README.md

snakemake-cluster

Simple snakemake cluster submission wrapper.

This command submits the snakemake master process to the cluster, which then launches the individual jobs with the correct cluster arguments.

Usage:
  snakemake-cluster -J myjob -- [snakemake-args]

Options:
  --cluster-config: Set cluster.yaml (default: cluster.yaml)
  --cluster-submitter: Set cluster submitter
    (default: ~/.config/snakemake/cluster-submitter.py)
  --jobscript: Set jobscript (default: ~/.config/snakemake/jobscript.sh)
  --verbose: Print actual submissions commands
  --lock: Don't set option --nolock
  --no-rerun-incomplete: Don't set option '--rerun-incomplete'
  --no-user-cluster-config: Don't use user's cluster config as the base template
    (default: ~/.config/snakemake/cluster.yaml)

Example: snakemake-cluster -J myjob --snakefile do_stuff.sf last_rule

Installing

This wrapper requires the qbsub wrapper script around qsub located here: https://github.com/broadinstitute/qbsub

For a simple one liner to download to ~/bin:

mkdir -p ~/bin && wget https://raw.githubusercontent.com/broadinstitute/qbsub/master/bin/qbsub -o ~/bin/qbsub && chmod +x ~/bin/qbsub

Simply clone the repository and run make. This will by default install the snakemake-cluster script to ~/bin. Therefore you will need to add ~/bin to your $PATH.

git clone https://github.com/yesimon/snakemake-cluster.git
cd snakemake-cluster
make

Snakemake Cluster Config

The cluster config is where you store cluster resource requirements for each type of rule. The first location is ~/.config/snakemake/cluster.yaml which applies to all projects, while the project specific config at <project_dir>/cluster.yaml can be used to add additional cluster configs, overriding the user cluster.yaml if there are any conflicts. The expected format is below:

<project_dir>/cluster.yaml

__default__:
  n: 1
  queue: short
  mem: 8
  log: '{log}'

bwa:
  queue: long
  time: '18:00'
  n: 2
  mem: 4

Time is listed in hours and is optional, n corresponds to the number of cores, mem is the maximum memory in gigabytes for the job. The log: '{log}' directive is necessary for logging to be redirected to the correct location. An example corresponding Snakefile:

<project_dir>/Snakefile

rule bwa:
    input: 'fastq/{sample}.1.fastq', 'fastq/{sample}.2.fastq'
    output: 'alignment/{sample}.bwa.sam'
    log: 'log/alignment/{sample}.bwa.log'
    shell:
        '''
        bwa mem -t 2 -a ~/db/bwa {input[0]} {input[1]} > {output}
        '''

rule mapped_bam:
    input: 'alignment/{sample}.bwa.sam'
    output: 'alignment/{sample}.bwa.bam'
    log: 'log/alignment/{sample}.samtools_mapped.log'
    shell:
        '''
        samtools view -h -b -F 4 -o {output} {input}
        '''

In this case, the bwa rule jobs will get the bwa cluster config resources, while the mapped_bam rule will get the __default__ cluster config resources.

Snakemake Pipeline Layout

This wrapper doesn't make any assumptions about the format of the pipeline. It is recommended to have a log directive for all rules.