change the default editor ===
On my Linux and mac, the default editor is "vi". To change it to "nano",
{{Pre}}
options(editor = "nano")
</pre>
 
=== Change prompt and remove '+' sign ===
See https://stackoverflow.com/a/1448823.
{{Pre}}
options(prompt="R> ", continue=" ")
</pre>
 
=== digits ===
* [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]
* [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>
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>
 
In Python,
{{Pre}}
>>> 100000.07 + .04
100000.11
</pre>
 
=== [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?]
 
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
 
# 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
</pre>
 
=== Suppress warnings: options() and capture.output() ===
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.
{{Pre}}
op <- options("warn")
options(warn = -1)
....
options(op)
 
# OR
warnLevel <- options()$warn
options(warn = -1)
...
options(warn = warnLevel)
</pre>
 
[https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/warning suppressWarnings()]
<pre>
suppressWarnings( foo() )
 
foo <- capture.output(
bar <- suppressWarnings(
{print( "hello, world" );
  warning("unwanted" )} ) )
</pre>
 
[https://www.rdocumentation.org/packages/utils/versions/3.6.2/topics/capture.output capture.output()]
<pre>
str(iris, max.level=1) %>% capture.output(file = "/tmp/iris.txt")
</pre>
 
=== Converts warnings into errors ===
options(warn=2)
 
=== demo() function ===
<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>
 
== sprintf ==
=== 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]
 
=== sep vs collapse in paste() ===
* sep is used if we supply multiple input objects to paste()
* 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=",")
[1] "a,A"
R> paste("a", "A", sep=",", collapse="-")
[1] "a,A"
R> paste(c("a", "A"), collapse="-")
[1] "a-A"
 
R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
[1] "a,A - b,B - c,C"
R> paste(letters[1:3], collapse = "-")
[1] "a-b-c"
</pre>
 
=== 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
 
{{Pre}}
# 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"
</pre>
 
=== formatC and prettyNum (prettifying numbers) ===
* [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"
 
R> x <- .000012345
R> prettyNum(x)
[1] "1.2345e-05"
R> x <- .00012345
R> prettyNum(x)
[1] "0.00012345"
</pre>
 
=== Format(x, scientific = TRUE) ===
Print numeric data in exponential format, so .0001 prints as 1e-4
 
== Creating publication quality graphs in R ==
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
 
== 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.
 
* 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.
 
== 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 ==
* [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()]
* 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].
<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 (alternative to histogram), bar chart on terminals ==
* https://en.wikipedia.org/wiki/Stem-and-leaf_display
* [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
 
== Plot histograms as lines ==
https://stackoverflow.com/a/16681279. This is useful when we want to compare the distribution from different statistics.
<pre>
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)
</pre>
 
== Histogram with density line ==
<pre>
hist(x, prob = TRUE)
lines(density(x), col = 4, lwd = 2)
</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).
 
== 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 [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.
 
# 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(). [https://www.rdocumentation.org/packages/graphics/versions/3.6.2/topics/par ?par].
# 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()]
# 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 Fit A Smooth Curve To Your R Data? loess() and lines()
# How To Add Error Bars In An R Plot? arrows()
# How To Save A Plot As An Image On Disc
# 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 Fix The Aspect Ratio For Your R Plots? asp parameter
# What Is The Function Of hjust And vjust In ggplot2?
 
== jitter function ==
* https://www.rdocumentation.org/packages/base/versions/3.5.2/topics/jitter
** 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://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]
 
:[[File:Jitterbox.png|200px]]
 
== Scatterplot with the "rug" function ==
<pre>
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")
})
</pre>
[[: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.
 
== Identify/Locate Points in a Scatter Plot ==
<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 ==
* http://www.gettinggeneticsdone.com/2015/04/r-single-plot-with-two-different-y-axes.html
 
== Draw Color Palette ==
* http://teachpress.environmentalinformatics-marburg.de/2013/07/creating-publication-quality-graphs-in-r-7/
 
=== Default palette before R 4.0 ===
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
 
=== svglite ===
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 ===
Using Inkscape. See [https://robertgrantstats.wordpress.com/2017/09/07/svg-from-stats-software-the-good-the-bad-and-the-ugly/ this post].
 
