R Packages with Packrat

Use packrat R package to create a personal environment of R packages.

Setup

Say your R code is in directory /scratch/$USER/projects/project1

Packrat package is already installed for module r/intel/4.0.3. You need to install it yourself if you use other R module version

You will see:

Add/remove, etc. packages

Error in library("reshape2") : there is no package called ‘reshape2’

Test

• • Create a script which calls reshape2 library

  • Run it from project1 directory

  • Now create new directory project2

  • Run the same file - shall fail

  • Option 1

    • copy packrat/packrat.lock file from project1 to project2 directory

    • Run the same file - all the packages shall be installed on the fly 

  • Option 2

    • copy whole packrat directory from project1 to project 2

    • Run the same file - will just run

  • Option 3

    • use bundle (see below)

Test R file

For testing run it as

Note: your '.Rprofile' file will include line source("packrat/init.R")

If you copy the whole directory of your project and don't want to use packrat remove it

Clean up

Keep only the packages that you use in this particular project (not all the packages available on the system)

Use with sbatch

Before you launch sbatch job, you need to make sure packrat environment is ready

You can restore as described below using packrat::restore() or packrat::unbundle()

When you launch a job with sbatch,  R will check if there is packrat directory, and if packrat is on it will pick up packages, installed using packrat in the current directory

Store and share your project in a fully reproducible way

Reproduce environment

If you already have file packrat.lock or bundle file skip step 1

1. In the original location (your own laptop for example) go to project directory and execute

(Make sure the whole path to project directory and names of your script files don't have empty spaces!)

2. Take file packrat/packrat.lock  OR bundle file created above, and copy it to a new location for the project

3. At the new location - restore environment: go to directory of the project and execute. (Make sure version of R is the same)

What is bundle

You can create a bundle file which will include all your project files, as well as a list of libraries with versions, plus the source files for libraries (no need to download anything later)

Packrat will install/compile what is needed on any system (Linux, Windows, etc)

You can share your code with other researchers no matter what system they use

Say you have a proejct1

• • go to the directory of project 1

  • start R

  • packrat::bundle()

  • You will see something like

  • The packrat project has been bundled at:

  • - "/scratch/$USER/projects/project1/packrat/bundles/project1-2019-11-08.tar.gz"

  • You can copy this file anywhere, then start R and unbundle it

  • packrat::unbundle("/scratch/$USER/projects/project1/packrat/bundles/project1-2019-11-08.tar.gz", where = "/scratch/$USER/projects/project3")

What to save/publish/Git commits

In order to have your work reproducible by you or/and others, save and/or commit your code in git, please including

• • packrat/packrat.lock (which lists all packages and versions that you use)

  • (optional) packrat/src/ (which will include all the source files for packages). This may be important if you use some custom packages, for which source code may be not available any more in the future (github repository deleted, etc.). Similar to bundle option described below

  • save version of R you are using

Useful links

• https://rstudio.github.io/packrat/walkthrough.html

• https://rstudio.github.io/packrat/commands.html