Studying the effect of parallelization on the performance of Andromeda Search Engine: A search engine for peptides

Jigna Shah, Purdue University

Abstract

Human body is made of proteins. The analysis of structure and functions of these proteins reveal important information about human body. An important technique used for protein evaluation is Mass Spectrometry. The protein data generated using mass spectrometer is analyzed for the detection of patterns in proteins. A wide variety of operations are performed on the data obtained from a mass spectrometer namely visualization, spectral deconvolution, peak alignment, normalization, pattern recognition and significance testing. There are a number of software that analyze the huge volume of data generated from a mass spectrometer. An example of such a software is MaxQuant that analyzes high resolution mass spectrometric data. A search engine called Andromeda is integrated into MaxQuant that is used for peptide identification. One major drawback of the Andromeda Search Engine is its execution time. Identification of peptides involves a number of complex operations and intensive data processing. Therefore this research work focuses on implementing parallelization as a way to improve the performance of the Andromeda Search Engine. This is done by partitioning the data and distributing it across various cores and nodes. Also multiple tasks are executed concurrently on multiple nodes and cores. A number of bioinformatics applications have been parallelized with significant improvement in execution time over the serial version. For this research work Task Parallel Library (TPL) and Common Library Runtime (CLR) constructs are used for parallelizing the application. The aim of this research work is to implement these techniques to parallelize the Andromeda Search Engine and gain improvement in the execution time by leveraging multi core architecture.

Degree

M.S.

Advisors

Springer, Purdue University.

Subject Area

Computer science

COinS