UdeAlignC: Fast Alignment for the Compression of DNA Reads

Abstract

Referential compression algorithms are one of the main strategies to cope with the exponential growth of DNA data available to scientists. One of the techniques used to build a referential compression is sequence alignment, which in turn requires a lot of computing. In this article we present UdeAlignC, a fast alignment algorithm for the compression of DNA reads. We demonstrate that our algorithm is 2× faster than prominent state of the art tools while optimality is only reduced by 5.6%. We also implement a GPU-accelerated version and show local speedups of up to 12×. Source code available at https://bitbucket.org/BioMRcomp/udealignc