Investigating Key Genes in Type 2 Diabetes Mellitus via Combining mAP-KL and Mutual Information Network

Background: The molecular mechanism of the type 2 diabetes mellitus (T2DM) remains unclear. Objective: This research aimed to investigate key genes in T2DM via combining mAP-KL and mutual information network (MIN) and give great insights to reveal pathological mechanism underlying this disease. Methods: First of all, the data of gene expression profile of T2DM were recruited and preprocessed; then mAP-KL was implemented to investigate clusters and exemplars in T2DM; in the following, support vector machines (SVM) model was selected to evaluate the classification performance of mAPKL; finally, MIN construction and topological analysis were performed to investigate key genes. Results: A total of 20,541 gene symbols were obtained from expression profile of T2DM. By applying mAP-KL, 12 clusters were identified. From Cluster 1 to Cluster 12, their exemplars were OGT, TTC22, LIMCH1, NENF, ROMO1, RGL2, TCF7L1, KRTAP4-4, POLR2F, KIF22, NDUFB11, and AGL, respectively. The results of evaluation by SVM model indicated that the mAP-KL methodology was feasible and suitable for identifying exemplars of T2DM. Finally, MIN construction and topological analysis indicated that there were four hub genes (degree centrality ≥ 100): TCF7L1 (degree = 104), LIMCH1 (degree = 102), NENF (degree = 101), TTC22 (degree = 101), which might be potentially novel predictive and prognostic markers for T2DM. Conclusion: We predict these hub genes (such as TCF7L1 and LIMCH1) might play key roles during the occurrence and development of T2DM and are potentially novel predictive and prognostic markers for T2DM.