We used stringent calling criteria (minimum 10 consecutive probes) to identify 51,997 putative CNV segments in the 41 samples (40 test samples and 1 reference sample). The median numbers of segments in CEU and YRI individuals were 1,117 and 1,488, respectively, reflecting both the higher genetic diversity in Africa and the use of a CEU reference sample. CNV sizes ranged from 443 bp to 1.28 megabases (Mb), with a median size of 2.9 kb. We merged these calls across samples to identify 11,700 putative CNV loci (median size of 2.7 kb), of which 49% were called in a single individual (Supplementary Methods and Supplementary Table 1). Using quantitative PCR (qPCR) for initial validation, we confirmed 79 of 99 randomly selected loci as varying in copy number, suggesting a preliminary false-discovery rate of ~20% (Supplementary Methods).
