BLOSUM matrices and multiple alignment packages

David J. States states at
Tue Jul 19 11:41:37 EST 1994

Has anyone had much experience using the BLOSUM matrices with any of
the multiple alignment (or for that matter pairwise gapped alignment)
packages such as clustalv, MSA, or treealign?  The experience with
BLAST has been that BLOSUM62 results in a significant sensitivity/
specificity improvement over PAM family of scoring matrices [Henikoff
and Henikoff, (1993) Proteins 17:49-61].  In theory they should do better
for multiple alignments as well.  Especially considering that the
BLOSUM matrices were derived from the BLOCKS multiple alignments.  Is
this born out in practice?  Test might include, for example,
correlation between sequence and structural alignments of distantly
related homologs.  Of course, gap penalties need to be established.
Any packages adapted to the newer matrices being distributed on the

David States
Institute for Biomedical Computing / Washington University in St. Louis

More information about the Comp-bio mailing list