Whole-exome imputation within UK Biobank powers rare coding variant association and fine-mapping analyses

Abstract

Exome association studies to date have generally been underpowered to systematically evaluate the phenotypic impact of very rare coding variants. We leveraged extensive haplotype sharing between 49,960 exome-sequenced UK Biobank participants and the remainder of the cohort (total N~500K) to impute exome-wide variants at high accuracy (R2>0.5) down to minor allele frequency (MAF) ~0.00005. Association and fine-mapping analyses of 54 quantitative traits identified 1,189 significant associations (P<5 x 10-8) involving 675 distinct rare protein-altering variants (MAF<0.01) that passed stringent filters for likely causality; 600 of the 675 variants (89%) were not present in the NHGRI-EBI GWAS Catalog. We replicated the effect directions of 28 of 28 height-associated variants genotyped in previous exome array studies, including missense variants in newly-associated collagen genes COL16A1 and COL11A2. Across all traits, 49% of associations (578/1,189) occurred in genes with two or more hits; follow-up analyses of these genes identified long allelic series containing up to 45 distinct likely-causal variants within the same gene (on average exhibiting 93%-concordant effect directions). In particular, 24 rare coding variants in IFRD2 independently associated with reticulocyte indices, suggesting an important role of IFRD2 in red blood cell development, and 11 rare coding variants in NPR2 (a gene previously implicated in Mendelian skeletal disorders) exhibited intermediate-to-strong effects on height (0.18-1.09 s.d.). Our results demonstrate the utility of within-cohort imputation in population-scale GWAS cohorts, provide a catalog of likely-causal, large-effect coding variant associations, and foreshadow the insights that will be revealed as genetic biobank studies continue to grow.

Efficient identification of trait-associated loss-of-function variants in the UK Biobank cohort by exome-sequencing based genotype imputation

“…During the time of the preparation of this manuscript, Barton et al published a similar UKB exome-sequencing imputation study [24]. The major difference between the two studies is the use of first tranche of ~50,000 WES participants as reference panel in that study versus the use of all ~200,000 WES participants in the present study.…”

Protein-coding repeat polymorphisms strongly shape diverse human phenotypes

“…Phenotypes in the latter set were derived from physical measurements and touchscreen interviews; blood count, lipid and biomarker panels of biological samples; and follow-up online questionnaires. For continuous traits, we performed quality control and normalization (outlier removal, covariate adjustment, and inverse normal transformation) as previously described 16,56 .…”

“…We further analyzed exome-sequencing read alignments and genotype calls available from whole-exome sequencing (WES) of 49,960 participants 14 (which achieved >20x coverage by 76bp paired-end reads for an average of 94.6% of targeted sites). We augmented the SNP-imputation data set with 4.9 million (predominantly rare) autosomal variants from the WES genotype call set that we previously imputed into the full cohort 16 .…”

“…We augmented the SNP-imputation data set with 4.9 million (predominantly rare) autosomal variants from the WES genotype call set that we previously imputed into the full cohort 16 .…”

“…This analysis found 180 statistically significant associations (Supplementary Table 2). To determine whether such associations were driven by VNTR length variation, or by other, nearby genetic variation with which VNTR alleles are in linkage disequilibrium (LD), we performed fine-mapping analyses 15 considering nearby genotyped and imputed variants 12,16 as well as the VNTR. Because variation at most VNTRs arises from three or more alleles, VNTR variation was only partially correlated with individual SNPs, enabling analysis to distinguish VNTR from SNP effects in a way that has been challenging for di-allelic variants but recently possible for multi-allelic structural variants such as those at the complement component 4 ( C4A/C4B ) and haptoglobin ( HP ) loci 17–19 .…”