Bioinformatics-led identification of pathophysiological hallmark genes in diabesotension via graph clustering method

Diabesotension, an overlapping triad of diabetes, hypertension, and obesity, remains a diagnostic challenge due to its complex underlying molecular mechanisms. Individuals with diabesotension face twice the risk of microvascular and macrovascular complications compared to those with either condition alone. However, the complexity of diabesotension poses significant diagnostic challenges due to limited knowledge of this disease trifecta.

The protein network was constructed, and the DPClusOST algorithm was applied to determine the protein clusters with a density ranging from 0.1 to 1.0 and those relevant to the pathophysiology of diabesotension. The significance score (SScore) was computed using the p-value from Fisher's exact test to evaluate each cluster, and the clusters containing proteins associated with diabesotension were classified using receiver operating characteristic (ROC) analysis. The significant density of the cluster, as indicated by the AUC, was determined and subsequently subjected to pathway enrichment analysis using ShinyGO.

At densities of 0.6 and 0.8, 14 proteins (STX3, VAMP2, STX4, SYT1, DNAJC5, HSD17B10, DLD, AIFM1, PDHA1, PDHB, DLAT, PDHX, OGDH, and STAT5A) from clusters 13 and 53 were significantly identified as potential diabesotension-related proteins. Key pathways associated with the tripartite interplay of the three pathologies were found to involve amino acid metabolism, glycolysis/gluconeogenesis, SNARE-mediated vesicle transport, insulin and salivary secretion, and the glucagon and HIF-1 signaling pathways, thus identifying novel candidates for diabesotension biomarkers and therapeutic targets.

This study highlights the use of graph clustering to identify potential biomarkers for the comorbid triad, which could enhance personalized future treatment strategies.