PROFILE Configuration

All nf-core pipelines have been successfully configured for use on the the Cybertron HPC at Seattle Children Research Institude (SCRI), Seattle, WA.

To use, run the pipeline with -profile PROFILENAME. This will download and launch the pipeline using seattlechildrens.config which has been pre-configured with a setup suitable for the Cybertron cluster at SCRI. Using this profile, a container with all of the required software will be downloaded.

Project info

This config file is created for the use on the Cybertron HPC at Seattle Children Research Institude (SCRI), Seattle, WA. Using this config will pre-configure a set up suitable for the Cybertron HPC. The Singularity images will be downloaded to run on the cluster. The nextflow pipeline should be executed inside of the Cybertron system.

Below are mandatory information SCRI

Before running the pipeline you will need to create a Nextflow environment on mamba. You can load Singularity using the environment module system on Cybertron.

Create a Nextflow mamba environment

  1. Create nextflow.yml file containing the following content. This YAML file can be utilized to set up a mamba environment, specifying both the version of Nextflow and the environment name.
name: nextflow
channels:
  - bioconda
  - conda-forge
dependencies:
  - python>=3.9
  - nextflow==23.10.0
  - nf-core==2.10
  - graphviz
  1. Setting channel priority

Make sure that channel priority is set to flexible using the following comments:

# print your current conda settings
mamba config --describe channel_priority
# set to flexible if not already done
mamba config --set channel_priority flexible
  1. Create the Nextflow mamba environment
mamba env create -f nextflow.yaml
  1. Running in HPC (Cybertron)

Please look into RSC-RP/nextflow_scri_config for details.

# activate enviornment
mamba activate nextflow
module load singularity
 
# to list all the projects with project codes you are authorized on HPC
project info
 
# example to run nextflow pipeline (please replace with your own project code and module)
nextflow run -c 'conf/seattlechildrens.config' \
    [nf-core/module_name] \
    -profile test,PBS_singularity \
    --project ["your_project_code"] \

You can find more information about computational resources here. You have to be an employee of SCRI to access the link.

Config file

See config file on GitHub

seattlechildrens.config
//Create profiles to easily switch between the different process executors and platforms.
 
//global parameters
params {
    config_profile_description = 'The SCRI (seattle childrens research institute) cluster profile'
    config_profile_contact = 'Research Scientific Computing (@RSC-RP)'
    config_profile_url = 'https://github.com/RSC-RP'
 
    // SCRI HPC project params
    queue                       = "paidq" // freeq
    project                     = "${params.project}"
}
 
 
profiles {
    //For running on an interactive session on cybertron with singularity module loaded
    local_singularity {
        process.executor = 'local'
        singularity.enabled = true
    }
    //For executing the jobs on the HPC cluster with singularity containers
    PBS_singularity {
        process.executor = 'pbspro'
        process.queue = "${params.queue}"
        process.clusterOptions = "-P ${params.project}"
        process.beforeScript = 'module load singularity'
        singularity.enabled = true
    }
}