Running R Singularity images in VSCode

Given the friction of maintaining R instances and libraries in the server, we have begun testing the use of all-in-one singularity images for standard R workflows. This has the advantage of using ipynb files for everything and not needing to deal with the jankiness of R in VSCode. In addition, it guarantees the versions of the libraries installed play nice.

There are however a few downsides: - The images are immutable. You have access to the libraries within the image and cannot add libraries from your personal R library - The Brickman lab assays/ and projects/ shortcuts do not work. All access to these folders must be done from the longhand path /maps/...

Instructions for using R Singularity images

The instructions assume you have set up your VSCode and are familiar with the current server structure.

1. Login to head node: ssh $USER@danhead01fl.unicph.domain
2. Start a new tmux session: tmux new -s rsingularity or attach to a running tmux session tmux -a
3. Start a new job: srun -w {SERVER} -c {CORES} --mem={MEMORY}gb --time={d-H:MM:SS} --pty bash
4. Load modules: module load miniconda/latest vscode_cli gcc/11.2.0 singularity/3.8.0
5. Run command: code tunnel
6. Go to your VSCode and on the left panel search for Remote Explorer
7. Click dancmpn01flunicphdom or dancmpn02flunicphdom

Instead of using .qmd files, now use .ipynb for all your analyses.

In the top right hand corner, you should be able to select kernels beginning with Singularity-

Available kernels

Kernel name	Description
Singularity-DESeq2	Differential expression analysis of bulk RNA-seq using DESeq2. Contains plotting libraries EnhancedVolcano and ggpubr
Singularity-Seurat	Single cell RNA-seq analysis with Seurat v4