zran_read returned error about halfpipe HOT 8 CLOSED

The error occurs when fmriprep tries to load the header of /mnt/BIACnfs/duhsnas-pri.dhe.duke.edu/morey_biac/Lab/Tours/rawdata/sub-COPTSDT09/func/sub-COPTSDT09_task-rest_bold.nii.gz to determine the number of volumes.
Could you try opening the file to see if it is a valid NIfTI image?

from halfpipe.

If it indeed is an invalid NIfTI image, then we may want to add code to HALFpipe to a) show a warning and then b) exclude the file from analysis.

from halfpipe.

I do open a few of the niftis and they seem fine in fslview. Is it saying a specific nifti is corrupt, or all of them?

from halfpipe.

The error refers to the file I mentioned above, so that would be the one to look at :-)

from halfpipe.

I checked all niftis and they look fine. I think I see where the issue is. Can subject names not have an underscore in them? The subjects are labeled like COPTSD_T01.nii.gz. In some of the error log, it is correct. But in the participant list, the underscore is dropped and the program is looking for "COPTSDT01." For this subject set, I do anat/{subject}.nii.gz. It seems like there is an error if an underscore is in the subject variable. I will try it as COPTSD_{subject}.nii.gz but that doesn't always work for other sites.

The subjects are all labeled as COPTSD_T**, but the list looks like this:

Running fMRIPREP version 20.2.1+halfpipe.2:
│ * BIDS dataset path: /ext/mnt/BIACnfs/duhsnas-pri.dhe.duke.edu/morey_biac/Lab/Tours/rawdata.
│ * Participant list: ['COPTSDT01', 'COPTSDT02', 'COPTSDT03', 'COPTSDT04', 'COPTSDT05', 'COPTSDT06', 'COPTSDT07', 'COPTSDT08', 'COPTSDT09', 'COPTSDT10', 'COPTSDT11', 'COPTSDT12', 'COPTSDT13', 'COPTSDT14', 'COPTSDT15', 'COPTSDT16', 'COPTSDT17', 'COPTSDT18', 'COPTSDT19', 'COPTSDT20', 'COPTSDV01', 'COPTSDV02', 'COPTSDV03', 'COPTSDV05', 'COPTSDV06', 'COPTSDV07', 'COPTSDV10', 'COPTSDV11', 'COPTSDV12', 'COPTSDV14', 'COPTSDV15', 'COPTSDV16', 'COPTSDV17', 'COPTSDV18', 'COPTSDV19', 'COPTSDV20', 'COPTSDV22', 'COPTSDV23', 'COPTSDV24', 'COPTSDV25', 'COPTSDV26', 'COPTSDV27'].

from halfpipe.

I don't think it's about the file names. The error comes from a low-level function for decompressing the .nii.gz and loading it into memory.

I think the next thing to try would be to unzip the input files, i.e. to run something like for i in *.gz; do gunzip $i; done in the folder with the image files.

from halfpipe.

Ok, I did some more digging. It seems like this is an issue with an older version of indexed_gzip (see pauldmccarthy/indexed_gzip#46). The most recent latest version has upgraded that library, so it should fix this issue. If you have time, it would be great if you could try again with the new version :-)

from halfpipe.

Will close this for now. Please re-open the issue if it turns out that the new release does not solve the issue :-)

from halfpipe.