duplex-tools icon indicating copy to clipboard operation
duplex-tools copied to clipboard

Try edlib for faster duplex matching

Open olawa opened this issue 2 years ago • 1 comments

Hi,

suggest to try edlib in place of parasail for alignment of pairs to get a speed up for longer comparisons. Potentially this can recover more good pairs if you go from 250 bp to full/long comparison. I tried it with 5kb and got ~15x speedup while keeping almost the same number of good matches.

olawa avatar Dec 15 '22 20:12 olawa

It's definitely worth a try, thanks for the suggestion. Initially we cared about getting template end and complement start match up really well, but getting more certain that we want to match up the full sequences. We would have to be careful about not allowing too short partial matches (SINEs/LINEs) as well though which will take some thinking about. If it looks ok I wouldn't be surprised if we switch to edlib.

Cheers

onordesjo avatar Dec 16 '22 10:12 onordesjo