Fig. 4
DeGAs applied to the protein-truncating variants (PTVs) dataset. a, b Phenotype (a) and gene (b) contribution scores for the top key components associated with BMI based on phenotype grouping (Methods, Supplementary Table 2). c Variant PCA plot with biplot annotations for the top two components (Methods). The identified targets for functional follow-up (main text) are marked as (I) rs114285050 (a stop-gain variant on GPR151) and (II) rs150090666 (PDE3B). d Phenome-wide association analysis for GPR151 rs114285050. The p-values (left) and log odds ratio (OR) (binary phenotypes, shown as red) or beta (quantitative phenotypes, shown as blue) (right) along with 95% confidence interval are shown for the phenotypes with minimum case count of 1000 (binary phenotypes) or 1000 individuals with non-missing values (quantitative phenotypes) and strong association (p < 0.001) and with this variants among all the phenotypes used in the study (n = 337,199 White British individuals in the UK Biobank for binary traits and n > 331,000 for each quantitative trait, Supplementary Table 3). L: left, R: right, %: percentage, pred: predicted. Source data are provided in Supplementary Data 2 (a, b), Source Data file (c), and Supplementary Table 3 (d)
