Seed-based Genomic Sequence Comparison using a FPGA/FLASH Accelerator

Dominique Lavenier (1) , Liu Xinchun (2) , Gilles Georges (1)

1 SYMBIOSE - Biological systems and models, bioinformatics and sequences
2 ICT - Key Lab of Computer System and Architecture, Institute of Computing Technology [Beijing]

This paper presents a parallel architecture for computing genomic sequence alignments using seed-based algorithms. Originality comes from the simultaneous use of FPGA components and FLASH memories. The FPGA technology brings the computer power while the FLASH memory provides high memory bandwidth able to feed a large array of specific operators. A 64 GBytes FLASH memory connected to a Xilinx Virtex-2 Pro PCI board has been developed and an array of 160 distance-computation operators have been implemented to perform the first step of seed-based alignment algorithms. Compared to the {\sc blast} reference software family, we measured a speed-up of 75 on a real intensive genomic sequence comparison application.

Domaines: Architectures Matérielles [cs.AR]; Bio-informatique [q-bio.QM]; Bio-Informatique, Biologie Systémique [q-bio.QM]

Dominique Lavenier, Liu Xinchun, Gilles Georges. Seed-based Genomic Sequence Comparison using a FPGA/FLASH Accelerator. International IEEE Conference on Field Programmable Technology, Dec 2006, Thailand. http://fpt.selfip.org/fpt06/program.php. ⟨hal-00179994⟩