A two-step procedure for multiple pairwise comparisons in microarray experiments
Abstract
Microarray technology has been widely used in biological and medical studies. Different statistical methodologies have been developed to identify differentially expressed genes under different treatment conditions or among different types of cell samples. In this dissertation, the similarities and differences between two commonly used analysis of variance (ANOVA) models are investigated in spotted microarray experiments. With tens of thousands genes on the array, multiple comparisons have become a challenging problem. A new method is proposed to estimate the proportion of true null hypotheses (π0) so that Benjamini and Hochberg's false discovery rate (FDR) controlling procedure with incorporation of this estimate of π0 controls the FDR below, but extremely close to a pre-chosen level α. A novel two-step multiple comparison procedure is also developed for pairwise comparisons of more than two treatment conditions in microarray experiments. Properties of this procedure are fully investigated for three treatments. Finally, a Bayesian approach is employed to test a group of genes which are differentially expressed between two treatments as a group by taking into account the possible interactions among the 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.