Fig. 3: Partial dependence plots of PRSs on the association between environmental risk factors and diseases in individuals from two biobanks.

Each panel represents a partial dependence plot of PRS on the association between obesity or smoking and T2D, DL, HTN, or CAD in individuals from UKB (a) and BBJ (b). In each panel, the individual risks of disease associated with an environmental risk factor (the vertical axis) are shown along with the PRS values of a disease (the horizontal axis). The color of each dot represents the density of individuals within that dot according to the color bar at the bottom. We showed (i) calibration coefficients and their P values at the upper left and (ii) Spearman’s correlation coefficients between PRSs and disease risks associated with exposure at the upper right. c Comparison of characteristics (the vertical axis) between groups with high and low disease risks associated with each environmental risk factor (the horizontal axis) in the UKB and BBJ cohorts. In each panel, color of the squares represents the standardized mean differences in characteristics between the two groups. The size of squares corresponds to the P values of a two-tailed t-test for continuous characteristics and a Chi-square test for categorical characteristics between two groups. The size and color scales are provided on the right of the figure. The P values in all the panels were calculated from n = 369,942 samples in UKB. For BBJ, the P values were calculated from down-sampled data based on n = 149,421 independent samples to match the disease prevalence in the UKB. UKB UK Biobank, BBJ BioBank Japan, PRS polygenic risk score, CAD coronary artery disease, T2D type 2 diabetes, DL dyslipidemia, HTN hypertension, Obes obesity, SM smoking.