Imputation of exome sequence variants into population- based samples and blood-cell-trait-associated loci in African Americans: NHLBI GO Exome Sequencing Project

Abstract

Researchers have successfully applied exome sequencing to discover causal variants in selected individuals with familial, highly penetrant disorders. We demonstrate the utility of exome sequencing followed by imputation for discovering low-frequency variants associated with complex quantitative traits. We performed exome sequencing in a reference panel of 761 African Americans and then imputed newly discovered variants into a larger sample of more than 13,000 African Americans for association testing with the blood cell traits hemoglobin, hematocrit, white blood count, and platelet count. First, we illustrate the feasibility of our approach by demonstrating genome-wide-significant associations for variants that are not covered by conventional genotyping arrays; for example, one such association is that between higher platelet count and an MPL c.117G>T (p.Lys39Asn) variant encoding a p.Lys39Asn amino acid substitution of the thrombopoietin receptor gene (p = 1.5 × 10(-11)). Second, we identified an association between missense variants of LCT and higher white blood count (p = 4 × 10(-13)). Third, we identified low-frequency coding variants that might account for allelic heterogeneity at several known blood cell-associated loci: MPL c.754T>C (p.Tyr252His) was associated with higher platelet count; CD36 c.975T>G (p.Tyr325(∗)) was associated with lower platelet count; and several missense variants at the α-globin gene locus were associated with lower hemoglobin. By identifying low-frequency missense variants associated with blood cell traits not previously reported by genome-wide association studies, we establish that exome sequencing followed by imputation is a powerful approach to dissecting complex, genetically heterogeneous traits in large population-based studies.

Figure 1

Regional Plots of Genetic Loci Associated with Blood Cell Traits Shown are regional association plots generated with LocusZoom for (A) the chromosome 16 α-globin locus and hemoglobin; (B) the chromosome 2 LCT region and WBC; (C) MPL on chromosome 1 and platelet count; and (D) the CD36 locus on chromosome 7 and platelet count. The color of each single-nucleotide variant (SNV) indicates the level of pairwise linkage disequilibrium (LD) based on r-squared relative to the lead SNV in the region. r-squared values were calculated from 1000 Genomes YRI. SNVs with missing LD information are shown in gray. Included in these plots are either SNVs that were imputed from exome sequence data or any regional SNVs that were genotyped with the Affymetrix 6.0 array.