Multiple Sequence Alignment using Hybrid Parallel Computing

Abstract

Multiple sequence alignment (MSA) is critical in several areas of science, especially in bioinformatics. Expressive advances have been developed in MSA and many methods, algorithms and tools have been proposed for it. Since the MSA is an NP-hard problem, efforts have led to the emergence of heuristics to solve it. More recently, heuristics based on progressive alignment have highlighted due to the quality of the alignment and relatively good performance. Despite significant advances, MSA remains a time-consuming task and parallel solutions have been investigated. We propose a novel algorithm for solving MSA based on progressive alignment using cluster of GPUs. Our experimental results showed encouraging speedups for instances containing sequences ranging in length between 60 and 10k.