Running on the Biowulf cluster (Slurm)

Prepare the pipeline

• clone the repository.

git clone https://github.com/NCI-CGR/GEMSCAN.git

This would clone the GEMSCAN github repository under your current folder, i.e. /GEMSCAN Alternatively, You could create a project folder say _my_variantCalling_project_ by ```mkdir my_variantCalling_project``` and then clone the GEMSCAN repo to it: ```git clone https://github.com/NCI-CGR/GEMSCAN.git ./my_variantCalling_project```

• create config/samples.txt and config/config.yaml as described in the Inputs and outputs section.

• go to workflow directory

  cd <cloned repo dir>/workflow

• You might want to create a submission script that looks like this

    #!/bin/bash
    #SBATCH --time=2-00:00:00
    #SBATCH -o <cloned repo dir>/workflow/snakemake.out
    #SBATCH -e <cloned repo dir>/workflow/snakemake.err
  	
    module load singularity
  	
    export SINGULARITY_CACHEDIR=/data/<yourusername>/
  	
    eval "$(conda shell.bash hook)"
        conda activate snakemake
  	
    snakemake  --profile profiles/biowulf --use-singularity --singularity-args "--bind /data/$USER,/fdb,/lscratch" --jobs 100
  

• If you have lots of samples, you may want to keep your main Snakemake workflow manager running until all the tasks finishes and it may exceed the maximum wall time of the norm partition on Biowulf. In that case you may want to use the unlimited partition, i.e.

    #!/bin/bash
    #SBATCH --partition=unlimited	
    #SBATCH -o <cloned repo dir>/workflow/snakemake.out
    #SBATCH -e <cloned repo dir>/workflow/snakemake.err
  	
    module load singularity
  	
    export SINGULARITY_CACHEDIR=/data/<yourusername>/
  	
    eval "$(conda shell.bash hook)"
        conda activate snakemake
  	
    snakemake  --profile profiles/biowulf --use-singularity --singularity-args "--bind /data/$USER,/fdb,/lscratch" --jobs 100
  

  One thing to pay attention to is that if any data you use is not on the list of directories listed in the example script above, you will have to add them to –singularity-args after –bind, so that singularity will bind those folders as well.

• Make sure you have a relatively updated Snakemake in your conda environment (>=5.26.1). Then activate the snakemake conda environment created in the installation step

  conda activate snakemake

• submit the script

  sbatch --time=48:00:00 <script>

• monitor the jobs

  squeue -u <your_username>

• The log files are in <cloned repo dir>/workflow/logs/, it is also useful to look at <cloned repo dir>/workflow/snakemake.err to see the snakemake running progress and potential errors