In pilot experiments conducted for this study, we recognized problems with the original PCR assay [Heils et al., 1996], as found by others [Kaiser et al., 2002; Yonan et al., 2006] and designed a new high-throughput assay [Wray et al., 2009]. The genotyping in duplicate of all samples, the genotyping of replicated samples, of MZ twins and of family members, all serve to provide a high level of quality control in our study. We found, in a subsample genotyped by our group [Gillespie et al., 2005] using the original assay, on which much of the literature is based, 16% showed inconsistencies with the new genotyping assay. Without our levels of quality control, it is likely other studies using the original assay will have suffered similar genotyping inaccuracy.
