FIX: Repair FreeSurfer Dependency in Dockerfile (tcsh)

[psadil]

Closes #403

FreeSurfer's recon-all requires tcsh, but that shell was no longer being added in the Dockerfile. This installs it.

The lack of tcsh escaped testing during CI because recon-all had been skipped entirely (generally too expensive to run). This change also removes stats/wmparc.stats from the FreeSurfer derivatives for ds005, which ensures that at least some of the recon-all scripts are run (any dependencies that are only used in earlier stages would still be missed).

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (3b0dcdb) 76.88% compared to head (13cacd8) 76.98%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #404      +/-   ##
==========================================
+ Coverage   76.88%   76.98%   +0.09%     
==========================================
  Files          28       28              
  Lines        2042     2042              
  Branches      236      240       +4     
==========================================
+ Hits         1570     1572       +2     
+ Misses        441      438       -3     
- Partials       31       32       +1     

Flag	Coverage Δ
ds005	∅ <ø> (?)
ds054	45.97% <ø> (ø)
pytest	66.45% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[psadil]

Not sure what to make of that failure on ds005. autorecon3 was looking for a file that wasn't present, surf/autodet.gw.stats.lh.dat. FWIW, that is a file that is generated with this version when FreeSurfer derivatives are not already precomputed (and so generated by sMRIPrep).

[psadil]

Seems like the autodet.gw.stats.[lr]h.dat files are generated during autorecon2 but are missing from the ReconAllDevTable. They don't seem to be referenced in the nipype freesurfer interface: https://github.com/nipy/nipype/blob/b066807402ee07d908ce4f1b24ca69fef6b91809/nipype/interfaces/freesurfer/preprocess.py#L1387, which would explain why they weren't generated during this CI run.

[effigies]

Seems so. Maybe we can pick a different thing to delete. Parcstats?

[psadil]

Sure, I'll test that locally and report back whether it works

[psadil]

Wow, sorry, same basic error (now with lh.orig.premesh). I was trying with a participant that I had locally, but clearly I should be experimenting with ds005

[effigies]

We should probably just update the prerun FreeSurfer with a fresh run. I've been really procrastinating on that since the right way to do it would be to set up something that is just punch and go and I haven't thought much about the best way to do that.

[psadil]

A fresh run does sound cleaner. From what I can tell,

• something that would trigger autorecon3 won't work, because then recon-all will try to $DoWhiteSurfs, which requires that autodet.gw.stats.?h.dat file
• something that would trigger autorecon-hemi would try to recreate autodet.gw.stats.?h.dat, but (without setting $DoRemesh), recon-all will fail for lack of ?h.orig.premesh
• something that triggers -cortribbon seems to work, though that's one of the slower ones

[effigies]

Spent too much time this week trying to create a datalad dataset that we could keep updated with each version of FreeSurfer. Guess we'll need to keep kicking the can. Thanks for this!