Looked at:

• Where R (loaded from a module) installed packages with respect to the version of R.

• Where an R environment built with Conda installs packages.

• Inspecting the paths returned from .libPaths() to understand and modify where packages are installed.

• Using R’s install.packages() command and that additional modules/libraries might need to be loaded into your environment for packages to successfully be installed.

• Using Conda’s install to install packages.

  • Similarly, you might need to conda install additional libraries.

• How R libraries can be shared across a user’s environments and potentially how updating a package version can impact these environments that share the library and packages.

• How to create an R library that can be shared by users across a project.

• How to use RStudio via OnDemand and an Interactive Desktop.

• How to configure a session to use an R environment within a Conda environment.

• How to take an existing R Conda Environment and update it to use as a kernel within Juypter via OnDemand.

• The more complicated an environment, the more packages you’re trying to install, the more likely you’re of hitting dependency hell.

  • Consider having a number of smaller environments.

• The notion of reproducibility of environments.

• Some high-level aspects to consider when parallel programming within R on the cluster.