Installing R packages on NeSI

These scripts are designed to facilitate installation of R software packages on the NeSI Pan Cluster. These take as many packages as inputs as given, allowing the installation of a single package or submission of multiple packages simultaneously.

Installing R packages in the terminal

This installs the packages specified into R from the terminal:

Rscript  install.R <package1> <package2> <package>

Installing R packages via shell scripts

This installs the packages specified into R from the terminal:

install.sh <package1> <package2> <package>

Installing R packages via slurm on the Cluster

This installs CRAN packages specified into R as a slurm job:

sbatch install.sl <package1> <package2> <package>

Bioconductor R Packages via slurm

This installs Bioconductor packages specified into R as a slurm job:

sbatch install_bioconductor.sl <user1/repo1> <user2/repo2> <user3/repo3>

GitHub R packages via slurm

This installs the packages specified into R as a slurm job:

sbatch install_github.sl <package1> <package2> <package>

Note that for installing from a private github repo you will need to add you personal access token (PAT) as the auth_token argument in install_github.sl.

##Configuring slurm

Note here that you will need to configure slum with the appropriate settings for your account:

#!/bin/bash
#SBATCH -A <projectID>
#SBATCH -J install_pkg
#SBATCH --mail-type=ALL
#SBATCH --mail-user=<user>@<domain>
#SBATCH --ntasks 1
#SBATCH --cpus-per-task=1
#SBATCH --time=10:00 #expected time to install
#SBATCH --mem-per-cpu=25M #expected memory to install (includes test of loading)
module load R/3.3.0-intel-2015a
srun Rscript install.R "$@"

The default settings here are to facilitate rapid queued installations for installing small packages. Note however that installing a package includes loading the data included and may require more memory, particularly for bioconductor packages.

##Installed packages

Packages will be installed in a personal library under your nesi user account if they are not included in the R module you have requested in the slurm file. Note that if you installing large packages (e.g., annotation packages on Bionconductor) this will be limited by your user quote, not your project, and you may need to request an increase in user disk quota from support@nesi.org.nz.

You can get a complete list of the personal packages you have installed (for a particular version of R) under ~/R. For instance:

ls /home/<user.name>/R/x86_64-pc-linux-gnu-library/3.3

##modules

Note that some packages are already supported by "R modules" on NeSI and will be available with library() without a user install. Most notably these include snow and Rmpi to run parallel code on NeSI. This also includes a number of base, stats, and Bioinformatics packages. These will not be installed in a personal library as they are already supported by NeSI.

The following command will list the available R modules on NeSI:

 module avail | grep "R/"

More information on the modules supported by the NeSI Pan cluster can be found below:

• https://github.com/nesi/applications/tree/master/R/modules/Auckland/pan
• https://support.nesi.org.nz/hc/en-gb/articles/209338087-R#available-modules

Updating R packages in the terminal

This updates the CRAN packages specified into R from the terminal:

sbatch update.sl <package1> <package2> <package>

Alternatively, this script defaults to updating all packages in the personal library:

sbatch update.sl

###Source

Based on functions provided by NeSI (Markus Binsteiner):

• https://github.com/nesi/r-dependencies/blob/develop/dependencies/R/dependencies.R