Molecular Classification Of Glioma And Development Of Disease-disease Association Database Based On Differential Coexpression Analysis

Differential coexpression analysis explores the gene pairs whose expression correlation is significantly different between two states (such as disease and normal, treatment group and control group). Due to the gene expression correlation suggesting the existence of some regulation relationships, changes of coexpression relationship or differential coexpression itself indicates changes of regulation relationships between different phenotypes. By analyzing differential coexpression of public glioma transcriptome data, we identify three transcription factors related to the prognosis of glioma (AHR, NFIL3 and ZNF423), and then we use them to classify the glioma into three different molecular subtypes (ZG, NG and IG subtypes), whose survival patterns are distinctively different. Compared with the old classification system, we provide a new system with less biomarkers. At the same time, we explore the functions of these three genes and find that they are enriched with drug targets and that the regulation relationships related to them play an important role in the growth of the glioma. In this research, we not only build molecular classification system of glioma, but also study the disease-disease association with the method of differential coexpression analysis, and develop the disease association database. Currently there is no database which starts from pure data to predict diseases’ relationship. In this study, we make the differential coexpression analysis of expression profile data of human diseases from the public database and find 1326 remarkably similar disease pairs, with which a disease association database is built. Disease Association Database is the first database that measures the relationship between diseases through the similarity of their regulation relationships, which is significant for the clinicians and the science researchers who study diseases.