Systematic assessment of copy-number variant detection via genome-wide single nucleotide polymorphism genotyping

Our strategy for genotyping polymorphic deletion variants within populations of moderate size (~100-1000 samples) is implemented by two algorithms: SCIMM, a clustering algorithm which derives genotypes from a set of copy-number-informative probes, and SCIMM-Search, a search algorithm which determines a set of copy-number informative probes for each annotated deletion variant. The central idea is to automatically estimate the location and extent of the 6 clusters that appear in intensity data for SNP probes specific to the site of a common deletion: specifically, the three diploid clusters (‘AA’, ‘AB’, and ‘BB’), the two hemizygous clusters (‘A -’ and ‘B -’), and the null (homozygous deletion) cluster (‘-’; see Figure 4 in the main text). Estimated locations are used to assign a copy-number label of ‘null’, ‘haploid’ or ‘diploid’ to each sample.