Replication is the best way to assess the credibility of individual-SNP associations. In addition, their credibility can be probed using the FDR formula, computed under plausible assumptions about genetic architecture. In our application below, we calculate what we call maxFDR, which is an upper bound for the FDR under certain assumptions (Online Methods). In particular, we assume that the effect-size distribution is a multivariate spike-and-slab distribution in which at least 10% of SNPs are non-null for each trait. Illustrative calculations indicate that a trait’s maxFDR can become high when the GWAS for the trait is low powered while the GWAS for another trait is higher powered (Supplementary Note).
