An Integrative Omics Approach to Identify Sub-Network Biomarker in Type 2 Diabetes Mellitus

摘要

In its most ambitious form, integrated multi-omics data technique aims to discover robust prognostic molecular signatures as disease biomarkers. Although, a massive collection of computational analysis approaches has emerged gradually for identifying genes or pathways (groups of genes) that contribute to diseases and other biological processes; gene expression values and confidence score/strength of interactions are hardly used in scoring/ranking the resulting pathways. We have introduced a simple but efficient approach for identifying subnetwork biomarkers can discriminate diseases versus control stage using readily available topological and network information. The procedure is applied to type 2 diabetes mellitus (T2DM) as a case study. We discover a novel gene signatures biomarker from mRNA expression data in skeletal muscle for T2DM. The resulting biomarker is highly enriched with related T2DM pathways and causal genes. Further, the identified subnetwork biomarker outperformed other well-known biokmakers in classification performance and has consistently high accuracies across different tissues and experiments. Taken together, the proposed approach proves the facility of identifying an accurate biomarker for T2DM disease prognosis due to the inclusion of important topological and network information in scoring the resulting pathways. Moreover, it could be successfully employed in biomarker discovery for further studies that characterized by expression profiles.