Comparaison intensive des séquences protéiques : indexation des banques et usage des instructions SIMD des microprocesseurs
Abstract
This paper presents a method for speeding up the comparison of two protein banks. It is based on two techniques: bank indexing and fast computation of short alignments using the SIMD instruction set of modern microprocessors. The algorithm follows the seed-based approach in which hits are first exhibited before further processing. Performances are compared with BLASTP and show a speed-up of 4 with comparable sensitivity.
Origin : Files produced by the author(s)