Mutual information network-based support vector machine for identification of rheumatoid arthritis-related genes

In recent years, the underlying roles of significant genes in the pathogenesis of rheumatoid arthritis (RA) have greatly aroused the interest of clinicians and researchers. In the current study, we aimed to identify the potential biomarkers of RA via mutual information network-based support vector machine (SVM) classifier. Firstly,microarray data E-GEOD-45291 was downloaded from ArrayExpress database. Next, differentially expressed genes (DEGs) identification was implemented. The differential pathways genes (DPGs) which were enriched in differential pathways were then screened through attract method. Subsequently, the mutual information network was constructed on the basis of the DPGs through context likelihood of relatedness (CLR) algorithm. Simultaneously, hub genes of the mutual information network were screened, followed by the extraction of the intersection of hub genes and DEGs (named as inter genes). Finally, the classification and evaluation via SVM with linear kernel was performed. Based on the results, we found that a total of 339 DEGs and 448 DPGs were screened. There were 54 hub genes. Then, 8 inter genes were identified such as mitogen-activated protein kinase kinase 2 (MAP2K2), tumor necrosis factor receptor (TNFR)-associated death domain (TRADD), BH3 interacting domain death agonist (BID) and so on. Accordingly, these 8 genes can classify unknown samples from patients with the highest AUC score of 1.00, MCC score of 1.00, specificity of 1.00, and sensitivity of 1.00. Our analysis might provide a novel insight into the pathogenic processes in RA. MAP2K2, TRADD and BID might be attractive biomarkers for the diagnosis and therapeutic intervention of RA.