General Departmental Seminar Series
Kmeans Clustering of Gene Expression Time Profiles in Glucose-Stimulated Beta Cells
Mindy Drum, Ph.D., Research Associate (Asst Professor), Department of Health Studies, The University of Chicago
Friday, May 2, 2003, 12-1 pm
6225 Medical Sciences Center, 1300 University Ave.
Microarray technology provides the ability to examine the transcription phase of gene expression for thousands of genes - often an entire genome - simultaneously. The vast quantities of high-dimensional data have prompted the use of traditional and newly-developed clustering techniques for data-reduction and identification of common patterns of expression. Clustering is particularly suitable when applied to expression levels measured over time. I will describe a kmeans cluster analysis to examine the time course of gene expression in pancreatic beta cells following increased glucose concentration. This analysis was conducted as one step in the process of identifying genes involved in signal transduction pathways utilized for glucose regulation. Subsequent annotation of highly expressed genes and those exhibiting specific patterns of expression confirmed known biological results and led to the identification of a novel kinase-encoding gene. Open methodological questions that will be discussed include selection of the number of clusters, starting values, scale of analysis and useful methods of summarizing somewhat cumbersome results.
Back to General Departmental Seminar Series