Ordered index seed algorithm for intensive DNA sequence comparison

This paper presents a seed-based algorithm for intensive DNA sequence comparison. The novelty comes from the way seeds are used to efficiently generate small ungapped alignments - or HSPs (high scoring pairs) - in the first stage of the search. W-nt words are first indexed and all the Aw possible seeds are enumerated following a strict order ensuring fast generation of unique HSPs. A prototype - written in C - has been realized and tested on large DNA banks. Speed-up compared to BLASTN range from 5 to 28 with comparable sensitivity.