the SNP equals to either 20% or 80% of the PVE in the first trait. We considered effect sizes for the two traits to be either in the same direction or in the opposite directions, and we added the simulated effects back to the original phenotypes to form the new simulated phenotypes. For each pre-specified PVE (ranged from 2% to 20% in HMDP and 0.04% to 0.4% in NFBC1966), we simulated 10,000 sets of phenotypes, one for each causal SNP, and calculated the p value for each SNP-phenotype pair. We calculated statistical power as the proportion of p values exceeding the genome-wide significance level at the conventional 0.05 level after Bonferroni correction (p=4.6×10−7 for HDMP and p=1.6×10−7 for NFBC1966). Notice that we simulated phenotypes based on HDL and TG in both data sets, and the two phenotypes are positively correlated in HMDP but negatively correlated in NFBC1966.
