Hi Nicole and Stephen.

Suzanne here. I did some trouble shooting back in December when I first got the error. Below are my findings:

1. The error was not caused by the late addition of an ROI. When I checked the relevant .mat files (ROI_Subject00*_Condition00*.mat files stored in the preprocessing directory), the error is caused by the ART Repair preprocessing not writing out either the QA or scrubbing data for any subject past subject #3. The cells corresponding to these data are empty for all subjects after subject #3. This error at the ART Repair step is not related to a lack of disk space. Also, it does not matter which data set is subject #3, so the error is not tied to a particular data set.

2. Since the error seems to be occurring at the ART Repair step, re-running the denoising step does not fix this. I also still get the same error at the same point in the CONN processing if I completely start the entire CONN analysis from scratch.

3. The error only occurs when I run the ART Repair preprocessing in CONN v17f. If I take the exact same set of subjects and run the ART Repair preprocessing in CONN v16f, I can complete the full CONN analysis with no errors. I also do not get the error if I run my analyses with no ART Repair preprocessing in CONN v 17f. I have not tried any earlier versions of CONN v17.

4. This error does not occur when running either the ICA or dynamic FC, only when running the first-level ROI-to-ROI and Seed-to-voxel analyses.

5. Just to make things more confusing, I ran a subset of my original data set (first 10 subjects only) in CONN v17f with the ART Repair step, and all processing steps could be completed without error.

I am using the conn_batch command to load in my data from a .mat file, but I wait until I am using the CONN gui to add any preprocessing steps or select any CONN analysis options.

Unless something is going wrong with the conn_batch command in CONN v17f, the error does not seem to be due to anything the user is doing. It really does seem to be buried in something fundamental with the I/O of CONN. I do not know whether there is some compatibility problem with Matlab v2016b and CONN v17f or something else completely. I stopped digging once I could not consistently reproduce the error in CONN 17f (see point #5).

suzanne