Guide to installing package on UQ’s bunya system

1. If you don’t already have a folder for storing personal R packages set up, do this first:
   1. Make a directory for your R packages.
      • For example: mkdir ~/Rlibs. Replace ~/Rlibs with your choice of folder name.
   2. Tell R where to find the R packages that will be saved here:
      • echo '.libPaths( c("~/Rlibs", .libPaths()) )' >> ~/.Rprofile, with ~/Rlibs modified to be the folder from step 1.
2. To install this package:
   1. Download the genomicSimulation source package (.tar.gz) from the Releases page on Github (https://github.com/vllrs/genomicSimulation/releases), and save that .tar.gz file somewhere in your directories on bunya.
      • For file transfer between your machine and bunya, the RCC recommends using scp or WinSCP. See their guides.
      • In this tutorial, the downloaded package is assumed to be located at ~/genomicSimulation_0.1.tar.gz on bunya
   2. Get an interactive job on an “epyc3” node: salloc --nodes=1 --ntasks-per-node=1 --cpus-per-task=1 --mem=5G --job-name=TinyInteractive --time=01:00:00 --partition=general --qos=debug --constraint=epyc3 --account=ACCOUNT srun , where ACCOUNT is replaced with your bunya group account string.
      • It’s important to compile software on bunya on an epyc3 node. Otherwise, if you are later running a job, and bunya allocates the job to an epyc3 node, the job may crash with an “invalid operand” or “illegal instruction” error when it tries to use a package that was compiled on a non-epyc3 node.
   3. On the interactive job node, open the R version of your choice.
      • For example, run module load r, then open the R commandline by typing R
   4. In R, on the interactive job node, install the package from the release file: install.packages("~/genomicSimulation_0.1.tar.gz", repos=NULL) with the location you saved the package release substituted for ~/genomicSimulation_0.1.tar.gz
      • Once this process has succeeded, the ~/genomicSimulation_0.1.tar.gz file can be deleted. The .Rprofile file and the contents of the ~/Rlibs folder should not be deleted. You can run q() to exit R, and exit to release your interactive job and return to the login node.

Once the package is installed, you can use library("genomicSimulation") in any R script. There should be no need to rerun the install.packages command, even after module unload and module load.