The heritable basis of gene–environment interactions in cardiometabolic traits

Fig. 1. Heat plot showing p values for (a) class 1 and (b) class 2 interactions. Experiment-wise significance threshold is p ≤ 1 × 10−4 (darkest blue in the heat plot). All environmental exposures are continuous variables except for sex, physical activity and smoking, which are dichotomous variables. TC, total cholesterol; TG, triacylglycerol

Fig. 2. Genotype–age interactions: (a) class 1 and (b) class 2 interactions. Dark blue full square, weight; dark blue empty square, BMI; dark blue full triangle, fasting glucose; dark blue empty triangle, 2 h glucose; light blue full square, SBP; light blue empty square, DBP; light blue full triangle, total cholesterol; light blue full circle, LDL-C. Only significant traits are represented in the figure. Experiment-wise significant interactions (p ≤ 1 × 10−4) are marked with an asterisk (SBP for class 1 interactions, and weight for class 2 interactions). α G, γ G and λ parameters were calculated based on individuals 30–60 years of age, as this was the age range in the dataset. A broader age range curve (0–80 years) based on estimates above is displayed in the x-axis to improve the visualisation

Fig. 3. Class 1 genotype–sex interactions for (a) BMI, (b) DBP, (c) LDL-C, (d) SBP and (e) triacylglycerol. Only significant traits are represented in the figure; *p < 0.05, **p < 0.01, ***p < 0.001. Experiment-wise significant class 1 interactions (p ≤ 1 × 10−4) are BMI (p ≤ 4 × 10−6) and triacylglycerol (p ≤ 4 × 10−7)

Fig. 4. Genotype–alcohol intake interactions: (a) class 1 and (b) class 2 interactions. Dark blue full square, weight; dark blue empty square, BMI; dark blue empty triangle, 2 h glucose. Only significant traits are represented in the figure. Experiment-wise significant interactions (p ≤ 1 × 10−4) are marked with an asterisk (weight for class 1 and class 2 interactions)