Analyses_v2 timecourses overwriting output csv (I believe)

[iamdamion]

I am using the most recent docker image from dcanumn/abcd-hcp-pipeline on docker hub. All seems to be working well in general, but I do see some weird behavior in the timecourses that are generated. This is not a pressing issue at all, but maybe something to look into.

What happened is we have two tasks and rest that we collected. There is onyl one csv per parcellation and the automatically created .csv had far fewer rows than TR in the concatenated rest. (I assumed it was only providing timecourse for the rest scans.) But when I looked into it and compared values on the ptseries to the csv, the values were from our saccade task, which is alphabetically last.

I think that the timecourses might be getting overwritten (and there is no task-specific naming in the file name). Example: Gordon_subcorticals.csv, Gordon.csv, etc.

Not an actual issue as we didn't even expect timecourses or any of those "extra" steps to be included in the pipeline, but I figured I'd let you know this is happening.

[arueter1]

Thanks @iamdamion for bringing this up! Something definitely seems off. I've assigned this to a few folks from our team to check out. We'll get back to you as soon as we are able. Thanks for letting us know that this isn't a pressing issue.

[iamdamion]

Thanks @iamdamion for bringing this up! Something definitely seems off. I've assigned this to a few folks from our team to check out. We'll get back to you as soon as we are able. Thanks for letting us know that this isn't a pressing issue.

I looked into the logs more. It looks to be happening during the DCAN Bold proc step at the end. And it looks like my idea above is what's happening: It's looping over the tasks, but not including the task name in the csv file paths, so it's just re-writing over and over. Hope that helps a little. :)