Figure 1

Workflow schematic. GWAS summary statistics and correlations between genotypes were used as input data. Each set of SNPs (all, non-coding, exonic and nonsynonymous) was analyzed separately. The first step of our study is the gene-based association analyses performed using the SKAT-O, PCA, ACAT-V methods and their results in combination. Next, to guard against the influence of strong GWAS signals outside the gene, we performed polygene pruning for genes with a p-value < 2.5 × 10^–5 obtained at least by one of the gene-based methods. After pruning, we repeated the gene-based analysis using the remaining SNPs. The genes identified using each set of SNPs were subdivided into known and new. All known genes confirmed in our study were compared with the rest of the known genes for their functional properties. We also compared the new and known genes identified in our study. For all identified genes, we performed an enrichment analysis; for the genes identified using protein-coding SNPs, a functional analysis of protein-coding variants was performed; for the genes identified using non-coding SNPs, SMR/HEIDI analysis was performed. Bold blue arrows point to the gene groups being compared and gene groups being under in silico analyses.