Targeted design strategy for choosing BACs for array CGH on the basis of genomic architecture. We hypothesized that the presence of large, highly homologous segmental duplications predisposes certain regions of the genome to microdeletion and/or microduplication by nonallelic homologous recombination (Bailey et al. 2002). Termed “rearrangement hotspots,” these are defined by the presence of flanking intrachromosomal duplications >10 kb in length with >95% similarity and separated by 50 kb to 10 Mb of intervening sequence. For each region, we selected BACs that were contained entirely within each rearrangement hotspot, BACs that overlapped the segmental duplications, and BACs in the peripheral unique sequence for microarray manufacture. The SD microarray comprised a total of 2,194 BACs.

Observations of minor alleles in CNP regions. The 47 samples were categorized as American Indian, sub-Saharan African (Biaka and Mbuti), white (Czech and Druze), and Asian (Chinese and Japanese), and the 119 CNPs identified were plotted (see table A2 in appendix A [online only]). In cases in which the reference sample represented the rare allele (i.e., the majority of individuals showed a change in fluorescence intensity ratio), we classified the CNPs as minor-allele variants. Some previously identified CNPs are indicated by numerals above the bars: 1 = CHRNA7/CHRFAM7A at 15q13.3 (Riley et al. 2002); 2 = IGVH/SLC6A8/CDM pseudogene cluster at 16p11.2 (Barber et al. 1999); 3 = CCL3-L1/CCL4-L1 at 17q12 (Townson et al. 2002); 4 = IGHG1 gene cluster at 14q32.33 (Sasso et al. 1995); 5 = IGL gene cluster at 22q11.22 (van der Burg et al. 2002); 6 = NF1/IGVH/GABRA5 pseudogene amplification at 15q11.2-q13 (Fantes et al. 2002); and 7 = β-defensin gene cluster at 8p23.1 (Hollox et al. 2003).

CNPs detected by array CGH, and verification by FISH. A, Genomic profile of GM10473A (a Biaka female). After quality filtering, 1,967 BAC clones were ordered sequentially from 1pter to Yqter on the basis of physical location in the July 2003 genome assembly (X-axis). Each data point represents the mean log₂ ratio of test:reference intensity of the three replicate spots of each BAC from a single hybridization experiment (Y-axis). Each hybridization was repeated with reverse labeling (dye swap), and clones that yielded log₂ ratios deviating >2 SDs from the mean of all autosomal clones in both experiments were classified as variant. The reference DNA is male; thus, clones located on the sex chromosomes show log₂ ratios consistent with female:male hybridization. Three previously reported CNPs are circled: the β-defensin gene cluster at 8p23.1 (Hollox et al. 2003) (solid circle), the IGHG1 gene cluster at 14q32.33 (Sasso et al. 1995) (dashed circle), and the IGVH/SLC6A8/CDM pseudogene cluster at 16p11.2 (Barber et al. 1999) (dotted circle). A novel variant locus, RP11-136P13 (chromosome 10:81097351–81263857 [green circle]), yielded log₂ ratios of 0.44 and 0.48 in replicate hybridizations. B, FISH confirmation of RP11-136P13, which is composed entirely of segmental-duplication sequence and therefore shows dual signals when used as a FISH probe (green). However, the presence of an additional signal in interphase nuclei shows the polymorphic nature of this locus in GM10473A. A control probe (red) confirms that the cells are diploid. This duplication was also observed in two other subjects. C, Array CGH profile of chromosome 8 in GM10493 (a Biaka). Clones are ordered by physical distance from 8pter in kb (X-axis), with error bars showing the SD of the log₂ ratios from the three replicate spots. Two adjacent clones, RP11-159F11 and RP11-46M15 (red circle), yielded log₂ ratios of 0.41, indicating the presence of a duplication. A BAC that yielded a log₂ ratio of −0.81 in this experiment (dashed circle) was not confirmed in the replicate dye-swapped hybridization; thus, it was classified as nonvariant. D, Use of RP11-159F11 as a FISH probe, showing increased signal intensity on a chromosome 8 homologue in metaphase cells, which resolves to dual signals in interphase cells, confirming the presence of a duplication of this region. An overlapping but nonidentical duplication of 8p23.2 was also observed in GM17051.

CNPs detected on chromosome 15. For each of the 47 individuals studied, relative duplications (green) and deletions (red) are represented, with invariant BACs shown in gray. The different sizes of the deletions and duplications in the three patients with known copy-number changes in the 15q11-q13 region (Locke et al. 2004) are clearly visible. Note that, as expected, none of the 47 normal individuals (94 chromosomes) in our population survey showed variation over these regions associated with disease. Also shown are the locations of segmental duplications >10 kb in size with >95% identity (pink) and of rearrangement hotspots represented on the array (blue), which illustrates the targeted nature of our array. Each tick mark represents 5 Mb, with gaps in the sequence assembly represented by gray bars.

Genomewide map of sites of copy-number variation detected by array CGH. The location of CNPs are indicated by colored bars on each chromosome. Relative deletions identified in a single individual are shown in orange, and deletions observed in multiple individuals are shown in red. Relative duplications identified in a single individual are shown in green, and duplications observed in multiple individuals are shown in blue. CNPs that showed both relative deletion and relative duplication in different individuals are shown in purple (not to scale). Also shown are the locations of segmental duplications >10 kb in size with >95% identity (pink bars above each chromosome). Each tick mark represents 20 Mb, with gaps in the sequence assembly represented by gray bars.