Table of contents

1. Using GIFT GUI
   1. Run Matlab GUI
   2. Add GIFT path
   3. Run GIFT
   4. Example analysis: Group ICA on auditory oddball data
2. Using GIFT Batch Scripts
   1. Serial/Parallel Mode
3. Memory considerations

GIFT is an application supported by the NIH under grant 1RO1EB000840 to Dr. Vince Calhoun and Dr. Tulay Adali. It is a MATLAB toolbox which implements multiple algorithms for independent component analysis and blind source separation of group (and single subject) functional magnetic resonance imaging data.

Using GIFT GUI

Run Matlab GUI

To use Matlab GUI, select Matlab from the TReNDS interactive apps on https://hemera.rs.gsu.edu/.

Add GIFT path

>> addpath(genpath('/trdapps/linux-x86_64/matlab/toolboxes/GroupICATv4.0b/'))

Run GIFT

>> gift

Example analysis: Group ICA on auditory oddball data

Change to the data directory and create a folder. Then click “Setup ICA Analysis” in GIFT and select the output folder.

$ cd /data/users2/salman/mind/data/
$ mkdir ica_example/

Select all subject data and all the ICA options.

Click the “Run Analysis” button in the GIFT window and start the analysis.

Once analysis is done, visualize the output.

Using GIFT Batch Scripts

Create the following batch script gift_batch.m:

inputFile = '/path/to/input/file';
addpath( genpath( '/trdapps/linux-x86_64/matlab/toolboxes/GroupICATv4.0b/icatb/' ) )
icatb_batch_file_run(inputFile)

Sample batch input files are located in the GIFT toolbox folder, e.g.: /trdapps/linux-x86_64/matlab/toolboxes/GroupICATv4.0c/icatb/icatb_batch_files/Input_spatial_ica.m. Please make a copy of it and make necessary changes.

Execute the script using the following commands (in a SLURM job script or interactive node):

$ module load matlab/R2022a
$ matlab -batch 'gift_batch'

Serial/Parallel Mode

If you select parallel_info.mode = 'serial'; in the GIFT batch input file, you can increase the number of allocated CPUs in the SLURM job script for faster analysis, e.g.

#SBATCH -c 5
#SBATCH --mem=80g

If you select parallel_info.mode = 'parallel'; in the GIFT batch input file, you can set the number of parallel pools equal to the number of allocated CPUs (in the #SBATCH -c directive) in the SLURM job script, e.g.

# in job script 
...
#SBATCH -c 5

# in batch input file
...
parallel_info.num_workers = 5;

Memory considerations

Note that GIFT tells you the amount of memory you need to run the analysis. Please see this section for how to properly configure SLURM and Matlab for running the analysis successfully.