Revision coming soon

Setting up you environment

Please, follow the instructions from this page

Pipeline usage

See the README.md on how to launch the pipeline outside of a job scheduler. You will probably want to submit the pipeline on the guillimin scheduler, we’ll see that in a few moments

vlpp-batch

The tool vlpp-batch helps you build scripts to submit to the guillimin scheduler automatically. The tool needs to know where is your data on the cluster. In this example, we suppose that your data is organized with this structure:

/path
  |--freesurfer
    |--subject_00_type1
    |--subject_01_type1
/other_path
  |--pet
    |--subject_00_20170308
      |--NAV_subject_00_20170308_1435.mnc
    |--subject_01_20170315
      |--NAV_subject_01_20170315_1126.mnc

Create a new directory in which you will run all the commands for the processing, for example:

cd <your-user-directory>
mkdir pet_processing
cd pet_processing

Participants file

With this kind of structure, you have to create a participants.tsv file with at least 3 columns. A tsv (tabulation-separated values) file is the same as a csv file but tabulation is used to separated data.

With participants.tsv created, submit your jobs to the scheduler. The tool creates batch files in the code directory and submit them.

vlpp-batch -f /path/freesurfer -p /other_path/pet -t participants.tsv --submit

It is best practice to launch vlpp-batch without the --submit option and review the batch files. After review, launch the line with the --submit option.

You can check the status of your jobs with qstat -u <your-username>. See this page to know more.

Configuration file

If you want to change the default parameter of the pipeline, create a config.cfg file inside the code directory and edit this file with your parameters. Read the default parameters file to understand the parameters of the pipeline.

mkdir code
touch code/config.cfg

For example, if you want to change the smoothing kernel to 8mm instead of the default 6mm, here is the content of code/config.cfg you should have :

pet2anat {
    pet {
        fwhm = 8
    }
}

Example with PAD dataset

Content of code/config.cfg

dataset = "PAD"

If you are working with the PAD dataset, base freesurfer and PET directories are already known. You still need to create participants.tsv but launching vlpp-batch is much simpler.

vlpp-batch -m pad -t participants.tsv --submit

vlpp-qa

When all the subjects are processed, launch this script from the same directory to produce Quality Assessment dashboards. Open qa/registration_T1w.html with firefox or chrome.

vlpp-extract-stats

When all the subjects are processed, launch this script from the same directory to extract the means of all of your subjects into one convenient file. It will be located inside the stats directory.

For example, group_time-4070_space-anat_ref-cerebellumCortex_suvr.tsv contains the SUVR means (cerebellum cortex as a reference) of each freesurfer and pibindex ROIs of all your subjects in the native space.