Trouble writing loop for pulling data out of separate files, rejecting data points, and saving into new file

I'm trying to pull one row of data out of a struct within files of different names(that are named according to a scheme represented by fileId below), reject some data points(which come from a different file with the same naming scheme), and save the data into a .mat file, so that all the data from the row of interest from each file is saved in one place. I'm having trouble constructing the loop so that it works. My code is below.

 n =[];
 fileId=[5 8 11 15 18 19 21 25 27 31 35 36 38 42 43];%length is 15
 out={};
 load('loop_test.mat')
 for m=1:15
    for n=fileId
    filename=sprintf('Subject%d.mat',n);
    load(filename);
    filename2=sprintf('EEGrej%d.mat',n);
    load(filename2);
    refEEG = EEG.data(8,:);%EEG is a struct within Subject%d.mat
        bad_times = [];
        for fill = 1:size(rej,1)%rej corresponds to the rejection indices in EEGrejection%d.mat
            bad_times = [bad_times round(rej(fill,1)):round(rej(fill,2))];
            end
            bad_times = unique(bad_times);%takes out non-unique values
            refEEG(bad_times) = [];
            out{1,m}=refEEG;
            save('loop_test.mat','out','-append')
    end
end

When I run it I get this error: Matrix index is out of range for deletion.

Error in looptest (line 21) refEEG(bad_times) = [];

But if I do this processing individually, it works. In my workspace filename and filename two end up having different fileID's and 'out' is only populated with data from one file.

[Walter Roberson]

We as outsiders have no reason to believe that the contents of rej column 1 and column 2 are definitely going to be valid row indices for EEG.data. For example we could easily suspect that there might be rej rows that contain 0 0

[Nicholas Jackson]

The round function ensures that rej columns correspond to valid row indices.

[Walter Roberson]

No, the round function ensures that rej columns correspond to integers. We as outsiders have no reason to believe that the integers will all happen to be valid row numbers. The numbers might include non-positives or might include values greater than the number of rows.

I suggest you

dbstop in error

and run. When it stops, examine bad_times to see what values are present and from there trace back in rej to see where they came from.

[Nicholas Jackson]

Thanks for your help!