Medical relevance of protein-truncating variants across 337,205 individuals in the UK Biobank study

Abstract

Protein-truncating variants can have profound effects on gene function and are critical for clinical genome interpretation and generating therapeutic hypotheses, but their relevance to medical phenotypes has not been systematically assessed. Here, we characterize the effect of 18,228 protein-truncating variants across 135 phenotypes from the UK Biobank and find 27 associations between medical phenotypes and protein-truncating variants in genes outside the major histocompatibility complex. We perform phenome-wide analyses and directly measure the effect in homozygous carriers, commonly referred to as "human knockouts," across medical phenotypes for genes implicated as being protective against disease or associated with at least one phenotype in our study. We find several genes with strong pleiotropic or non-additive effects. Our results illustrate the importance of protein-truncating variants in a variety of diseases.

Fig. 1

Schematic overview of the study. We prepared a data set of 18,228 protein-truncating variants and 135 medical phenotypes from the UK Biobank data set of 337,205 individuals. From these data, we analyzed the clinical effects of predicted protein-truncating genetic variants

Fig. 2

Identification of risk and protective alleles for 135 phenotypes. a–c Manhattan plots for logistic regression for all PTVs and all phenotypes stratified by minor allele frequency a > 1%, b between 0.1% and 1%, and c between 0.01% and 0.1%. Scatter points are colored according to phenotype. Fourteen associations with -log₁₀ p values> 20 were plotted at 20. PTVs in genes near or in the MHC region have smaller scatter points. d Effect size “cascade plot” for all associations outside the MHC with BY-adjusted p < 0.05. Error bars represent 95% confidence intervals. e–f Manhattan plots for PTVs in or near the MHC with minor allele frequency e > 1% and f between 0.1 and 1%. The p values for gray points are the same as in a and b, respectively. The p values for the color points have been re-calculated conditional on HLA alleles

Fig. 3

PheWAS for IFIH1. Phenome-wide associations (logistic regression, p < 0.01) for four PTVs in IFIH1 with minor allele frequency > 0.01%. The left panel shows the number of cases per phenotype in thousands. The middle panel shows the logistic regression −log₁₀ p values. The right panel shows the estimated odds ratios and 95% confidence intervals

Fig. 4

Non-additive associations for FUT2. Association results under additive and non-additive logistic regression models for predicted FUT2 heterozygous or homozygous knockouts (KOs) with a difference between non-additive model AIC and additive model AIC < −1. The left panel shows the number of cases per phenotype. The middle-left panel shows the -log₁₀ p value for the KO association analysis. The middle-right panel shows the estimated log odds ratios and 95% confidence intervals under an additive model (orange) and under a non-additive model for heterozygote KOs (blue) and homozygote KOs (green)