Identification of differentially expressed spatial clusters using humoral response microarray data

The protein microarray is a powerful chip-based technology for profiling hundreds of proteins simultaneously and is being increasingly used. To study humoral response in pancreatic cancers, scientists have developed a two-dimensional liquid separation technique and built a two-dimensional protein microarray. However, identifying regions of differential expression on the protein microarray requires the use of appropriate statistical methods to assess the large amounts of data generated. A permutation-based test is proposed that incorporates spatial information of the two-dimensional antibody microarray. By borrowing strength from neighboring differentially expressed spots, the procedure is able to detect differentially expressed regions with high power while controlling the familywise type I error at 0.05 in simulation studies. The proposed methodology is also applied to a real microarray dataset.