Identification of Differentially Expressed Genes and Gene Sets Using a Modified Q-Value

Abstract

Gene expression technologies allow expression levels to be compared across treatments for thousands of genes simultaneously. Statistical methods exist for identifying differentially expressed (DE) genes and gene sets while controlling multiple testing error. Most methods do not take into account the distribution of effect sizes or the overrepresentation of observed patterns. This paper compares a recently proposed modified q-value method that takes into account such patterns to a traditional q-value method for experiments with three treatments. The results of simulation studies performed suggest that the proposed methods improve upon the traditional method in the identification of DE genes in certain settings, but are outperformed by the traditional method in other settings. Analysis of data sets from real microarray.