[ENH] start bidsifying output #80
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #80 +/- ##
==========================================
+ Coverage 89.80% 90.28% +0.48%
==========================================
Files 11 11
Lines 500 525 +25
==========================================
+ Hits 449 474 +25
Misses 51 51
|
|
There's a big problem with this structure - our file system handles small files poorly, hence I came up with the solution of dumping everything into one h5 file for |
at the group level? or the participant level? in any case, how about we add an output_format flag to the CLI to allow BIDS output, otherwise it keeps the previous format? |
An illicit use of using
Sounds good. |
docs/source/usage.md
Outdated
There was a problem hiding this comment.
A bit extra to this PR but it felt like a minimum description on how to run a demo was necessary
docs/source/outputs.md
Outdated
|
|
||
| The output file name is: `group/atlas-<atlas_name>_meas-PearsonCorrelation_desc-<denoising_strategy>.h5` | ||
|
|
||
| ## Examples |
There was a problem hiding this comment.
added examples of the output generated by the demo
|
Note that this probably should trigger a minor version bump because the output data structure will be defo incompatible with what previous versions of the app were doing. |
There was a problem hiding this comment.
-
For the BIDS compliant output, I am inclined to just use
tsvas the output format. -
For the non-BIDS compliant h5 output, I want to make everything that's not the json file into one h5 file, so essentially revert to what it was like before but just change the dataset names inside the h5 file. Does this make sense? Sorry for the trouble
|
@Remi-Gau i had a chat with Pierre and reached the conclusion. we will follow bids through and through, and save things in tsv, so this will make things a lot easier. |
giga_connectome/postprocess.py
Outdated
| # dump timeseries to h5 | ||
| h5_filename = connectome_path / utils.output_filename( | ||
| source_file=Path(img.filename).stem, | ||
| atlas=atlas["name"], | ||
| extension="h5", | ||
| strategy=strategy["name"], | ||
| desc=desc, | ||
| ) |
There was a problem hiding this comment.
If we want to be fully BIDS compliant, shall we save time series to h5 as well?
There was a problem hiding this comment.
yeah am working on this
ideally maybe even in nifti, no?
There was a problem hiding this comment.
no nifty since we are not outputting seed based maps?
The only sensible nifti we can output from here is the denoised and smoothed nifti, but that would be out of scope IMO
There was a problem hiding this comment.
ok will dump that in tsv then
| if analysis_level != "participant": | ||
| raise Warning( | ||
| "Only participant level analysis is supported. Other levels " | ||
| "would be ignored, and will run the participant level." | ||
| ) |
There was a problem hiding this comment.
realizing I just completely removed this from the code
but I feel that given that this is group level is not allowed by the CLI this warning will never be hit in practice
closes #32
measentity to filename