Identifying regulated genes through the correlation structure of time dependent microarray data

Martina Muehlbach Bremer, Purdue University

Abstract

Since microarray technology has become widely available, it is possible to study the transcription of thousands of genes simultaneously. Experiments can be conducted in which measurements of transcription levels on the same set of genes are taken repeatedly over time. Often these time course gene transcription experiments aim to understand the behavior of genes in a certain process, such as the cell cycle, or the organism's reaction to injury or disease. The transcription levels of genes are influenced by many factors: genes may be regulated by other genes, as well as by enzyme or protein levels in the cell, or by processes such as DNA methylation. Understanding and describing an organism's entire gene regulatory network is an ambitious goal that is considered here in the context of time dependent microarray data. A method is proposed that uses a state space model to represent a gene regulatory network. An algorithm is developed that estimates the optimal model parameters, as well as the behavior of hidden regulators. Based on the model parameter estimates, a criterion is proposed that describes the degree of regulation of every observed gene. Biological assumptions are incorporated to place restrictions on the model parameters, while mathematical restrictions assure statistical validity of the model. The power of the proposed method to identify regulated genes in time dependent microarray data is investigated via simulations and the algorithm is applied to several real microarray time series data sets. Recommendations are made for a minimum number of time point observations that a microarray experiment should include in order to achieve a desired degree of statistical separation between regulated and unregulated genes.

Degree

Ph.D.

Advisors

Doerge, Purdue University.

Subject Area

Statistics

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS