The percentage variance explained for RR-interval was calculated using data from all subjects in the 1958BC study. The SNV genotypes were extracted from the 1958BC Exome Chip data and considered in two different sets: the 12 previously reported SNVs covered on the chip including proxies (r2 ≥ 0.8; see Supplementary Material, Table S1); and the lead SNVs from the five unreported novel loci (see Table 1A). First, RR-interval was regressed in a linear model against the sex and BMI covariates (not age, as all 1958BC subjects are of same age). Then the trait residuals from this first model were used as the phenotype in a second linear regression model, with all SNVs in the given set analysed jointly as multiple predictors, and adjusted for the top 10 PCs. The percentage trait variance explained by the set of SNPs was estimated from this second model, according to the adjusted R2 value.