=== svg -> png ===
[https://laustep.github.io/stlahblog/posts/SVG2PNG.html SVG to PNG] using the [https://cran.rstudio.com/web/packages/gyro/index.html gyro] package
 
== read.table ==
=== clipboard ===
{{Pre}}
source("clipboard")
read.table("clipboard")
</pre>
 
=== inline text ===
{{Pre}}
mydf <- read.table(header=T, text='
cond yval
    A 2
    B 2.5
    C 1.6
')
</pre>
 
=== http(s) connection ===
{{Pre}}
temp = getURL("https://gist.github.com/arraytools/6743826/raw/23c8b0bc4b8f0d1bfe1c2fad985ca2e091aeb916/ip.txt",
                          ssl.verifypeer = FALSE)
ip <- read.table(textConnection(temp), as.is=TRUE)
</pre>
 
=== 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.
{{Pre}}
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))
</pre>
 
To know the number of columns, we might want to read the first row first.
{{Pre}}
library(magrittr)
scan("var_annot.vcf", sep="\t", what="character", skip=62, nlines=1, quiet=TRUE) %>% length()
</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]
 
=== 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 1,672: Line 5,685:
> writeBin(as.integer(a_length), connection, endian="big")
> writeBin(as.integer(a_length), connection, endian="big")
> serialize(a, connection)
> serialize(a, connection)
 
