A new (experimental) release of the FASTA program package is now
available from virginia.edu in pub/fasta/fasta20x.shar(.Z). Version
2.0x incorporates several major improvements in the FASTA and SSEARCH
(Smith-Waterman) sequence searching programs:
(1) Explicit statistical estimates are now available from FASTA,
TFASTA, and SSEARCH. Expectation values (a la BLAST) are provided for
library similarity scores. In addition, raw similarity scores are
normalized to correct for the expected length-dependence of similarity
scores. The statistical estimates are very accurate for SSEARCH and
for FASTA if the "-o" option is used. The "-o" option improves
dramatically the performace of FASTA - at a cost in execution time - I
recommend strongly that you use it as often as possible for proteins.
For DNA it serves no purpose.
(2) FASTA now uses the rigorous Smith-Waterman algorithm to produce
alignments. Thus, there is no limit to length size in alignments.
(3) The BLOSUM50 matrix is now used by default.
(4) It is easy to change the penalties for the first residue (now -12
by default) and additional residue (-2) in a a gap.
(5) A new alignment option "-m 4", makes it easy to display the part
of the query sequence that is aligned to the library sequence. For
many protein families, one expects for the alignment to extend from
one end of the sequence to the other. This display makes it easy to
see when the alignments become much shorter.
This version has only been tested for a few weeks, thus the version
number 2.0x. Please let me know if you have any problems with it.
Bill Pearson
wrp at virginia.edu
--
wrp at virginia.EDU
Dept. of Biochemistry #440
U. of Virginia
Charlottesville, VA 22908
