[Bioperl-l] What is MPsrch?

William R.Pearson wrp at alpha0.bioch.virginia.edu
Thu Apr 17 09:49:53 EDT 2003


> Thanks for your reply. How does it compare to ssearch? ssearch is using
> Phil Green's algorithm which is a SW algorithm with a little heuristic.
> How "true" is the SW in MPsrch?

I think it is unfair to characterize Phil Green's optimization of 
Smith-Waterman
as heuristic.  With his optimization, SSEARCH is guaranteed to 
calculate the optimal score for any alignment that has a score greater 
than the cost of a one residue gap.
For sequence database searches with sensible gap penalties, this has no 
effect
on sensitivity.

The FASTA package also includes "OSEARCH", which is about 50% slower, 
and provides
a full Smith-Waterman.

The current version of the FASTA package (fasta34t21b2, soon to be 
fasta34t21b3)
includes an Altivec optimized version of SSEARCH (Aaron Mackey's work) 
that is
about 50% slower than FASTA ktup=2, and is faster than FASTA ktup=1.

Bill Pearson



More information about the Bioperl-l mailing list