Extraction of Discriminative Gene Modules between Two-Class DNA Microarray Data and Its Application to Chronic Loneliness Patients' Dataset

Development of techniques for assisting to discover useful biomarkers for disease diagnosis is a crucial issue in the field of Biometrics as well as clinical area. In this paper, we propose a novel method that searches for informative gene sets (called modules) for discriminating between different but similar two disease classes in DNA microarray data. This method enables us to identify discriminatory modules that are not only differential between the classes but also specific within respective classes. The proposed method is applied to a dataset obtained from peripheral bloods of individuals with chronic loneliness, and the biological functions of the extracted discriminative modules are evaluated by the functional enrichment analysis. As a result, we show that the proposed method can identify not only biologically meaningful genes as suggested by earlier studies but also novel findings not shown so far.