[BiO BB] More Smith-Waterman internals. Gaps extensions this time.
Theodore H. Smith
delete at elfdata.com
Sun Feb 19 16:54:53 EST 2006
Is there anyone out there with enough knowledge of the arcane field
of Smith-Waterman matrixes to answer this?
OK, so Smith-Waterman is best done using a gap penalty that can has a
start cost, and an extension cost.
I want to implement a Smith-Waterman algorithm.
Must I maintain a "gap or not" matrix alongside the cell score matrix?
I'd rather not if it were possible and not wasting CPU time. Because
maintaining and processing two matrixes, effectively makes my
algorithm run at 2x as slow as it would otherwise!
If there is a nice simple trick to avoid needing two matrixes, that
would be great. If not, well at least I know.
The answer could really be something obvious like "well I can't see
any way to avoid needing two matrixes, it looks pretty much like
you'll have to use two" :)
Maybe someone better than me with this stuff has a good answer.
More information about the BBB