We next turn to SNP genotyping (Fig. 3). For any given genomic segment containing a SNP, samples with two copies of the locus per diploid genome are expected to have one of the canonical SNP genotypes of AA, AB or BB. For most autosomal SNPs we expect all samples to have two copies, but for those overlapping a CNP other possibilities may be observed13 (Fig. 3a). For such SNPs, we use the information from Canary (above) to restrict initial SNP genotyping to those samples whose CNP genotype (integer copy number) is two (Fig. 3b). Such a partitioning allows for the model of diploid SNP clustering not to be misled by samples that have fewer or extra copies (as might happen if one clustered the raw SNP data shown in Figure 3a).
