Computational genomics of type ۲ diabetes: an integrative approach to shed light on genome-wide statistics

Genome-wide association studies (GWAS) have revealed genomic loci for different complex traits; but the biological interpretation of these loci is still elusive. A number of methodologies have been developed to prioritize the most likely causal genes and pathways. Here we follow a post-GWAS pipeline and add an extra step to analyze GWAS results of type ۲ diabetes (T۲D) and to gain insight about underlying biological knowledge. For this goal, after defining independent associated loci, we prioritized the most likely causal genes using a phenotype- and mechanism-agnostic algorithm, which is predicated on a previously formulated assumption that truly associated genes share functional annotations. Co-expression data with annotated gene sets, are used to predict gene function and gene set enrichment analysis. Mendelian Randomization (MR) analysis is performed to integrate GWAS results with expression data and investigate genes which their expression level is likely causal. ۱۵ genes from associated loci were prioritized (FDR < ۰.۲) according to the abovementioned functional predictions. MARK۱ PPI subnetwork (P = ۷.۱۳e-۷) and positive regulation of cell cycle (FDR < ۰.۰۵) were the most significant pathways enriched for genes from associated loci with P < ۵e-۸ and P < ۱e-۵ respectively. MR analysis of gene expression causality showed a robust effect for PLEKHA۱ among eQTL studies; this gene is also present among top ۱۰ genes of MARK۱ PPI subnetwork. PLEKHA۱ is not directly reported for T۲D, but regarding its involvement in macular degeneration and eye disease, this result suggests a role for it, maybe via eye related implications, in diabetes.