Discussed with AFNI group at OHBM 2023

[cmaumet]

Interesting thoughts we had w/ @afni-dglen and @mrneont :

• we could use apsearch to update the inputs/outputs of all commands in https://github.com/bids-standard/BEP028_BIDSprov/blob/master/bids_prov/afni/description_functions.json
• use 3dinfo -history to get the provenance rather than starting from the proc file.
• uvars.json file provides a description of the important output files in an afni_proc folder
• the other idea is to use thd_mastery to remove a "input:" or "output:" selectors that could be added to each command by afni_proc when writing the proc file.

TODO: camille to email Paul with https://github.com/bids-standard/BEP028_BIDSprov/blob/master/bids_prov/afni/description_functions.json

[afni-dglen]

We had also discussed on how atomic the providence could be - afni_proc.py as the only single step, processing blocks within the generated script, or the individual AFNI commands, with the processing blocks choice as probably the most difficult to automate, I think. afni_proc.py has a relatively small set of inputs, but it has a large set of outputs, many described in the uvars.json file.

You already have a great start with the individual steps, so it's probably a good way to go.

Paul didn't get tagged properly in your post, so I'll add it in again here, and also Rick Reynolds, who will certainly be the best judge of these proposals. @afni-rickr , @mrneont

[cmaumet]

Thank you @afni-dglen! I have now updated my original comment to correctly tag Paul :)