Selecting Genes for Classification with Microarray Data by Exploiting Third-Order Gene Interactions

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Title: "Selecting Genes for Classification with Microarray Data by 
       Exploiting Third-Order Gene Interactions"

by

Miss LI Xuyang

Abstract:

Microarray technology provides an innovative approach to the collection of
massive gene profiling data from a single experiment. However, it poses a
challenge to data analysts in that a typical microarray dataset usually
involves only several tens of samples but each sample contains information
of up to tens of thousands of genes. This significant disproportion
seriously affects the performance of a classifier of any kind. Therefore,
effective methods for gene selection, or dimensionality reduction, are
needed to filter out a large portion of irrelevant or redundant genes from
the original data.

Among the existing gene selection methods, most pay no or little attention
on higher-order gene interactions, that is, gene interactions within a gene
group of size larger than two. In this project, we propose a new gene
selection method which takes third-order gene interactions into account. The
method searches for informative gene triplets based on higher-order mutual
information measures and constructs new features for each triplet to put
emphasis on gene interactions. An experiment has been conducted to evaluate
our gene selection method and to study the effect of incorporating
higher-order gene interactions on the classification performance.


Date		:  May 15, 2009

Time		:  2pm - 2:45pm

Venue		:  Room 3405

Advisor		:  Prof. D.Y. Yeung

2nd Reader	:  Dr. James Kwok


*************** All are welcome *******************