aslprep
aslprep copied to clipboard
--skip-bids-validation not skipping all bids validation
Summary
ASLPrep is exiting with a BIDS validation error even when using the "--skip_bids_validation" or "--skip-bids-validation" options at submission.
Additional details
- ASLPrep version: 0.6.0rc1
- Singularity version: 3.8.3
What were you trying to do?
I had a bunch of pCASL data processed using ASLPrep 0.2.8 that had "ok" results. I updated to 0.6.0rc1 to try reprocessing some data to see if I saw any improvements. I wanted to use the same command to best compare across versions.
What did you expect to happen?
I expected similar results, with likely some improvements.
What actually happened?
No output, just logs with error messages about BIDS validation.
Reproducing the bug
# LSBATCH: User input
singularity run -e -B /input/gfapprelim/:/whatever/gfapprelim/,/input/templateflow-d259ce39a/:/templateflow /input/pmacsPreps-0.2.3/containers/aslprep-0.6.0rc1.simg /whatever/gfapprelim//squids /whatever/gfapprelim//squids_aslprepout-0.6.0rc1 participant --participant_label sub01xxses01 --nprocs 2 --mem 40GB -w /whatever/gfapprelim//squids_wrkdir-0.6.0rc1 --skip-bids-validation --basil --scorescrub --output-spaces T1w asl MNI152NLin2009cAsym:res-2
Traceback (most recent call last):
File "/usr/local/miniconda/bin/aslprep", line 8, in
Can you share the tree for your dataset?
I had to make a "pseudo-BIDS" directory to get my data processed correctly using 0.2.8, and with the "--skip-bids-validation" flag I had not needed the dataset_description.json at all. With the 0.6.0rc1 it looks like it checks for the dataset_description.json regardless of the flag. But, I just created a dummy dataset_description.json file that contained the example contents ( {"Name": "Example dataset", "BIDSVersion": "1.0.2"} ) in my "pseudo-BIDS" directory ("/input/gfapprelim/squids/", in my command above) and things seem to be running now!
That's good to hear. It appears that pybids has some requirements, even when validation is disabled.
Fair enough. Feel free to close unless you have any reason to look into this further
I think it's worth at least improving the documentation a bit, so I'll keep this open until I've had a chance to update that.