University of Houston
Department of Computer Science

In partial fulfillment of the Requirements for the Degree of
Master of Science


Lulu Shi
will defend her thesis

Using Linear Correlation as a Discriminative Tool for Gene Expression Data Analysis

Abstract

Discriminative analysis of gene expression data is typically achieved on the single gene level using statistic tests such as t-test. While effective in determining the change of an individual gene’s behavior, these methods fail to capture alteration of gene-gene interaction. Cluster analysis offers a good means of studying gene-gene interaction, but it is not designed for discriminative analysis.

We propose the method of using pair-wise correlation to conduct discriminative analysis of gene expression data. Gene pairs that demonstrate high level of correlation under one condition but low level of correlation under another condition are identified. A permutation-based criterion is established to determine the proper correlation threshold. A handy browser is created to visualize the “Correlation Alternation Networks” (CAN) of genes - networks composed of genes that show differential behaviors in correlation under different conditions. Correlation-based analysis offers a valid tool that can reveal information for discrimination between different conditions. This analysis may be useful for better understanding of gene-gene interaction

Date: Wednesday, November 19, 2003
Time: 12:00 PM
Place: 550-PGH



Faculty, students, and the general public are invited.
Thesis Advisor: Dr. Yuriy Fofanov