Extended Data Fig. 2: Effects of coding variants in LXRα on cardiometabolic traits.
A: The effect of coding variants in LXRα on a panel of pre-selected cardiometabolic traits. Each dot represents the effect of a damaging mutation on the trait in standard deviation, the error bars represent 95% confidence intervals, each derived from a generalized linear model (see methods). Red points indicate statistical significance after adjustment for multiple testing (Bonferroni-corrected P-value – 3.9×10−4), NS = non-significant. B: Scatter plot demonstrating the relationship between the average effect of loss of function mutations and all dominant negative (DN) or protein truncating variants (PTV) on cardiometabolic trait. Each dot plots the effect of the Loss of function mutations on a single trait against either PTVs (blue) or DN (red). The dashed line represents y = x where there would be perfect agreement between the effects of Loss of function mutations and DN/PTV. C: Forest plot of sensitivity analyses. ‘Full’ illustrates the effect of damaging mutations in LXRα (LOF or DN or PTV) from the primary analysis, Drop R415Q – analysis after removing the most common rare variant (R415Q, N(carriers in all UKBB) = 565) or ‘Adjusted common variants’ - adjusting for common variants in the region significantly associated with the listed traits of interest. D: Scatter plot of scaled, centred log normalized ALT values (0 represents mean ALT, units are standard deviation) according to tertile of liver fat polygenic risk score (PRS) and presence/absence of a damaging variant in LXRα. Error bars represent 95% confidence intervals. LP(a) – Lipoprotein-a, LDL – low density lipoprotein, BMI – body mass index, WHR adj BMI – waist hip ratio adjusted for BMI, ALP – alkaline phosphatase, ALT – alanine aminotransferase, GGT – gamma glutamyl transferase, AST – aspartate aminotransferase, ApoA1 – Apolipoprotein A1, HDL – high density lipoprotein.
