Next, we extended our investigation of mutation mechanisms to identify probable dispersed duplications among the CNVs. The array data themselves do not identify chromosomal location, but polymorphic dispersed duplications can be identified by considering other sources of information. We took five complementary approaches to identify dispersed duplications among our CNVs: (1) precise mapping to inter-chromosomal segmental duplications; (2) evidence for inter-chromosomal mappings from sequence data34; (3) inter-chromosomal linkage disequilibrium; (4) poly-A and target site duplication signatures of retrotransposition; and (5) in silico splicing of CNV discovery data in known transcripts to identify retroposed genes (Supplementary Methods and Supplementary Fig. 1.9). By integrating these different sources of data we identified 75 probable dispersed duplications (Fig. 4 and Supplementary Tables 1.5 and 1.6). We developed PCR assays for four of these and genotyped them across 270 HapMap samples, with complete concordance with the array-based genotypes (Supplementary Notes, Supplementary Table 1.7 and Supplementary Fig. 1.10). These dispersed duplications appear randomly distributed among chromosomes. Some of the dispersed duplications can be confidently ascribed to retrotransposition using the signatures described earlier, but other mechanisms may
