Example Scripts¶

This page list a couple of example jobs to submit to SLURM for running Step 3 of HeuDiConv where actual BIDS sorting occurs. In general, the time, partition, number of cores, and memory should be sufficient for converting any dataset. Paths inside the script should be changed to fit the user's dataset and file structure.

Note

Replace anything inside <> (including those symbols) with the relevant information, whether it is a path, or something else. If they are left in, the scripts will not run.

Create Conda Environment¶

#!/bin/bash
#
#SBATCH --job-name=create-bids-env
#SBATCH --output=bids-env.txt
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --partition=express
#SBATCH --time=20:00
#SBATCH --mem-per-cpu=8G
#SBATCH --mail-type=FAIL

module load Anaconda3/2020.11

conda create -n bids
conda activate bids

pip install heudiconv==0.9.0

conda install -c conda-forge dcm2niix

Single Subject Job¶

This job assumes each subject only has a single session of data to convert and does not use the session variable.

Note

The conversion scripts assume dicoms are stored in the same folder structure as in the practical HeuDiConv example. If your directory structure differs, or there are multiple sessions, you will need to alter the path for the -d input to account for these differences

#!/bin/bash
#
#SBATCH --job-name=bids
#SBATCH --output=array-out/bids.txt
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --partition=express
#SBATCH --time=30:00
#SBATCH --mem-per-cpu=8G
#SBATCH --mail-type=FAIL

# set necessary variables
base_dir=<path/to/dataset>
subject=<base subject name>

# load the necessary Anaconda module and activate the virtual environment. The heudiconv executable is in ~/.local/bin, so add that to the PATH
module load Anaconda3/2020.11
conda activate bids

PATH=$PATH:~/.local/bin

# Run heudiconv
heudiconv -s $subject -d <path/to/find/all/subject/dicom/files> -o $base_dir/nifti -f <path/to/heuristic> -c dcm2niix -b --overwrite

# change the permissions of all of the files in the BIDS directory to have user and group write permissions
find $base_dir/nifti/<BIDS subject name> -exec chmod ug+w {} \;
find $base_dir/nifti/.heudiconv/<BIDS subject name> -exec chmod ug+w {} \;

Array Job¶

The array job will replicate the job across multiple subjects' data without having to create individual job scripts to submit. For more information about array jobs and how to submit them, read the UAB Research Computing documentation.

base_dir should be changed to the path to the dataset folder, the directory that contains the dicom folder and the newly created nifti folder.

This job assumes each subject only has a single session of data to convert and does not use the session variable.

Note

The conversion scripts assume dicoms are stored in the same folder structure as in the practical HeuDiConv example. If your directory structure differs, or there are multiple sessions, you will need to alter the path for the -d input to account for these differences

#!/bin/bash
#
#SBATCH --job-name=bids
#SBATCH --output=array-out/bids-%A-%a.txt
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --partition=express
#SBATCH --time=30:00
#SBATCH --mem-per-cpu=8G
#SBATCH --mail-type=FAIL

# set the path to the base directory and change directory to the dicom
# directory inside it.

base_dir=<path/to/dataset>

cd $base_dir/dicom

# load the necessary Anaconda module and activate the virtual environment. The heudiconv executable is in ~/.local/bin, so add that to the PATH
module load Anaconda3/2020.11
conda activate bids

PATH=$PATH:~/.local/bin

# Get the subject name based on the array index
# For example, if I wanted to get all the folders the began with UAB, it would be subs=(UAB*)
# These subjects should be in $base_dir, so plan accordingly.
subs=(<glob to get the names of all the subjects>) 
pid="${subs[$SLURM_ARRAY_TASK_ID]}"

# Run heudiconv
heudiconv -s $pid -d <path/to/find/all/subject/dicom/files> -o $base_dir/nifti -f <path/to/heuristic> -c dcm2niix -b --overwrite

# change the permissions of all of the files in the BIDS directory to have user and group write permissions
find $base_dir/nifti/sub-${pid//_} -exec chmod ug+w {} \;
find $base_dir/nifti/.heudiconv/${pid//_} -exec chmod ug+w {} \;