Thanks for providing a detailed question.
From what I understand, I think you are just looking for sequences that have only one repeated pattern for G's i.e. "GGG". Anything else besides this pattern is unwanted. So I thought we could just do a "strfind" http://www.mathworks.com/help/matlab/ref/strfind.html to look for a pattern of "GGG". If you go through the documentation link I provided, you will notice how "strfind" will return the values of starting indices for the pattern it is searching for. These indices will be helpful in eliminating the sequences of the form that have "AGGGA", for example, in N. So the idea is simple, first do an "strfind" and locate indices for each string that has the "GGG" pattern. Second, eliminate sequences with indices that are not allowed, for example, only the indices of 7 and 16 correspond to valid indices if the length of N is 6. You can even come up with a formula for the "valid sequence indices". For example, length of N = (total sequence length - 6)/3. Valid indices for "GGG" pattern = (length of N + 1) and (2*length of N + 3 + 1) etc. I apologize in advance, if I have committed any arithmetic errors in providing the above example formula.
Also, you may need to loop through your data for this or if all of your data is stored in a cell array, you can take the shorter route of using "cellfun" http://www.mathworks.com/help/matlab/ref/cellfun.html.
Have fun!
