Statistica Sinica 12(2002), 203-217

STATISTICAL ANALYSIS OF A GENE EXPRESSION

MICROARRAY EXPERIMENT WITH REPLICATION

M. Kathleen Kerr$^1$, Cynthia A. Afshari$^2$, Lee Bennett$^2$,

Pierre Bushel$^2$, Jeanelle Martinez$^2$, Nigel J. Walker$^2$ and Gary A. Churchill$^3$

$^1$University of Washington,

$^2$National Institute of Environmental Health Sciences and

$^3$The Jackson Laboratory

Abstract: Common ratio-based approaches for analyzing gene expression microarray data do not provide a framework for handling replication, although replication is clearly desirable for these noisy data. In contrast, replication fits naturally into analysis of variance (ANOVA) methods. We use ANOVA to analyze data from a microarray experiment to compare gene expression in drug-treated and control cells lines. We discuss issues that commonly arise in the analysis of microarray data, and present practical solutions to some common problems.

Key words and phrases: Analysis of variance, bootstrap, cDNA microarray, gene expression, orthogonal design.