all other GWAS included in table 1 can be found in Supplementary Figures 8a–w. As with any inference procedure that relies on a model of genetic architecture, it is possible that our results may be biased by model misspecifications other than those that we have simulated directly (e.g., if independent effect sizes are a poor model, perhaps because coupled alleles have a tendency to have effects in the same direction). This may explain the moderate inflation in the LD Score regression intercept that we observe in some large GWAS that are likely well-calibrated. Note that upward bias in the LD Score regression intercept means only that the intercept may be conservative as a correction factor.
