Lestropie
commented
3 years ago (note that it believes there are 2 sessions, but there is only one session...but mostly, it cannot find the data to work on)
This indicates that it believes that there is a total of two sessions within the BIDS dataset across all subjects, not necessarily two sessions within the one subject. But the answer to this is below.
Error when attempting load from "/bids_dataset": No raw or pre-processed T1-weighted images could be found in input directory "/bids_dataset" for session derivatives_MRtrix3_connectome-preproc_sub-219_ses-itbs
There is an issue arising here somehow in the parsing of session information. The session name should be "sub-219_ses-itbs", but instead the parent directories are being accumulated into such for some reason. This can be seen here:
mrtrix3_connectome.py: [DEBUG] mrtrix3_connectome.get_sessions() (from mrtrix3_connectome.py:4110): ['derivatives/MRtrix3_connectome-preproc/sub-219/ses-itbs', 'sub-219/ses-itbs']
I'm thinking one possibility here is that you have created the "derivatives" directory as a sub-directory of the input BIDS dataset, as can be seen in your singularity run call:
... --bind /groups/dkp/Nifti:/bids_dataset --bind /groups/dkp/Nifti/derivatives:/outputs ...
This I would have expected to result in the BIDS validator failing. A quick search suggests that the BIDS validator may have been hard-coded to ignore the contents of any derivatives/ sub-directory from root, in which case it's been deemed permissible and my code probably needs to do the same. I also seem to recall having had a discussion with the BIDS crew regarding what relative filesystem arrangements between raw and derived data should be permitted or not, though I'm having trouble finding it. For instance, the BIDS raw directory would ideally be mounted read-only by an application, and so having the output being written to a sub-directory of such is counter-intuitive to me personally.
So at your end, I would suggest placing your "derivatives" directory somewhere other than within the BIDS raw directory, as this seems to be confusing my session parsing code.
dkp
Hi Robert, I'm feeling dumb here. The summary is, I ran the preproc level with no issue, but at the participant level, the container cannot find the relevant data. I've tried specifying the session and not specifying it (figuring it would do everything it found if I did not specify...consistent with the preproc step).
The preproc stage created appropriate looking files:
Participant Level
Next, I tried the participant stage several times, both with and without specifying the session.
Within seconds, I get the following log:
(note that it believes there are 2 sessions, but there is only one session...but mostly, it cannot find the data to work on)
Debug Mode