The polygenic score was estimated using the first trait in a sample of 4000 unrelated subjects. The score was constructed using P-value thresholds of 0.1 and 0.001 for π0 = 0.95 and 0.99 respectively; these thresholds yielded the highest R 2 and AUC values. The score was then tested for association with the second trait in an independent sample of 4000 subjects. The correlation and mean square error between the score and the second trait were also estimated in the second sample. The association tests were used in equation 6 to estimate the explained genetic variances in the first and second samples in turn, and then the covariance between effects in the two samples, each time keeping other parameters fixed to their simulation values.
