Getting started with Base R
Contributed by Dr. Howard Castelyn.
Follow these steps to start working with base R:
-
Login into the HPC
-
In the terminal (if you are in a vnc desktop session, open a terminal window) type the following:
module load R/<version>
IMPORTANT NOTE: Tab complete to view the versions availible and make sure to load a specific version of R. This is important for using / compiling / installing packages (See below)
Using / installing / compiling R packages
In the next steps you will learn how to install and load R packages. However, it is important to understand that R packages you install (and likely build/compile) will likely only work with the version of R that you installed them with. It is thus imperative that you ensure to module load the exact version of R everytime (and thus the warning in step 2). Should you wish to use a different version of R, you will need to create a new library directory, ensure that you point R to that directory and re-install all the packages you need into that library.
Getting errors when loading packages related to library files (the linux kind) likely indicates the problems mentioned above.
-
When using the particular R version for the first time, when need to make new library. First make a directory for the library:
$ mkdir -p $R_LIBS_USER
-
Now we need to check if R can see your created library path. Launch R and type the following:
.libPaths()
The output for me is as follows:
[1] "/home/castelynhd/R/library/gen5/4.0.4" [2] "/data/soft/sl7/gen5/R/4.0.4_gcc_9.3.0/lib64/R/library"
The first item lists the path I just created. So it is fine. The second directory is the system default library.
-
In the event that your user path does not appear in step 4, execute the following in R:
.libPaths( c("~/R/your/library/path/here/, .libPaths() ) )
Now repeat step 4 to check if the path is recognized by R.
-
Finally, you need to set the CRAN repository to the local mirror, especially if you plan on using R scripts with Rscript. To do this, copy this file to your home directory and rename the file to .Rprofile
-
You can now install and load packages as you are accustomed to.
IMPORTANT NOTE: If you attempt to load and receive error messages about .so files that are missing, then you are probably trying to load a package that is was not built with the currently loaded version of R. If you are unsure, please send an email to hpc@ufs.ac.za