Ilifu slurm cluster configuration

To use, run the pipeline with -profile ilifu. This will use the ilifu.config which has been pre-configured with a setup suitable for the Ilifu slurm cluster. In particular it configures the use of Singularity containers and chooses an appropriate queue based on the memory required and the need (or lack of need) for a GPU (as specified using the accelerator attribute of the Nextflow process).

Running the workflow on the Ilifu slurm cluster

Mextflow is made available using the Environment Modules system. Load the Nextflow module with module load nextflow.

Do not run Nextflow directly on the login node, rather write a script that executes module load nextflow and then runs the workflow with nextflow -profile ilifu.

Singularity containers are cached in the directory specified by NXF_SINGULARITY_CACHEDIR. Ensure that this directory exists and that the environment variable is set to ensure that you re-use Singularity container images.

An example script illustrating the execution of a (fictional) workflow follows:

#SBATCH -J my_pipeline
module add nextflow
nextflow run nf-core/my_pipeline -profile ilifu \
    -w /path/to/some/dir/work \
## clean up on exit 0 - delete this if you want to keep the work dir
if [[ $status -eq 0 ]]; then
    rm -r /path/to/some/dir/work

Memory, CPU and time limits can be imposed by providing --max_memory, --max_cpus and --max_time paramaters to the workflow run.

To check which versions of Nextflow are installed on the cluster and which is the default version, use the module avail nextflow command. If a newer version is required, contact Ilifu support.

Config file

See config file on GitHub

params {
    config_profile_description = """
    Ilifu ( slurm cluster profile provided by nf-core/configs.
    config_profile_contact = 'Peter van Heusden (@pvanheus)'
    config_profile_url = ''
    max_memory = 1500.GB
    max_cpus = 96
    max_time = 336.h
singularity {
    enabled = true
    autoMounts = true
process {
    beforeScript = """
    module load singularity
    executor = 'slurm'
    queue = { task.accelerator != null && task.accelerator.contains('nvidia') ? 'GPU' : (task.memory >= 250.GB ? 'HighMem' : 'Main' ) }