Design and implementation of a CUDA-compatible GPU-based core for gapped BLAST algorithm

This paper presents the first ever reported implementation of the Gapped Basic Local Alignment Search Tool (Gapped BLAST) for biological sequence alignment, with the Two-Hit method, on CUDA (compute unified device architecture)-compatible Graphic Processing Units (GPUs). The latter have recently emerged as relatively low cost and easy to program high performance platforms for general purpose computing. Our Gapped BLAST implementation on an NVIDIA Geforce 8800 GTX GPU is up to 2.7x quicker than the most optimized CPU-based implementation, namely NCBI BLAST, running on a Pentium4 3.4 GHz desktop computer with 2GB RAM.