</pre>
In C++ process, I receive one int variable first to get the length, and
In C++ process, I receive one int variable first to get the length, and
then read <length> bytes from the connection.
then read <length> bytes from the connection.
== socketConnection ==
See ?socketconnection.
=== Simple example ===
from the socketConnection's manual.
Open one R session
<pre>
con1 <- socketConnection(port = 22131, server = TRUE) # wait until a connection from some client
writeLines(LETTERS, con1)
close(con1)
</pre>
Open another R session (client)
<pre>
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)
</pre>
=== 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
<pre>
nc localhost 22131  [ENTER]
</pre>
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
<pre>
nc -v -w 2 localhost -z 22130-22135
</pre>
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,
<pre>
con1 <- socketConnection(port = 8080, server = TRUE)
</pre>
On the client,
<pre>
curl --trace-ascii debugdump.txt http://localhost:8080/
</pre>
Then go to the server,
<pre>
while(nchar(x <- readLines(con1, 1)) > 0) cat(x, "\n")
close(con1) # return cursor in the client machine
</pre>
=== Use telnet command in client ===
On the server,
<pre>
con1 <- socketConnection(port = 8080, server = TRUE)
</pre>
On the client,
<pre>
sudo apt-get install telnet
telnet localhost 8080
abcdefg
hijklmn
qestst
</pre>
Go to the server,
<pre>
readLines(con1, 1)
readLines(con1, 1)
readLines(con1, 1)
close(con1) # return cursor in the client machine
</pre>
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 ==
[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].
The result of the command '''x[3:5] <- 13:15''' is as if the following had been executed
<pre>
`*tmp*` <- x
x <- "[<-"(`*tmp*`, 3:5, value=13:15)
rm(`*tmp*`)
</pre>
=== Avoid Coercing Indices To Doubles ===
[https://www.jottr.org/2018/04/02/coercion-of-indices/ 1 or 1L]
=== Careful on NA value ===
See the example below. base::subset() or dplyr::filter() can remove NA subsets.
<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
xnam <- paste("x", 1:25, sep="")
fmla <- as.formula(paste("y ~ ", paste(xnam, collapse= "+")))
</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]
{{Pre}}
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
</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 and signature ==
* 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
* 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://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
* [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
* https://www.coursera.org/lecture/bioconductor/r-s4-methods-C4dNr
* https://www.bioconductor.org/help/course-materials/2013/UnderstandingRBioc2013/
* http://adv-r.had.co.nz/S4.html, http://adv-r.had.co.nz/OO-essentials.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]
=== Debug an S4 function ===
* '''showMethods('FUNCTION')'''
* '''getMethod('FUNCTION', 'SIGNATURE') ''' 
* '''debug(, signature)'''
{{Pre}}
> 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
</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>
</pre>
* '''getClassDef()''' in S4 ([http://www.bioconductor.org/help/course-materials/2014/Epigenomics/BiocForSequenceAnalysis.html Bioconductor course]).
{{Pre}}
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"
</pre>
=== 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.
* [https://kasperdanielhansen.github.io/genbioconductor/html/R_S4.html#slots-and-accessor-functions R - S4 Classes and Methods] Hansen. '''getClass()''' or '''getClassDef()'''.
=== setReplaceMethod() ===
* [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?]
=== 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::"+"
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")
</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.
* 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 [http://www-huber.embl.de/DESeq2paper/vignettes/posterior.pdf Shrinkage of logarithmic fold changes] vignette of the DESeq2paper package
{{Pre}}
> 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
</pre>
== Pipe ==
* [https://www.tidyverse.org/blog/2023/04/base-vs-magrittr-pipe/ Differences between the base R and magrittr pipes] 4/21/2023
* [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]
<ul>
<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>
</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() ==
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://adv-r.had.co.nz/Rcpp.html Hadley Wickham]
== Assign operator ==
* Earlier versions of R used underscore (_) as an assignment operator.
* [https://developer.r-project.org/equalAssign.html Assignments with the = Operator]
* In R 1.8.0 (2003), the assign operator has been removed. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
* In R 1.9.0 (2004), "_" is allowed in valid names. See [https://cran.r-project.org/src/base/NEWS.1 NEWS].
: [[File:R162.png|200px]]
== 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]'''.
<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
> 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
</pre>
=== OS-dependent results on sorting string vector ===
Gene symbol case.
<pre>
# mac:
order(c("DC-UbP", "DC2")) # c(1,2)
# linux:
order(c("DC-UbP", "DC2")) # c(2,1)
</pre>
Affymetric id case.
<pre>
# 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"
</pre>
It does not matter if we include factor() on the character vector.
The difference is related to locale. See
* [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>
# 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>
=== unique() ===
It seems it does not sort. [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/unique ?unique].
<pre>
# mac & linux
R> unique(c("DC-UbP", "DC2"))
[1] "DC-UbP" "DC2"
</pre>
== 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>
> 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
</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
$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-"
</pre>
* ''environment'' and ''quote'' arguments.
{{Pre}}
> 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
</pre>
* Good use case; see [https://stackoverflow.com/a/11892680 Get all Parameters as List]
{{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>
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
</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]
=== do.call vs mapply ===
* 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.
{{Pre}}
> 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"  "-"
</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>
=== do.call vs lapply ===
[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())'''
* 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.
{{Pre}}
> 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"
</pre>
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 [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.
{{Pre}}
> 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
</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"
> 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"
</pre>
=== do.call + rbind + lapply ===
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.
{{Pre}}
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"
</pre>
A 2nd example is to [http://datascienceplus.com/working-with-data-frame-in-r/ sort a data frame] by using do.call(order, list()).
Another example is to reproduce aggregate(). aggregate() = do.call() + by().
{{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>
== Run examples ==
When we call help(FUN), it shows the document in the browser. The browser will show
<pre>
example(FUN, package = "XXX") was run in the console
To view output in the browser, the knitr package must be installed
</pre>
== How to get examples from help file, example() ==
[https://blog.r-hub.io/2020/01/27/examples/ Code examples in the R package manuals]:
<pre>
# How to run all examples from a man page
example(within)
# How to check your examples?
devtools::run_examples()
testthat::test_examples()
</pre>
See [https://stat.ethz.ch/pipermail/r-help/2014-April/369342.html this post].
Method 1:
<pre>
example(acf, give.lines=TRUE)
</pre>
Method 2:
<pre>
Rd <- utils:::.getHelpFile(?acf)
tools::Rd2ex(Rd)
</pre>
== "[" and "[[" with the sapply() function ==
Suppose we want to extract string from the id like "ABC-123-XYZ" before the first hyphen.
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), "[", 1)
</pre>
is the same as
<pre>
sapply(strsplit("ABC-123-XYZ", "-"), function(x) x[1])
</pre>
== Dealing with dates ==
* Find difference
:<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")
# 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)
</syntaxhighlight>
* http://cran.r-project.org/web/packages/lubridate/vignettes/lubridate.html
:<syntaxhighlight lang='rsplus'>
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
</syntaxhighlight>
* 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.
* [https://cran.r-project.org/web/packages/anytime/index.html anytime] package
* weeks to Christmas difftime(as.Date(“2019-12-25”), Sys.Date(), units =“weeks”)
* [https://blog.rsquaredacademy.com/handling-date-and-time-in-r/ A Comprehensive Introduction to Handling Date & Time in R] 2020
* [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'''.
* [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
== Nonstandard/non-standard evaluation, deparse/substitute and scoping ==
* [https://www.brodieg.com/2020/05/05/on-nse/ Standard and Non-Standard Evaluation in R]
* [http://adv-r.had.co.nz/Computing-on-the-language.html Nonstandard evaluation] from Advanced R book.
* [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
** Labelling: turn an argument into a label
** Formulas
** Dot-dot-dot
* [https://www.rdocumentation.org/packages/base/versions/3.5.1/topics/substitute substitute(expr, env)] - capture expression. The return mode is a '''call'''.
** 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>
* 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))
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) # 👍
</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"   
> 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"
</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(…))?]
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).
{{Pre}}
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
</pre>
=== library() accept both quoted and unquoted strings ===
[https://stackoverflow.com/a/25210607 How can library() accept both quoted and unquoted strings]. The key lines are
<pre>
  if (!character.only)
    package <- as.character(substitute(package))
</pre>
=== Lexical scoping ===
* [https://lgreski.github.io/dsdepot/2020/06/28/rObjectsSObjectsAndScoping.html R Objects, S Objects, and Lexical Scoping]
* [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]
== The ‘…’ argument ==
* 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://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://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?]
== Functions ==
* https://adv-r.hadley.nz/functions.html
* [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!
=== Function argument ===
[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.
Argument matching is augmented by the functions
* [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/match.arg match.arg],
* [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].
Access to the partial matching algorithm used by R is via [https://www.rdocumentation.org/packages/base/versions/3.6.1/topics/pmatch pmatch].
=== Check function arguments ===
[https://blog.r-hub.io/2022/03/10/input-checking/ Checking the inputs of your R functions]: '''match.arg()''' , '''stopifnot()'''
'''stopifnot()''': function argument sanity check
<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>
stopifnot(condition1, condition2, ...)
</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>
=== 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
'''R function arguments are lazy — they’re only evaluated if they’re actually used'''.
* Example 1. By default, R function arguments are lazy.
<pre>
f <- function(x) {
  999
}
f(stop("This is an error!"))
#> [1] 999
</pre>
* Example 2. If you want to ensure that an argument is evaluated you can use '''force()'''.
<pre>
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.
<pre>
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"
</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]
=== 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.
<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'>
> (function(x) x * x)(3)
[1] 9
> (\(x) x * x)(3)
[1] 9
</syntaxhighlight>
</ul>
== 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 [http://adv-r.had.co.nz/Functions.html Advanced R] and [https://stackoverflow.com/a/36229703 What do backticks do in R?].
<pre>
iris %>%  `[[`("Species")
</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
})
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)
</pre>
</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>
=== Using $ in R on a List ===
[https://r-lang.com/dollar-sign-in-r-with-example/ How to Use Dollar sign 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
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
> .libPaths(c("~/R/x86_64-pc-linux-gnu-library/3.4", .libPaths()))
> library(devtools) # Works
</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].
== 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 ==
* [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.'' '''
* ‘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
== 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 ==
[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>
== 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.
=== 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()'''.
== License ==
[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://moderndata.plot.ly/why-dash-uses-the-mit-license/ 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.

