Optimizing Reconfigurable Hardware for Genomic Sequences Comparison

Abstract

Last advances in the bioinformatics field have lead to improve models and computational resources to satisfy the high computational cost needed to search and identify DNA sequences. This paper provides a FPGA-based systolic array implementation of the Smith-Waterman algorithm, which improves previous proposals, reducing the amount of hardware and the computing power needed. Besides, optimizations in memory access using prefetching and different sizes of the systolic array let us to increase the system's performance up to 60%.