[Bioperl-l] An update for my DNA Smith-Waterman code

[Aaron J Mackey]

On Mon, 27 Jan 2003, Yee Man wrote:

> 	It seems to me that means Phil Green's implementation technically
> isn't really the correct one.

Correct; in some borderline cases (as those mentioned), the "PG-opt" SW
will not give the true optimal SW score.  It is a small heuristic that
doesn't affect 99% of cases.

> And as I reported before ssearch also complains when it aligns those
> two sequences with their default scores

I haven't been able to reproduce that effect; what version of ssearch are
you using (what does the version string say?)

> I guess ssearch34 is really a linear space algorithm but it probably
> used quite a lot of memory in something else.

Among other things, the scoring matrix gets turned into a profile, with
length equal to that of the query sequence; this allows for more efficient
memory access as we "walk" down the profile.  There are other odds and
ends that add up to the extra memory you saw.

-Aaron

-- 
 Aaron J Mackey
 Pearson Laboratory
 University of Virginia
 (434) 924-2821
 amackey at virginia.edu