Rapid parallel genome indexing with MapReduce

  • Authors:

  • Rohith K. Menon;Goutham P. Bhat;Michael C. Schatz

  • Affiliations:

  • Stony Brook University, Stony Brook, NY, USA;Stony Brook University, Stony Brook, NY, USA;Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, USA

  • Venue:
  • Proceedings of the second international workshop on MapReduce and its applications
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Sequence alignment is one of the most important applications in computational biology, and is used for such diverse tasks as identifying homologous proteins, analyzing gene expression, mapping variations between individuals, or assembling de novo the genome of organism. Except for trivial cases involving just a small number of short sequences, virtually all other sequence alignment tasks rely on a precomputed index of the sequence to accelerate the alignment. Two of the most important index structures are the suffix array, which consists of the lexicographically sorted list of suffixes of a genome, and the closely related Burrows-Wheeler Transform (BWT), which is a permutation of the genome based on the suffix array. Constructing these structures on large sequences, such as the human genome, requires several hours of serial computation and must be performed for each genome, or genome assembly, to be analyzed. Here we present a novel parallel algorithm for constructing the suffix array and the BWT of a sequence leveraging the unique features of the MapReduce parallel programming model. We demonstrate the performance of the algorithm by greatly accelerating suffix array and BWT construction on five significant genomes using as many as 120 cores leased from the Amazon Elastic Compute Cloud (EC2), reducing the end-to-end runtime from hours to mere minutes. The source code is available under an open source GPL License at: http://code.google.com/p/genome-indexing/