Running R on Linux systems

Running R on Linux systems#

Linux allows us to run specific versions of R, but may also pose particular challenges when it comes to installing packages.

BioHPC specific

See R notes for BioHPC.

Specific versions of R#

You may either need to use module or docker to run specific versions of R. Both can work on BioHPC.

BioHPC specific

Use module avail to see which versions of R are available. Use module load R/4.2.1-r9 to load a specific version.

Installing packages#

R Packages Through CRAN#

R on Linux will usually install packages from CRAN, and compile from source. This may be feasible to circumvent by using Package Archives, and defining the specific operating system you are using.

BioHPC specific

BioHPC uses Rocky Linux 9.

This is the default for rocker/ provided Docker images.

R Packages through Conda#

On BioHPC, installing packages through R can fail, with errors such as


----------------------------- ANTICONF -------------------------------
Configuration failed to find libgit2 library. Try installing:
 * brew: libgit2 (MacOS)
 * deb: libgit2-dev (Debian, Ubuntu, etc)
 * rpm: libgit2-devel (Fedora, CentOS, RHEL)
If libgit2 is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a libgit2.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 MESSAGE] ---------------------------
<stdin>:1:10: fatal error: git2.h: No such file or directory
compilation terminated.
----------------------------------------------------------------------

This occurs since libraries on BioHPC are often installed in directories which R does not expect. To solve this, we use conda. Conda handles dependencies of both R packages and other libraries. See installation details [here][https://biohpc.cornell.edu/lab/userguide.aspx?a=software&i=574#c]. To create a conda environment, run the command

conda env create --name <name> r-base=<version>

  • This creates a conda environment with the proper R version installed. After activating the environment, install R packages with the command

conda install conda-forge::r-tidyverse

R packages on conda often follow the format r-<package>. You can search for packages [here][https://anaconda.org/].

On Anaconda, R packages are often hosted by the R repository, and conda-forge. Packages from the R repository tend to be pickier about version compatibility than those from conda-forge.

Errors#

Package not available#

If you get

3: package ‘NameOfPackage’ is not available for this version of R

check

install_version("mypackage", "1.15")

  • your version of R is too young (see above for specific versions)

  • your version of R is too old (see above for specific versions)

MRAN no longer available#

Some older packages may attempt to use MRAN - the “Microsoft R Application Network”. This was shut down on July 1, 2023 (see this notice).

Solution: Switch to Posit Package Manager (PPM).

Contents