BIDS App PyMVPA chnk length error

pymvpa
#1

I’ve used the participant_prep on my subjects, resampled my masks (ref), both successful. However, when I attempt participant_test, I receive an error that the chnk[sic] length is incorrect. If I select individual participants, the number can change (despite the protocol being the same for everyone). I have four runs of 238 volumes each, for a total of 952 volumes. The chnk error is sometimes that 1904 doesn’t match 952, or 3504 (as I recall) doesn’t match 952.
My data were previously processed with fmriprep.
Thanks for any insight.

#2

I’ve been around the world looking at an SPM solution (magic numbers in the code), and attempting to install PyMVPA on macOS (lots of clashes with python versions, etc.). So installed NeuroDebian on VirtualBox, which comes with its own challenges.

I’d really like to get the Docker going, and would really appreciate any guidance on the “chnk length” errors.

#3

Would appreciate any input.

#4

After working on my other aims awhile, I’m back to this error message. Would appreciate any help.
I preprocessed with fmirprep, then for each subject I ran
./mvpa 01 etc. for each subject 01-15, which ran the script:

docker run -i --rm \
	-v /Users/john/CloudStation/School/Research/fmri/aim1:/data:ro \
	-v /Users/john/CloudStation/School/Research/fmri/aim1/derivatives/pymvpa:/outputs \
	bids/pymvpa \
	--skip_bids_validator \
	/data /outputs participant_prep -k ao -p $*

Then ran:
./mvpa2 -c altdev famsalt famssame intalt intsame which calls the following script:

#!/bin/bash

docker run -i --rm \
	-v /Users/john/CloudStation/School/Research/fmri/aim1:/data:ro \
	-v /Users/john/CloudStation/School/Research/fmri/aim1/derivatives/pymvpa:/outputs \
	bids/pymvpa \
	--skip_bids_validator \
	/data /outputs participant_test -k ao $*

Which produced deprecation warnings and the following error:

WARNING: Detected incorrect (nan) scl_ fields. Resetting to scl_slope=1.0 and scl_inter=0.0
 * Please note: warnings are printed only once, but underlying problem might occur many times *
Traceback (most recent call last):
  File "/code/run.py", line 281, in <module>
    fds.sa['chnks'] = chunks_labels  # we call this sample attribute 'chnks' so later it won't be mistaken for
  File "/usr/lib/python2.7/dist-packages/mvpa2/base/collections.py", line 599, in __setitem__
    str(self)))
ValueError: Collectable 'chnks' with length [2856] does not match the required length [952] of collection '<SampleAttributesCollection: time_coords,time_indices>'.

As noted back in March, the length is a multiple of the number of volumes. Should by 952, but varies.