Support output spaces to arbitrary templates and cohorts

[mattcieslak]

The HBCD data will be registered to MNIInfant+cohort templates. We want to make sure these file names are used in the qsirecon scalar mappers

[tsalo]

This looks like a really clean solution. Do you know what would happen if there were multiple transforms available?

[mattcieslak]

there would be a crash. Probably better to exit with an error than let it crash downstream?

[tsalo]

Oh I see that now:

qsirecon/qsirecon/utils/bids.py

Lines 204 to 217 in 0b53f53

	for name, query in queries.items():
	files = layout.get(
	return_type="file",
	subject=subject_id,
	session=session_id,
	**query,
	)
	if len(files) == 1:
	anat_data[name] = files[0]
	elif len(files) > 1:
	files_str = "\n\t".join(files)
	raise ValueError(
	f"More than one {name} found.\nFiles found:\n\t{files_str}\nQuery: {query}"
	)

Users would get a pretty clear error message. Something like:

More than one acpc_to_template_xfm found.
Files found:
    /path/to/sub-01_from-ACPC_to-MNIInfant+1_mode-image_xfm.h5
    /path/to/sub-01_from-ACPC_to-MNIInfant+2_mode-image_xfm.h5
Query: {'datatype': 'anat', 'from': ['T1w', 'T2w', 'ACPC'], 'to': 'MNI152NLin2009cAsym', 'mode': 'image', 'suffix': 'xfm', 'extension': '.h5'}

[mattcieslak]

I'm happy with that error message!

[tsalo]

Actually, since the to value is set by QSIRecon as either MNI152NLin2009cAsym (default) or MNIInfant (infant mode), I think (1) you wouldn't get an error and (2) you won't catch MNIInfant+<cohort>. Does that sound right?

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 46.66667% with 8 lines in your changes missing coverage. Please review.

Project coverage is 47.90%. Comparing base (0b53f53) to head (f63f7ea).

Files with missing lines	Patch %	Lines
qsirecon/interfaces/anatomical.py	20.00%	4 Missing ⚠️
qsirecon/interfaces/scalar_mapping.py	33.33%	2 Missing ⚠️
qsirecon/utils/bids.py	66.66%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #184      +/-   ##
==========================================
+ Coverage   47.64%   47.90%   +0.26%     
==========================================
  Files          57       56       -1     
  Lines        7231     7198      -33     
  Branches      986      978       -8     
==========================================
+ Hits         3445     3448       +3     
+ Misses       3581     3544      -37     
- Partials      205      206       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[mattcieslak]

In my local testing with the cohorts it still finds them

[tsalo]

Oh... maybe that's related to how parse_file_entities doesn't find the +<cohort> in the entities?

[mattcieslak]

Right, the query finds it while ignoring the cohort part

[tsalo]

I worry that's more of a convenient bug than a feature, but I also think we'll have plenty of notice if pybids starts parsing these plus signs correctly and searching 'MNIInfant' stops finding 'MNIInfant+1' etc.

[mattcieslak]

We'll definitely need to add a test for MNIInfant inputs, but it will be a big task to get simulated data that looks like HBCD

[tsalo]

I think SDCFlows has a few tests where they just generate a dataset layout from a dictionary and make sure that the queries work as expected. We could probably add something like that just to make sure the collection steps are working.

[tsalo]

I think things will change once we let users specify the output spaces they want, but this is a huge step in that direction. Everything looks good to me!