Parallel Prefiltering for Accelerating HHblits on the Convey HC-1

Abstract

HHblits is a bioinformatics application for finding proteins with common ances- tors. To achieve more sensitivitythe protein sequences of the query are not compared directly against the database protein sequencesbut rather their Hidden Markov Models are compared. ThusHHblits is very time-consuming and therefore needs to be accelerated. A multi-FPGA system such as the Convey HC-1 is a promising candiate to achieve acceleration. We present the design and implementation of a parallel coprocessor on the Convey HC-1 to accelerate HHblits after analyzing the application toward acceleration candidates. We achieve a speedup of 117.5× against a sequential implementation for FPGA-suitable data sizes per kernel and negligible speedup for the entire uniprot20 protein database against an optimized SSE implementation.