Fastest way to preprocess data from 1 subject with multiple (60) sessions on computing cluster

[Avijit-Chowdhury1]

Dear all,

Does fmriprep run preprocessing of multiple sessions from one subject (it is case-study data) in parallel?

I am trying to run fmriprep on my Uni's computing cluster but it is taking more than a week. What would be the fastest way to accomplish this task?

[effigies]

Probably the fastest thing would be to use --anat-only to run through the anatomical workflow. Once that's done, I would run sessions more-or-less indepdendently:

Job 1:

fmriprep /data /out participant -w /work --anat-only $OPTS

Job N+1:

echo '{"bold": {"datatype": "func", "session": "N", suffix="bold"}}' > filter_file_N.json
fmriprep /data /out participant -w work --bids-filter-file filter_file_N.json $OPTS

If it makes sense to run multiple sessions in the same job, you can pass a list of sessions:

echo '{"bold": {"datatype": "func", "session": ["M", "N"], suffix="bold"}}' > filter_file_MN.json
fmriprep /data /out participant -w work --bids-filter-file filter_file_MN.json $OPTS

---

After the anatomical job is done, then the sessions should be runnable in parallel, reusing the same working directory.