A baseline scenario was defined to reflect that seen in recent studies, as follows. Two normally distributed traits were simulated with explained genetic variances 0.4 and 0.3 and correlation of genetic effects of 0.65. Genotypes from 100,000 independent SNPs were simulated, with minor allele frequencies uniformly distributed on (0.01, 0.5). This reflects current marker panels that directly explain about half the heritability [1]. The proportion of null SNPs was 0.95 or 0.99 [9], with the same SNPs having effects for both traits. Their effect sizes were drawn from the bivariate normal distribution such that the desired variances and covariance were attained. The traits were then generated from the quantitative model in equation 1.
