The WTCCC study is widely held to have been a success, yet the P–value threshold used by that study is an order of magnitude higher than our estimate. Why were there not many false positives? One reason is that the rationale for their choice of 5 × 10−7 was based on prior odds for a region being associated, rather than a single SNP, and adjustment towards single SNP tests could bring the estimate closer to ours. Another practical aspect is that, for reasons of quality control, the study only followed up regions with multiple SNPs showing evidence of association. A formal combination of evidence from neighboring SNPs could produce stronger P–values exceeding those of the single SNPs. Finally, of the 21 single-SNP associations reported by the study, only three had P–values less than 5 × 10−7 but greater than 8 × 10−8, so that chance may have played a role.
