Using `nf-core` Pipelines in the HPC

This page details running nf-core pipelines in the biostat partition of the KUHPC using the shared Nextflow configuration file.

nf-core is a global community of people that collaborate to build open-source Nextflow pipelines. They have pipelines for many common bioinformatics data analyses. For more information about the nf-core community and the curated pipelines available, please visit their documentation. If you have questions about using a pipeline, you are encouraged to connect directly to the team that maintains the specific pipeline you are using via their Slack, which you can request to join on this page.

Using the shared `ku_crc.config` file

In order to have the Nextflow pipeline properly run on our HPC system, users need to use a configuration file, or config file, specifically created to the properties of our HPC. Rather than have every user create their own, the biostat partition maintains a shared Nextflow config file located at /kuhpc/work/biostat/sw/internal_pipelines/nextflow/ku_crc.config. This file configures Nextflow to submit processes to the SLURM scheduler on the correct partition/account, sets up the shared Singularity container cache and bind paths, and sets sensible default resource requests (CPUs, memory, time) per process. Passing this file with -c when you launch a pipeline means you don’t need to write or maintain your own infrastructure config.

Benefit of the shared ku_crc.config file

Using the shared config means you don’t need to figure out SLURM account/partition settings, container cache directories, or per-process resource defaults yourself. This saves setup time and keeps Singularity images from being re-downloaded by every user.

To run an nf-core pipeline with the shared config, first load Nextflow:

module load nextflow

Then run your pipeline, passing the shared config with -c. No -profile is needed since ku_crc.config already configures the executor and container settings:

nextflow run nf-core/<pipeline> -c /kuhpc/work/biostat/sw/internal_pipelines/nextflow/ku_crc.config

Submitting `nf-core` pipeline jobs

nf-core pipelines are typically launched from a submitted job whose only responsibility is to run the head Nextflow process. Nextflow itself then submits each pipeline step as its own SLURM job using the executor settings from ku_crc.config. A submitted job to run an nf-core pipeline would look something like this:

#!/bin/bash
#SBATCH --job-name=nfcore_pipeline
#SBATCH --output=nfcore_pipeline_%j.out
#SBATCH --error=nfcore_pipeline_%j.err
#SBATCH --time=24:00:00
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=4G

source ~/.bashrc

module load nextflow

nextflow run nf-core/<pipeline> -c /kuhpc/work/biostat/sw/internal_pipelines/nextflow/ku_crc.config

Important note for submitted jobs

Submitted jobs must source ~/.bashrc before calling module load nextflow. Without it, the module system will not be initialized in the job’s shell and module load will fail, even if it works fine when you run it interactively. This allows the submitted job to have access to the variables such as $WORK and $HOME that are associated with your account.

Using the shared ku_crc.config file

Submitting nf-core pipeline jobs

Using the shared `ku_crc.config` file

Submitting `nf-core` pipeline jobs