Hi All,
I am using xcpengine (singularity) to process my results from fmriprep. Overall I’ve found it to be an excellent way to obtain ROI timecourses for subsequent network-based analyses. One hurdle, however, has been figuring out how to properly model task-driven activity as a confound.
From my understanding, I should (and have) create(d) a file with columns for each confound of interest and then 1s or 0s in each row (corresponding to each TR) for whether or not the condition of interest is present. There is no header to this file. The number of rows = num TRs. I then edit my design file to include a line such as:
condound2_custom[2]=/path/to/file
So far I have tried unsuccessfully to implement this for a single subject. xcpengine finishes without throwing an error, but the confound appears to be simply ignored.
Does xcpengine only accept 1D files for this? If so, what is the best way to convert my csv/txt/mat file to a 1D file? Also, I assume the path should include the bind point? Do I need to somehow reference this confound in the regress module? And, lastly, once I figure out how to make this work (with help from this community!), I will need to reference a different task file for each subject, which I think should be done using {[sub]}, but I’m not completely sure on this either.
Any guidance on these matters would be much appreciated. Thank you!
Adam