Using the results from the SNP‐based meta‐analysis, we computed R 2 (the procedure is described in Supplementary Methods II). To give an indication of how the resulting effect size estimates impacted power, we conducted post‐hoc power analyses for the meta‐analysis. The analysis was conducted based on the observed effect sizes as a function of the minimum and maximum sample size. We used the compromise power analysis option from the G*power package for the F test family with a single predictor. 45
