Mining Order-preserving Submatrixes from replicated data model

Speaker: CHUI Chun Kit

Abstract

Most of the previous analytical studies on gene expression datasets assume a data model under which the expression value of a gene in an experimental condition is measured once only. This implies that each entry of the data matrix contains a single expression value. Recently, researches in microarray data analysis have shown that any single microarray output is subject to substantial variability. Replication is well recognized by the biologists as a straightforward approach for improving the quality of inferences made from experimental studies. With repeated experiments, the data can be organized in a matrix with each entry being a set of expression levels of a gene under an experimental condition. Analyzing this new replicated data model brings challanges to conventional data mining algorithms. In this talk, I will talk about the issues of mining the Order-preserving Submatrixes (OPSM) under the replicated data model. I will mainly focus on the computataional issues arised and introdce a bounding technique to efficiently mine the OPSMs under the replicated data model.