SCHOOL OF COMPUTING,NUS 
 
POSTGRADUATE SEMINAR BY 
 
MS XU XIN
 
Data Mining Techniques in Gene Expression Data Analysis
 
Executive Classroom, SoC-1 level 5
 
1 September 2006, 10.00am

 
Abstract: 
With the rapid development of microarray chip technology, gene
expression data are being generated in huge quantities. The
indispensable task of data mining, as a result, is to effectively and
efficiently extract useful biological information from gene expression
data. However, high-dimensionality and complex relationships among
genes impose great challenges for the existing data mining methods.

In this thesis, we systematically study the existing problems of
state-of-the art data mining algorithms for gene expression data in
class association rule mining, associative classification and subspace
clustering of genes of nonlinear and shiftingand-scaling correlations.
Specifically, we propose the concept of top-k covering rule groups
(TopKRGs) for each gene expression sample, and design a row-wise
mining algorithm to discover TopKRGs efficiently. We further develop a
new associative classifier by combining the nl rules consisting of the
most significant genes based on the entropy test on the top k covering
rule groups. To address the nonlinear correlation problem and the
shifting-and-scaling correlation problem, we introduce our Curler and
RegMiner algorithms respectively to identify the subset of genes which
exhibit non-linear or shifting-and-scaling correlation patterns across
a subset of conditions.

Extensive experimental studies are conducted on synthetic and
real-life datasets. The experimental results show the effectiveness
and efficiency of our algorithms. While we mainly use gene expression
data in our study, our algorithms can also be applied to
high-dimensional data in other domains.