Microarray data sets contain a wealth of information on the gene expression levels for thousands of genes for small number of different conditions called assays. But, the information is hidden by high noise levels, and low signal levels. Data mining techniques are used to extract the information of genes related to the assays. This work proposed a powerful principal component analysis (PCA) based method in extend of PCA approach of Rollins et. al. (2006). The proposed method is able to generate gene signatures that expressed the most differently between two assay groups in a microarray data set.

This work developed and evaluated two new test statistics based on PCA and they were found to be effective as evaluated in different case studies including real and simulated data. The methods proposed in this work were compared to the current method. The proposed method was favor in term of high statistical power and low false discovery rate. Therefore, the PCA based approach is highly recommended for use in gene expressions data analysis.