Mixture distribution approach for identifying differentially expressed genes in microarray data of Arabidopsis thaliana

Abstract

The basic aim of analyzing gene expression data is to identify genes whose expression patterns differ in the treatment samples, with respect to the control or healthy samples. Microarray technology is a tool for analyzing simultaneous relative expression of thousands of genes within a particular cell population or tissue in a single experiment through the hybridization of RNA. Present paper deals with mixture distribution approach to investigate differentially expressed genes for sequence data of Arabidopsis thaliana under two conditions, salt-stressed and control. Two-component mixture normal model was fitted to the normalized data and the parameters were estimated using EM algorithm. Likelihood Ratio Test (LRT) was performed for testing goodness-of-fit. Fitting of two-component mixture normal model was found to be capable of capturing more variability as compared to single component normal distribution and was able to identify the differentially expressed genes more accurately.