These instructions assume that Anaconda is already installed (or can be loaded using module load, e.g. module load Anaconda3/5.1.0 on Phoenix HPC).
Create a new conda environment for the R install:
conda create --name R_env
Activate the environment using source activate R_env and install the following in this order:
conda install r-essentials r-base
conda install -c conda-forge r-devtools
conda install -c conda-forge r-biocmanager
Going in this order ensures that the version of R is 3.6+, which is required for some Bioconductor packages.
Type R to start R.
Because the BiocManager may be out of date, use the command BiocManager::valid() and copy and paste the command it outputs in order to get the packages to be valid. This may take a while.
Desired packages can now be installed using install.packages() within R or using conda, which is the safer option. Note, if packages installed using install_github gives the following error:
> install_github("satijalab/Seurat", ref="develop")
Using GitHub PAT from envvar GITHUB_PAT
Downloading GitHub repo StatsWithR/statsr@master
from URL
Installation failed: Bad credentials (401)
This can be resolved by unsetting the envvar GITHUB_PAT by using this command: Sys.unsetenv("GITHUB_PAT"), see ref.