Title: "Identifying Regulated Genes through the Correlation Structure of Time Dependent Microarray Data"
Speaker: Martina Muehlbach Bremer, Department of Statistics, Purdue University
Place: Krannert (KRAN) G016; October 24, 2006, Tuesday, 4:30pm

Abstract

Recent advances in the area of microarray technology make it possible to study complex biological processes over time. In experiments gene expression data is collected at an ever increasing number of time points. An important goal of these experiments is to study the gene regulatory network. Due to the extreme dimensions of microarray data sets, most statistical methods for the analysis of regulatory networks that have been proposed rely on a significant amount of data reduction prior to analysis. However, the criteria currently used to exclude genes from analysis are not necessarily biologically meaningful.

Based on a state space model for gene regulation, a criterion is formulated that allows one to rank all of the observed genes in a microarray experiment according to their degree of regulation in a biological process. Hidden regulators in the model allow the inclusion of unobserved components, e.g., genes not measured on the array or other cell components such as proteins. The dimension of the state space is estimated using the autocovariances of the observed gene expression values. Applications to real microarray data sets will be presented as well as simulations that demonstrate the performance of the proposed criterion under a variety of circumstances. Partitioning methods allow for efficient computation even in the case of microarray data sets of extreme dimension.




Click here for a full schedule of BIOINFORMATICS SEMINARS, past and present.