Accelerating Genome Assembly Using Hard Embedded Blocks in FPGAs

Sharatchandra Varma Bogaraju, Kolin Paul, M Balakrishnan

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

Modern FPGAs consist of a number of Hard Embedded Blocks (HEBs) like BRAMs, DSPs, PCI controllers, etc. for performing specialized operations/functions. This helps in developing efficient architectures for accelerating compute intensive kernels. Computational genomics is a fairly recent domain which helps in the understanding of the biological properties of an organism with applications in medicine, evolutionary biology and social sciences. A key problem in this research domain is the assembly of DNA sequences obtained from sequencing machines. Next Generation DNA Sequencing (NGS) technologies generate a large number of short fragments (sequences) of fixed length known as reads, which are a part of the genome sequence being assembled. It is computationally very time consuming to assemble these short reads to form the genome sequence purely with software. FPGAs have been effectively used to build accelerators to speed up the computation process. In this paper, we propose HEBs to accelerate the genome assembly process. We show 20% increase in speedups using FPGAs with HEBs over existing FPGA implementation. Speedups of up-to 11x can be obtained over Velvet [1] which is a commonly used software implementation.
Original languageUndefined/Unknown
Title of host publication2014 27th International Conference on VLSI Design and 2014 13th International Conference on Embedded Systems
Pages306-311
Number of pages6
DOIs
Publication statusPublished - 01 Jan 2014

Keywords

  • field programmable gate arrays
  • genomics
  • FPGA
  • Velvet
  • genome assembly acceleration
  • hard embedded blocks
  • speedups
  • Acceleration
  • Assembly
  • Bioinformatics
  • Field programmable gate arrays
  • Genomics
  • Software
  • Vectors
  • FPGA based Acceleration
  • Genome Assembly
  • Hard Embedded Blocks
  • NGS assembly

Cite this