Multi-omics integration of epigenetics, proteomics, and metabolomics identifies putative drug targets and improves early prediction for diabetes

Diabetes holds significant social importance due to its high incidence rate and the multitude of associated complications. The identification of diabetes biomarkers and the understanding of the intricate biological mechanisms underlying diabetes are crucial for the early diagnosis and treatment of diabetes. In this study, we conducted comprehensive omics profiling of CpGs, plasma proteins, and serum metabolites in NSPT cohort of 3,451 individuals, among whom 293 were patients with diabetes. Global association analysis identified 175 CpGs, 29 proteins, and 93 metabolites significantly linked to diabetes, among which 43 CpGs and 25 metabolites were validated in an independent cohort comprising 532 individuals. MR and mediation analysis identified 20 causal biomarkers and 190 signaling pathways linking biomarkers from different layers. By integrating the cross-omics evidence, we provided a list of putative causal biomarkers of diabetes to serve as a valuable resource for the diabetes community. Cross-omics integration prioritized biomarkers for therapeutic targeting, highlighting COLEC11 as an example of potential targets, whose function was further validated in vitro. The early-prediction model using the prioritized biomarkers improved the AUROC by 27.5% compared to the baseline model using clinical features alone. Our findings provide a comprehensive list of prioritized multi-omics biomarkers and elucidate specific signaling pathways in diabetes, contributing significantly to the selection of therapeutic target and the understanding of diabetes pathophysiology.