Revision as of 13:56, 22 April 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)

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.

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

creating directed networks with 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

RDCOMClient where excel.link depends on it.

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

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.

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 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

Tibbles are data frames, but slightly tweaked to work better in the tidyverse.

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

Tibbles Vignette

> data(pew, package = "efficient")
> dim(pew) 
[1] 18 10
> class(pew) # tibble is also a data frame!!
[1] "tbl_df"     "tbl"        "data.frame"

> tidyr::gather(pew, key=Income, value = Count, -religion) # make wide tables long
# A tibble: 162 x 3
                                                       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

To show all rows of a tibble object, use the print() method.

print(tbObj, n= Inf)

tbObj %>% print(n= nrow(.))

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

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)

To print all rows of a tibble object, use print(tbl_df, n=Inf) or tbl_df %>% print(n=Inf)

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().
    library(glue)
    name <- "Fred"
    glue('My name is {name}.')  # My name is Fred.
    
  • 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

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

summary_factorlist() from the finalfit package.

table1

gtsummary

gt

Introduction to Clinical Tables with the {gt} Package

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).

> format.pval(c(stats::runif(5), pi^-100, NA))
[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"

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

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 input objects to paste()
  • 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=",")
[1] "a,A"
R> paste("a", "A", sep=",", collapse="-")
[1] "a,A"
R> paste(c("a", "A"), collapse="-")
[1] "a-A"

R> paste(letters[1:3], LETTERS[1:3], sep=",", collapse=" - ")
[1] "a,A - b,B - c,C"
R> paste(letters[1:3], collapse = "-")
[1] "a-b-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)

Print numeric data in exponential format, so .0001 prints as 1e-4

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

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

  • 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)
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

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))

Using $ in R on a List

How to Use Dollar sign 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

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())

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

“Do More with R” video tutorials. Search for R video tutorials by task, topic, or package. Most videos are shorter than 10 minutes.

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.