[BiO BB] Looking for researcher, to assist on blast-like invention
dorjetarap at googlemail.com
Tue Feb 12 09:30:18 EST 2008
"Faster than blast" or even "more accurate than blast" type algorithms have
been around for some time now. Some interesting examples are patternhunter
(commercial) http://www.bioinformaticssolutions.com/products/ph/ and YAST
(opensource); Both claim to be significantly faster and more accurate than
BLAST, unfortunately they are not as popular.
I suspect this is for a few reasons: Blast has been around for a while, and
has gained some confidence in the bioinformatics sector; The reporting of
the statistical significance (e-values) is easy to interpret; And it has a
large genomics database readily available. For an algorithm to replace
blast, it would have to tick a lot of these boxes.
Your approach seems pretty interesting as you mention it is not a heuristic
algorithm, whereas the main approach recently seems to be using the
"spaced-seeds" concept introduced in PatternHunter. Your approach sounds
somewhat similar to the four-russians speedup and any way to speed up the
dynamic programming algorithm, without sacrificing speed would benefit a
number of feilds, not just bioinformatics.
I guess, the first step would be to outline your algorithm into a draft
paper to get a better understanding of your approach.
On 12/02/2008, Mike Marchywka <marchywka at hotmail.com> wrote:
> >> I also use many tricks to reuse computations across various proteins.
> >> For example, the matrix for protein "ABCDE", is identical, at first
> Have you gotten any blast source code? This would be a good thing to
> start with for a number of reasons. But, don't assume that a given
> is either well optimized of naive. Sure, they could have
> code like
> if ( metric
> Helping your favorite cause is as easy as instant messaging. You IM, we
> BBB mailing list
> BBB at bioinformatics.org
More information about the BBB