containerize modules

[sreichl]

• Containerize using snakemake i.e., add a Dockerfile?
• does not cost anything, can be automatically created (every time something changes in the software)
• automatize this step using GitHub actions?
• Singularity ****support: use singularity functionality to automatically create a container and set it up on DockerHub to trigger image build
  • issue in rule_all: /usr/bin/bash: line 1: snakemake: command not found
    • two solutions: install snakemake additionally (snakemake inception?!) or manual report generation (if this, what signals the successful completion to the user?) → solved
  • issue: errors occur immediately when submitting jobs with singularity: try it with local execution only → works (suspicion: module singularity is not loaded in the jobs → it is...)
  • sounds like that issue, but path solution did not work: Combining --use-conda and --use-singularity fails in some cases snakemake/snakemake#304
  • https://snakemake.readthedocs.io/en/stable/snakefiles/deployment.html#containerization-of-conda-based-workflows
  • https://slides.com/johanneskoester/snakemake-6#/10
  • idea from PC
    you could always try the exact opposite:
    put export SBATCH_EXPORT=ALL
    in your .bashrc, open a new terminal, activate a new conda env, start a new jupyterhub interactive job, and then launch snakemake

modules

• put list of all modules here