Accelerating Genomic Sequence Analysis with Graphics Processors

PhD Thesis Proposal Defence


Title: "Accelerating Genomic Sequence Analysis with Graphics Processors"

by

Mr. Mian LU


ABSTRACT:

Sequence alignment and Single-Nucleotide Polymorphism (SNP) detection are 
two fundamental tasks in DNA sequence analysis. Sequence alignment, in 
particular, short read alignment, matches DNA fragments generated from 
second-generation sequencers to a reference sequence. Subsequently, 
through SNP detection, the variation on a single nucleotide is identified 
between each aligned read and the reference sequence. As these analysis 
tasks handle millions to billions of base pairs of gene data and perform 
intensive computation, we propose to improve the analysis speed by (1) 
improving the IO, memory access, and computation of each task; and (2) 
tightly integrating the two tasks to reduce redundancy. In particular, we 
explore the use of graphics processors, or the GPU, for both tasks. 
Specifically, we propose a filtering-verification algorithm to utilize the 
GPU's massive parallel processing power and high memory bandwidth in short 
read alignment; we design a sparse data representation format to improve 
memory access and reduce branch divergence on the GPU in SNP detection; we 
propose a partition-based storage layout and GPU-based, customized 
compression techniques for alignment results to reduce the IO cost and to 
improve the overall speed of the two tasks. Our initial results show that 
our proposed approach accelerates state-of-the-art tools by an order of 
magnitude.


Date:                   Wednesday, 21 March 2012

Time:                   2:30pm - 4:30pm

Venue:                  Room 3311
                         lifts 17/18

Committee Members:      Dr. Qiong Luo (Supervisor)
                         Prof. Qiang Yang (Chairperson)
 			Prof. Dik-Lun Lee
 			Prof. Frederick Lochovsky


**** ALL are Welcome ****