Target setups thought dump

[yarikoptic]

is a BIDS derivative dataset like

the goal:

.datalad/
code/
   babs_config.yaml  
   containers/  -- submodule for repronim/containers (WiP)
sourcedata/
   rawbids/ -- submodule for "source dataset" (no point to push to ria if it is not in ria) (TODO)
rias/  # for now and .gitignored (TODO: aim to get rid of)
logs/
... outputs for processing ... 

and make babs init --config code/babs_config.yaml ...options . to dump results right in here.

meanwhile could differ due to babs 'intricacies' but strive to get to above.

So would it be result of running

mechababs --derivative mriqc THISPATH

and if fails.

[asmacdo]

Reframe (planning): this issue = decide the requirements for pipeline output layouts — settle what the output layouts must be, including fmriprep-anat published as a citable standalone deliverable (smriprep-like, "Option B") and the target derivative-dataset shape. Overlaps #7, #24 (compose-into-study), #25 (study dataset_description), and PennLINC/babs#369. (Title kept; this comment records the reframe.)

[asmacdo]

Next step: taking the derivitives produced by the june 1 shakeout, compose a bids-study in the style of opennueuro studies.

• 1 dataset per study, with both derivatives
• determine metadata layers, and how they can "bubble up"

maybe need a tool specific top level something, see _desc for BIDS so thinking like desc-mechababs_studies+derivatives.tsv

Reference:

• Formalize the concept of [{leading entities}_]{entity_plural}.{tsv,json} file(s) bids-standard/bids-specification#2283
• Formalize joint index support in .tsv via {"+".join(entities)}s.tsv files bids-standard/bids-specification#2273
• Community Review: BEP028 - Provenance bids-standard/bids-specification#2405

Build a "perfect" version composing the right metadata at the right level.. Clone openneurostudies to start with the correct structue, then compose 2 sample studies into it.

[asmacdo]

Decision to bank: derivative naming needs a stage axis.

{tool}-{ver}+mb1 collides when one tool emits multiple staged derivatives — fmriprep-anat / fmriprep-minimal / fmriprep-full are the same tool+version. So the layout has to carry the stage:

study-*/derivatives/{tool}-{ver}-{stage}+mb1     e.g. fmriprep-24.1.1-anat+mb1

Each stage is its own derivative subdataset under study-*/derivatives/. (Surfaced from the multi-cluster brainstorm in #13 — relevant because later stages consume earlier ones: minimal/full reuse anat.)

TODO: [ ] update the fmriprepDerivatives README to the staged naming