Exploratory analysis of multiple sequence alignments using phylogenies.

Abstract

The significance of an alignment between two sequences can be determined using well-known techniques but cannot be easily evaluated with multiple alignments due to the computational complexity. Therefore multiple alignment algorithms may produce an alignment between sequences even when they have little homology with other sequences. A program is presented that makes use of a phylogeny to explore the implications of an alignment. Using the phylogeny, branch lengths are inferred and a search is conducted for regions of unusually rapid or slow rates of change given the observed rates in the rest of the sequence. A very rapid rate of change can be due to either poor homology or due to rapid divergence because of selection. Phylogenies are calculated using either the neighbor joining algorithm of Saitou and Nei (Mol. Biol. Evol., 4, 406-425, 1987) or a phylogeny supplied by the user. The program also permits randomization of subsections of the sequences to determine the significance of the multiple alignment for these individual regions. The combination of these two simple methods permits rapid and interactive exploration of multiple sequence alignments.