I'd introduced raising an error if there was only one sequence to
cluster, but this was a mistake.
You can have only one sequence to cluster if the other sequences are <
min_match_len (or also I believe if all sequences are 'N's, in which case the an interval is considered a non-match interval).
Unit tests now ensure this is fixed.
I will be using make_prg quite a bit in the near future, on thousands of sequences, so any other bugs should pop up. Sorry about this!
I'm afraid the refactoring I did added a bug.
I'd introduced raising an error if there was only one sequence to cluster, but this was a mistake.
You can have only one sequence to cluster if the other sequences are < min_match_len (or also I believe if all sequences are 'N's, in which case the an interval is considered a non-match interval).
Unit tests now ensure this is fixed.
I will be using make_prg quite a bit in the near future, on thousands of sequences, so any other bugs should pop up. Sorry about this!