R and RStudio
R is a statistical programming language and Rstudio is a development environment for R. R/Rstudio is good software for running advanced statistical analysis, e.g. statistical modelling with the lme4 or brms packages and nice plotting features with ggplot2.
Neither R nor RStudio is pre-installed on Compute. You have to install it yourself. At the moment it is not possible to install directly from the source. However, with minimal hacks, it can be installed through Anaconda.
However, it appears that R/Rstudio only has little support from the developers, so this does not appear to give the latest version of R.
For an up to date list of known issues with R/Rstudio on Compute see here.
Install RStudio through Anaconda
The first step is to make sure you have Anaconda installed: https://docs.anaconda.com/anaconda/install/linux/#installation
Make a new environment for RStudio. This will create an environment called r_env:
mamba create --name r_env
source activate r_env
Install RStudio from Conda Forge (https://anaconda.org/conda-forge/rstudio-desktop). This will install Rstudio and all dependencies (like R itself):
mamba install -c conda-forge rstudio-desktop
rstudio in the terminal and you are ready to do some fun statistics :)
I get several error messages in the terminal when I start Rstudio. Those I ignore. Rstudio starts and works just fine.
Update R
When starting RStudio it informs you that it is an outdated version and you should get the latest version from the website. Just click ignore. To update R your need to do it through Anaconda:
mamba update r-base
Install packages
Most packages can be installed as usual though R with install.packages. Some packages (like lme4) cannot be installed through RStudio as a default. To fix this, I needed to first install CMake. Cmake can be installed through Anaconda, like this:
mamba install -c anaconda cmake
After this, it was possible to install lme4 through R as usual:
install.packages("lme4")