HALFpipe

HALFpipe

HALFpipe can be run in a apptainer/singularity container on ALICE.

In the manual is some information about running on a HPC cluster:

GitHub - HALFpipe/HALFpipe: ENIGMA HALFpipe is a user-friendly software that facilitates reproducible analysis of fMRI data

The container images are located at: 

/zfsstore/shared/singularity_images/h/halfpipe-1.3.2.sif

The singularity command is now called apptainer, so starting an interactive shell and starting the container will start the userinterface:

$ salloc -p cpu-short  --time=30:00

$ cd $TMPDIR

$ apptainer run /zfsstore/shared/singularity_images/h/halfpipe-1.3.2.sif --use-cluster

According to the manual HALFpipe will expect the in and output files in a specific path /ext which you may have to bind mount for the container to find.

Ie. --bind /data1/projects/<PROJECT_DIR>:/ext but in general --bind /:/ext also works.

An example of a resulting halfpipe batch job:

#!/bin/bash #SBATCH --job-name=halfpipe #SBATCH --output=halfpipe.log.txt #SBATCH --partition="cpu-zen4" #SBATCH --time=1-0 #SBATCH --ntasks=1 #SBATCH --cpus-per-task=2 #SBATCH --mem=21504M #SBATCH --array=1-221 if ! [ -x "$(command -v apptainer)" ]; then module load apptainer fi apptainer run \ --contain --cleanenv \ --bind /:/ext \ /zfsstore/shared/singularity_images/h/halfpipe-1.3.2.sif \ --workdir /ext/<PATH_TO_DATA> \ --only-run \ --uuid <UUID> \ --subject-list /ext/<PATH_TO_DATA>/subject-list.txt \ --subject-chunks \ --only-chunk-index ${SLURM_ARRAY_TASK_ID} \ --nipype-n-procs 2 \ --keep some