With so many SNPs and individuals studied simultaneously, I is unavoidable. While practical phasing procedures will have to allow for some errors, the importance of high quality genotypes cannot be overstated. Based on 50 individuals who were typed twice, for individuals and SNPs that passed our quality control criterion, we put the error rate at around 0.01% or lower, which is consistent with the discrepancies we observed for the trio test. Given proper treatment, isolated genotyping errors which impact SNPs individually do not pose a substantial problem. With II, the best approach is to identify them in advance based on extra information such as probe intensities and SNPs that are specifically designed to capture CNVs, with special attention paid to known locations harbouring such variants. Our study of the deletion on 15q11.2 is one such example. There phasing is performed two ways, with and without SNPs in the deletion region (Supplementary Results). The impact of III is discussed below using the MHC region as an example.
