Question: Workflow for post-merge processing (e.g., MRIQC group stats) #326
Description
After babs merge and unzipping into the final derivative dataset form, some pipelines need additional processing that requires all subjects' outputs. For MRIQC, this is mriqc group which produces group_bold.tsv, group_T1w.tsv, etc.
This post-merge job:
- Needs all subject outputs (can only run after merge)
- Still needs to go through SLURM (can't run on login node)
- Is a single job, not parallelizable per-subject
Is there a suggested workflow for this? I considered a second babs project, but that seems a little heavy.
If theres not currently a way within babs, one approach we could adapt is Felix Hoffstaedter's bootstrap_MRIQC which handles this by including the group step in the merge script itself.