I've been asked: how does BLAST improve dynamic programming accuracy?

I do know that BLAST is based on heuristic approach that approximates the Smith-Waterman algorithm and that the Smith-Waterman is too slow for large databases but more accurate.

Can someone settle my conflict?

  • 3
    $\begingroup$ It doesn't improve DP, is less accurate but usable for large databases with excellent results $\endgroup$ – Martin Beracochea Jun 7 '13 at 16:11

So I got this answer from my lecturer:

Since Smith-Waterman algorithm is based on DP, we will get the best performance on accuracy, but there is a change that the homologous sequence is not with the highest probability (so better matching sequences will be "hidden" behind worse ones). Since blast is based on the heuristic approach, it overcomes the disadvantage described above, by adding biological information, and therefore make it more accurate from biologist's perspective.

  • 1
    $\begingroup$ BLAST uses a heuristic, but mainly to cut down the time spent looking for homologs, not to produce better alignment. So BLAST is not more accurate than S-W, but often times accurate enough. $\endgroup$ – Michael Kuhn Jun 8 '13 at 20:12
  • $\begingroup$ ok another constraint I forgot to mention which I believe is crucial - the two sequences are not homologous $\endgroup$ – Ranco Jun 8 '13 at 20:20
  • 2
    $\begingroup$ As Michael said, this answer is wrong; actually the question itself is wrong. BLAST uses heuristic for speed up, not for accuracy. It is less accurate than DP. BLAST does NOT "improve dynamic programming accuracy". With DP, we can also find suboptimal alignments. If we had an extremely fast computer, we would use DP, not BLAST. $\endgroup$ – user172818 Jun 14 '13 at 19:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.