Figure 5

Population distribution and functional characterization of GWAS and LD SNPs. (a) GWAS-identified SNPs (left panel) and LD SNPs (right panel) by population are shown in a Venn diagram. In the population where GWAS (index) SNPs were detected, we selected SNPs in LD (linkage disequilibrium) with index SNPs using a threshold of r² > 0.8. Only index SNPs were analyzed in mixed populations with different ethnicity or in populations, which could not be assigned to one of the four populations: EUR (European), ASN (East Asians), AFR (Africans including African Americans) and AMR (Admixed American). (b) Proportions of index and LD SNPs in the four populations. Asterisks indicate significant differences between EUR and ASN (P = 0.004) in the proportion of index and LD SNPs. (c) Density and box plots for functional IW-scores (K10) for index and LD SNPs. IW-scores (K10) were obtained from ten different scoring systems via IW-Scoring tool: an integrative weighted scoring framework to annotate and prioritize noncoding variations. (d) Distribution of genomic region annotations for index and LD SNPs (INT, intronic variant; NSM, non-synonymous missense variant; SYN, synonymous variant; U3, 3′untranslated region variant; U5, 5′untranslated region variant). (e) Distribution of potential regulatory sequences for index and LD SNPs. (f) Distribution of additional functional annotations for index and LD SNPs. SIFT and PolyPhen prediction scores were used to predict pathogenicity of amino acid substitutions. (d, e) Data were obtained from HaploReg v4; (f) Data were obtained from SNPnexus.