Add information about extracting echo-wise preprocessed data from fMRIPrep to FAQ

[tsalo]

Summary

Our current recommendation when using fMRIPrep + tedana (instead of just t2smap) is to extract echo-wise preprocessed data from the working directory to the output directory, then run tedana on those native-space preprocessed files, and then apply the saved transforms to the native-space denoised data.

Additional Detail

Next Steps

1. Add info about this recommendation to FAQ.
2. Link to example code to extract the semi-preprocessed data from the fMRIPrep working directory.

[tsalo]

This script by @JulioAPeraza works for 20.2.1 (at least with the settings we've been using).

[tsalo]

It's probably best to point to a gist that we can easily update: https://gist.github.com/tsalo/83828e0c1e9009f3cbd82caed888afba.

We could even embed it in the docs, I think

<script src="https://gist.github.com/tsalo/83828e0c1e9009f3cbd82caed888afba.js"></script>

[mateuszpawlik]

The script seems to work with fMRIPrep 20.2.5
In the lab, we'd like to go through the fmriprep -> Tedana -> Tedana BIDS derivative process. Any hints on that? The FAQ section is rather vague.

[AaronAngJW]

Hi there,

I've recently tried to run the collect_fmriprep() script on https://tedana.readthedocs.io/en/stable/faq.html#collecting-fmriprepped-data but I could not successfully retrieve the required files needed. I'm currently running fmriprep v21.0.0.

For me, the only output seen was the string "Wrangling subject n" and then the script ended almost instantaneously. Any guidance on this?

**Update: The first change seems to be changing the following line, where bold_transform is I supposed, the new folder instead of the previous _bold_file_:

bf_dirs = sorted(glob(op.join(bb_wf_dir, "bold_transform*"))) #_bold_file_*

The next part I believe, has got to do with:

idx = bf_dir_list.index("sub-{0}".format(sub))

where files in the bf_dir_list do not follow the 'sub-' label. I am not sure how to fix this part as I don't know how the output of the working dir of earlier fmriprep versions look like (first time using!).

[tsalo]

While I understand why you might not want to re-run fMRIPrep on your data, fMRIPrep 21.0.0 does have an option to output the semi-preprocessed echoes (--me-output-echos). Given that your situation reflects an edge case, I think it would be best to open this as a tedana/fmriprep-tagged usage question on NeuroStars instead of here.