Cluster

If you’re using a computing cluster with SLURM as the resource manager, make sure to adjust the nextflow.config file to set the appropriate executor. You can then submit the job using the following command:

Singularity Convert

container_name=cmrlab/megprep
version=0.0.3
singularity build ${container_name}_${version}.sif docker-daemon://${container_name}:${version}

Run MEGPrep on Slurm

#!/bin/bash

#SBATCH --job-name=multi
#SBATCH --partition=cpu1,cpu2,fat

#SBATCH --nodes=1
#SBATCH --ntasks=1

#SBATCH --mem=4G
#SBATCH --cpus-per-task=4

#SBATCH --output=%x.%j.out
#SBATCH --error=%x.%j.err

nextflow run megprep.nf -c nextflow.config -resume

This command will run the MegPrep pipeline on the cluster while managing resource allocation automatically.

nextflow.config

//megprep.slurm.cpu.config

// Define the working directory for Nextflow
workDir = "/lustre/grp/gjhlab/liaop/datasets/SMN4Lang/output_dir_v3/work"

singularity.enabled = true
singularity.autoMounts = true
singularity.runOptions = '-e \
    --env NUMBA_CACHE_DIR=/tmp/NUMBA_CACHE_DIR \
    --env MPLCONFIGDIR=/tmp/MPLCONFIGDIR/ \
    --env DISPLAY=99 \
    --env QT_QPA_PLATFORM=xcb \
    --env MESA_GLSL_VERSION_OVERRIDE=150 \
    --env MESA_GL_VERSION_OVERRIDE=3.2 \
    --env XDG_RUNTIME_DIR=/tmp/NUMBA_CACHE_DIR \
    -B /lustre/grp/gjhlab/liaop/ \
    -B /lustre/grp/gjhlab/liaop/datasets/SMN4Lang/smri/ \
    -B /lustre/grp/gjhlab/liaop/license.txt:/fs_license.txt \
'



process {

    executor = 'slurm'

    queue = 'cpu1,cpu2,fat'

    clusterOptions = { " --chdir=${workDir}" }

    container = '/lustre/grp/gjhlab/liaop/codes/megprep/megprep_0.0.3.sif'
}


// ------------------------------
// Global Parameters Configuration
// ------------------------------
params {
    dataset_dir = "/lustre/grp/gjhlab/liaop/datasets/SMN4Lang" // "/input"   // Input data directory
    output_dir = "/lustre/grp/gjhlab/liaop/datasets/SMN4Lang/output_dir_v3" //"/output" // nextflow logs directory
    preproc_dir = "${params.output_dir}/preprocessed" // Output results directory

    code_dir = "/program/megprep"   // all codes for preprocessing.
    t1_input_type = "nifti"                      // 'dicom' or 'nifti'
    is_bids = true                     // Whether the data is in BIDS format

    do_fs = true
    anatomy_preprocess_method = "deepprep"
    fs_subjects_dir = "/lustre/grp/gjhlab/liaop/datasets/SMN4Lang/smri_v3" //"/smri"

    //deepprep
    deepprep_device = "cpu"
    t1_bids_dir = "${dataset_dir}"
    fs_license = "/lustre/grp/gjhlab/liaop/license.txt"
    ...