tl;dr: Dependencies hurts when you try to set up R on server After a while of playing around, I’ll say the best way to use R with a Notebook-style interface on a server where you are no superuser would be using Anaconda, and then run R inside Anaconda to get whatever package you need. It is designed to run for a normal user, so there’s no need for superuser permission, and dependency issue is also taken care of most of the time. If you get the permission on server and are comfortable with R Studio, R Studio Server might be a even better option. That is because even if the packages are in Conda R , Conda Forge , or Bioconda , sometimes they are out of date, which might break dependencies, and conda skeleton and conda build unfortunately come with a lot of dependency issue as well. Besides, the visualization of environment and integration of documentation is better in R Studio for me. What worked for me starts from creating a new environment in Anaconda and install t...