TREEBLAST, AN ALTERNATIVE VISUALIZATION OF BLAST OUTPUT

National Center for Genome Resources, 1800A Old Pecos Trail, Santa Fe, NM 87505 USA

Blast is a useful and popular tool for searching large databases for sequences similar to, and therefore usually homologous to, a query sequence. The results of a Blast search contain far more information than is overtly conveyed by the standard text output. In particular, although Blast output presents pairwise comparisons of the query sequence to each hit, it does not fully exploit relationships among hits. To mine this hidden information, we have created TreeBlast, a tool that processes Blast results to yield trees showing similarity relationships among sequences. It uses rapid heuristic methods and so is not intended as an attempt at thorough phylogenetic analysis, but can serve as the starting point for more rigorous treatment. These trees are built from distance measures based on multi-sequence alignments assembled from selected pairwise alignments found by Blast. Alignments are selected to target regions of maximum sequence density, favoring longer regions spanned by larger numbers of sequences. These alignments can be viewed, edited, saved, or submitted for further analysis via links to other NCGR tools. To make this interface more useful, trees and alignments can be labeled with specific information, such as taxon name and, where applicable, gene name or function. The goal is to provide heuristic visualization of the neighborhood of homology around a query sequence annotated by both taxonomy and function.