We use an exact test for Hardy-Weinberg equilibrium [Haldane 1954] (HWE) performed on unrelated control subjects with relatively homogenous ancestry. In quantile-quantile (QQ) plots, all four projects show deviations of observed from expected p-values at about 0.01 for autosomal SNPs (e.g. Figure S11). It is not clear how many of these deviations are due to genotyping artifacts and how many are due to true genotypic frequency deviations from HWE, but examination of cluster plots indicates that most of the extreme deviations are due to poorly performing SNP assays.
