Dynamic clustering of time series gene expression

Lingling An, Purdue University

Abstract

Clustering time course gene expression data allows one to explore functional co-regulation of genes via their co-expression. Typically, clustering methods seek to place genes with similar expression patterns or profiles into the same group or cluster. However, since a gene may be involved in more than one biological process at any one time, co-regulated genes may not have visually similar expression patterns. Furthermore, the starting or ending time points for co-regulated genes may differ and the number of co-regulated patterns or biological processes shared by two genes may be unknown. Based on this reasoning, biologically realistic gene clusters gained from gene co-regulation may not be accurately identified using traditional clustering methods. By taking advantage of techniques and theories from signal processing, it is possible to cluster time series gene expression profiles using a dynamic perspective under the assumption that different spectral frequencies characterize different biological processes. The proposed dynamic clustering method provides insight into the dynamic associations among the time-limited co-expressed genes which have not been discovered by other clustering techniques. The novel contributions of this research to the areas of Statistics and Bioinformatics are two-fold and include the concept of time-varying clusters, as well as an approach to differentiate significant, or biologically meaningful, clusters from noisy clusters.

Degree

Ph.D.

Advisors

Doerge, Purdue University.

Subject Area

Genetics|Statistics

COinS