EEGlab LIMO data analysis issue

[arnodelorme]

Transferred from https://github.com/sccn/eeglab/issues/350

Hi,

Do you provide assistance with data analysis using the LIMO plugin? I have run the data set twice and both times the beta files are empty, although the program is generating the output files for each subject. The other steps to the analysis worked fine and I can compute channel and component measures when using just EEGlab without LIMO.

Thanks, Jay

[arnodelorme]

@sasis001 I have transferred your issue. Would you mind providing access to the data?

[sasis001]

I have shared a google drive folder consisting of the files.

Thanks, Jay

[arnodelorme]

Thanks. We will look at it.

[CPernet]

step 1 - run OLS (although WLS should have give you a message) if it works - but no message that WLS fail let us know, thx

[sasis001]

I ran the data with the OLS option and the analysis went through for the channel measures. However, when I tried the same thing with the component analysis it did not go through for any of the measures (ERP, Spectrum, ERSP). I could view the components on EEGlab with preclustering. Hence, the components are all generated for the different measures.

Excerpt from the error message generated- Error using load Unable to read file 'Yr'. No such file or directory. File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\limo_glm_handling.m at line 18 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\limo_eeg.m at line 473 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\external\psom\psom_run_job.m at line 214 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\external\psom\psom_run_job.m at line 128 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\external\psom\psom_run_script.m at line 437 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\external\psom\psom_run_script.m at line 323 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\external\psom\psom_pipeline_process.m at line 591 File C:\matlab\eeglab_current\eeglab2021.0\plugins\LIMO3.0\external\psom\psom_run_pipeline.m at line 417 File C:\Program Files\MATLAB\R2021a\toolbox\matlab\lang\parallel_function.m at line 832 File C:\Program Files\MATLAB\R2021a\toolbox\parallel\remoteParallelFunction.m at line 67

---

Checking outputs

---

The output file or directory C:\matlab\eeglab_current\eeglab2021.0\subj files\set\data files\sub-8\design1_GLM_Components_Time_OLS\Betas.mat has not been generated!

Thanks, Jay

[CPernet]

if you look into the psom report of subjects, it looks like to me it cannot generate the design and associate the IC data (Yr) and thus lino_glm cannot load something that doesn't exist -- see https://github.com/LIMO-EEG-Toolbox/limo_tools/wiki/debugging-1st-level-stats to get the actual error

thx

[CPernet]

PS: when running WLS didn't you get an error or warning message? like those here https://github.com/LIMO-EEG-Toolbox/limo_tools/blob/master/limo_batch.m#L536

[arnodelorme]

Is there something for me to do/test?

[CPernet]

let's see what the error is - I cannot see why it can process channels but not IC, maybe a bug related to clustering (as one reorganize the IC data so each subject has the same IC*frame matrix) ; unless of course you have time to play with that

[arnodelorme]

No time. Will help if my expertise is needed though or if there is a problem with EEGLAB/LIMO integration

[arnodelorme]

Message from Jay

I appreciate your help with this. I have uploaded the files from the EEGlab component analysis and the errors from the LIMO component analysis (also see attached) to the shared google drive folder. I have also uploaded the files from the channel WLS analysis. I followed the instructions provided in the EEGlab workshops and 11 components were identified across all subjects via the initial EEGlab analysis.

https://drive.google.com/drive/folders/1opwBe6zwenfDK5tww9PfSFh7hUGJkwXq

Also, the channel level analysis only worked with OLS and not WLS. Is this ok and if not, why is the analysis not working for WLS?

Thanks, Jay

[CPernet]

For both channels and component, only your subject 1 is processed because in all other cases there is an issue with the number of trials extracted vs. the number of conditions defined -- that is your conditions do not match trials

[sasis001]

A few sets had to be excluded due to noisy data or other issues. This is the reason for unequal trials across the four conditions in the study. Is there an option to find out how many trials in each condition in the entire data set and a workaround to fixing the issue?

Thanks, Jay

[CPernet]

Unequal number of trials is not an issue, the problem is triggers and segments don't match, it's a preprocessing problem - not a LIMO issue.

On Thu, 22 Jul 2021, 20:14 sasis001, @.***> wrote:

A few sets had to be excluded due to noisy data or other issues. This is the reason for unequal trials across the four conditions in the study. Is there an option to find out how many trials in each condition in the entire data set and a workaround to fixing the issue?

Thanks, Jay

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/LIMO-EEG-Toolbox/limo_tools/issues/71#issuecomment-885129785, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABENIDWY2B2KEVLWUAXX7HDTZBNWRANCNFSM5AKB5S6A .

[sasis001]

the problem is triggers and segments don't match. This error is surprising. Can you say which step in preprocessing or how this can be fixed?

and/or should I do epoching at the subject instead of the study level?

[CPernet]

sorry, check with EEGLAB email-list, help, I do not know how this kind of problem occurs. As you can see from the LIMO tutorial (https://github.com/LIMO-EEG-Toolbox/limo_meeg/wiki/1.-Preprocessing) I epoch on the whole dataset and this usually works