All nf-core pipelines have been successfully configured for use on the ABiMS cluster.
To use, run the pipeline with -profile abims
. This will download and launch the abims.config
which has been pre-configured with a setup suitable for the ABiMS cluster. Using this profile, a docker image containing all of the required software will be downloaded, and converted to a Singularity image before execution of the pipeline.
You will need an account to use the HPC cluster on ABiMS in order to run the pipeline. If in doubt see http://abims.sb-roscoff.fr/account.
Nextflow is installed on the ABiMS cluster.
You need to activate it like this:
module load nextflow slurm-drmaa graphviz
Nextflow manages each process as a separate job that is submitted to the cluster by using the sbatch command. Nextflow shouldn't run directly on the submission node but on a compute node. Run nextflow from a compute node:
# Load the dependencies if not done before
module load nextflow slurm-drmaa graphviz
# Run a downloaded/git-cloned nextflow workflow from
srun nextflow run \
/path/to/nf-core/workflow \
-profile abims \
--email my-email@example.org \
-c my-specific.config
...
# Or use let nf-core client download the workflow
srun nextflow run nf-core/rnaseq -profile abims ...
# To launch in background
sbatch --wrap "nextflow run nf-core/rnaseq -profile abims ..."
Or write a sbatch script
nfcore-rnaseq.sh
#!/bin/bash
#SBATCH -p fast
#SBATCH --mem=4G
module load nextflow slurm-drmaa graphviz
nextflow run nf-core/rnaseq -profile abims ...
Launch on the cluster with sbatch:
sbatch nfcore-rnaseq.sh
nf-core provides some test for each workflow:
module load nextflow slurm-drmaa graphviz
nextflow run nf-core/rnaseq -profile abims,test
To reduce the disk usage, nf-core images can be stored in a mutualized directory: /shared/software/singularity/images/nf-core/
The environment variable NXF_SINGULARITY_CACHEDIR: /shared/data/cache/nextflow
will indicate this directory to nextflow.
A local copy of several genomes are available in /shared/bank/
directory.