Statistical tests for identification of differentially expressed genes in cDNA microarray experiments

Abstract

Microarrays experiments are becoming a common laboratory tool for monitoring expression level in cells for thousand of genes simultaneously. The new data promise to enhance fundamental understanding of life on a molecular level and may prove useful in medical diagnosis, treatment and drug design. The greatest challenge to array technology lies in the analysis of gene expression data to identify which genes are differentially expressed across tissue samples or experimental conditions. A simple fold change was used to test the differential expression of genes. Ordinary t-test and t-test approaches with minor variations are usually used in finding differentially expressed genes under two conditions. Analysis of variance (ANOVA) and mixed model ANOVA proved to be powerful under multiple conditions or several sources of variation. Since thousands of hypotheses are tested simultaneously there is increased chance of false positives and it becomes necessary to adjust for multiple testing when assessing statistical significance of findings. Bayesian variable selection and empirical Bayesian approaches offer yet another avenue.