A Copula-based approach to differential gene expression analysis

Abstract

Microarray technology has revolutionized genomic studies by enabling the study of differential expression of thousands of genes simultaneously. The main objective in microarray experiments is to identify a panel of genes that are associated with a disease outcome or trait. In this thesis, we develop and evaluate a semi-parametric copula-based algorithm for gene selection that does not depend on the distributions of the covariates, except that their marginal distributions are continuous. A comparison of the developed method with the existing methods is done based on power to identify differentially expressed genes (DEGs) and control of Type I error rate via a simulation study. Simulations indicate that the copula-based model has a reasonable power in selecting differentially expressed gene and has a good control of Type I error rate. These results are validated in a publicly available melanoma dataset. The copula-based approach turns out to be useful in finding genes that are clinically important. Relaxing parametric assumptions on microarray data may yield procedures that have good power for differential gene expression analysis.