Multiple alignment of protein sequences

Bill Pearson wrp at cyclops.micr.Virginia.EDU
Thu Feb 6 13:29:40 EST 1992


	If you have access to a UNIX machine, you should consider the MSA
program, which should be available from Stephen Altschul at the NCBI/NLM.

	%A D. J. Lipman
	%A S. F. Altschul
	%A J. D. Kececioglu
	%T A tool for multiple sequence alignment
	%J Proc. Natl. Acad. Sci. USA
	%D 1989
	%V 86
	%N 12
	%P 4412-5

This program works well for up to about 5 sequences of 200 residues if they
are > 25% identical.  It does a rigorous alignment.

	I would be very cautious about successive pairwise alignment schemes,
such as the Feng-Doolittle method (Pileup) and Clustal.  Any multiple
alignment method, including MSA, can do well on sequences with very
few gaps.  The reason you need a multiple alignment is to rationally
place gaps based on the information from several sequences. Pairwise
schemes can have serious problems in appropriately weighting large
gaps in multiple alignments.  I am familiar with one Feng-Doolittle
tree in the literature which is very unlikely to be correct, because
most of the topology was defined by gap weights.

Bill Pearson



More information about the Proteins mailing list