A P-value of 5.0 × 10−8 was the threshold for GWS in the GWAS. Results were not adjusted for testing in two populations because we tested three distinct a priori hypotheses: SNPs are associated with AD and related traits in AAs; SNPs are associated with AD and related traits in EAs; and associations are evident with the same SNPs meta-analyzed in AAs and EAs. Results were not adjusted for testing two measures of AD, which are highly correlated: the rank order correlation between the symptom count and case–control variables is 1.0 (that is, no affected subject can have a lower symptom count than a control), and the point-biserial correlation is 0.90. In the replication analyses, we tested in each model and population whether the percentage of False Discovery Rate (FDR) q-values that were <0.05 was greater than would be expected by chance.
