Copy-Number Variation (CNV) Analysis of Twin Pair 291/292
We applied 32K BAC array (A, D, F, and G), Illumina HumanHap 300 Duo beadchip (B, C, and E), and high-resolution array based on repeat-free nonredundant PCR fragments (H and I).
(A) and (D) illustrate combined array-CGH profiles of chromosomes 11 and 4, respectively, which were derived from two dye-swap experiments on 32K BAC array for twin 292 versus 291. Relative fluorescent intensity was plotted as a straight ratio. The data points shown were scored in both experiments and do not deviate more than one global standard deviation (SD) from each other. The global mean and one standard deviation for these two combined experiments are 1.003 and 0.039, respectively. Two large aberrations on chromosomes 11 (∼22 Mb) and 4 (∼85 Mb), present in a minority of blood cells from twin 292, are highlighted by white areas. The red line shows the moving average with a period of 30 data points.
(B) shows the values of SNP allele frequencies, obtained by the Illumina beadchip, in twin 292 on chromosome 11. The region corresponding to the 22 Mb deletion is highlighted in the white box. Heterozygous SNPs are distributed around a value of 0.5.
(C) and (E) plot the values of absolute difference (Abs. Diff.) between the heterozygous SNP allele frequencies in twin 292 versus 291, as measured by the Illumina beadchip for chromosomes 11 and 4, respectively. The solid and the dotted orange lines represent the average and the standard deviation of the absolute difference, respectively. The larger the absolute difference between the allele frequencies in two tested individuals, the larger the variance in copy number. The red line shows the moving average with a period of 30 data points. The data from (A), (C), (D), and (E) were used to compute the statistical significance of the observed aberrations (Figure 4).
(F) displays the relative fraction of BACs per chromosome that deviate more than two SDs from the mean, point in the same direction (i.e., show deletion or gain), and are located less than 1 Mb from each other. The two combined 32K array experiments used for these calculations are displayed in (A) and (D). A higher relative score was indicative of a larger region of deviation for the particular chromosome, as seen for chromosomes 11 and 4. Although the 11q deletion was approximately four times smaller than the deletion on chromosome 4, it was present in a higher number of cells and thus contains higher number of BACs deviating by more than 2 SDs.
(G) displays the summary of five 32K array experiments between each of twins 291 and 292 against healthy female control (F1), displaying 155 clones within the 100–127 Mb interval of 11q. The blue dots represent the average intensity ratios of data points scored in at least two 292-versus-F1 hybridizations. Red dots show the average value of two 291-versus-F1 experiments. The mean and SD shown were calculated with data points from the 102–124 Mb interval for the respective hybridizations. These experiments show that the CNV detected on 11q represents a deletion in twin 292.
(H) Array-CGH analysis of DNA from twin 292 versus 291 on the confirmatory PCR-based array covering the four shaded clones. The mean and SD for the six chromosome 11 data points are 0.899 and 0.022, respectively. Corresponding numbers for the nonchromosome 11 data points are 1.006 and 0.018.
(I) Summary of two array-CGH hybridizations of twin 291 against F1 (red) and twin 292 against F1 (blue) on the same array shown in (E). The average intensity ratio for autosomal PCR fragments for twin 291 is 1.000, SD = 0.031. The average intensity ratio for nonchromosome 11 autosomal fragments for twin 292 is 1.011, SD = 0.021. The average ratio for the chromosome 11 clones for 292 is 0.907, SD = 0.039.

CNV Analysis of Twin D8 Showing the 1.6 Mb Deletion on Chromosome 2
(A) Profile of the entire chromosome 2 from Illumina HumanHap 300 Duo beadchip showing the values of SNP allele ratios. True heterozygous SNPs are expected to be distributed around a value of 0.5. In the highlighted region (white box), the allele ratios differ significantly from 0.5, indicating an imbalance in the allele signals caused by a 1.6 Mb deletion.
(B) Two enlarged views of the deleted region, plotted as values of absolute difference between the heterozygous SNP allele frequencies in twin D8 versus twin D7, calculated in a similar way as shown in Figures 1C and 1E. The red line in both graphs displays the moving average, with a period of 30 and five data points in the left- and right-hand graphs, respectively. The solid and the dotted orange lines represent the average and the standard deviation of the absolute difference, respectively. The green arrows indicate the SNP rs11674089, used for confirmatory experiments using Melting Curve Analysis (C) and pyrosequencing (D).
To confirm the presence of the deletion and to estimate the number of cells carrying it, we performed high-resolution Melting Curve Analysis (hrMCA, [C]) and pyrosequencing (D). For hrMCA, reference samples were made from homozygous AA and BB individuals and mixes of AA:BB of 1:1, 3:1 and 1:3. MCA shows the signals for the homozygous AA and BB samples as nearly identical, whereas the signal of twin D7 (heterozygous for this SNP) matches that of the reference AB sample. The signal for twin D8 does not follow that of the reference AB sample but follows that of the AA:BB 3:1 sample instead, confirming the deletion and estimating it as being present in 70%–80% of the cells.
(D) displays pyrosequencing analysis of the SNP rs11674089 for twins D7 (left panel) and D8 (right panel). Taking the G to A ratio of twin D7 as normal (50:50), the G to A ratio in twin D8 shifts to 3:1, suggesting that ∼70%–80% of the cells analyzed carry the 1.6 Mb deletion.

Parallel CNV Analysis with Two CNV Platforms
We show five loci deviating in twin pairs 701/702 and 491/492 on the 32K BAC array (A, C, E, G, and I) and the Illumina HumanHap 300 Duo beadchip (B, D, F, H, and J). Regions altered between the two twins in each pair were defined as having at least two overlapping BAC clones that differ by ≥2 global standard deviations and are supported by at least one corresponding and one deviating SNP on the Illumina beadchip.
32K BAC array-CGH profiles were derived from two dye-swap experiments. Relative fluorescent intensity was plotted as a straight ratio. The data points shown were scored in both experiments and were not deviating more than one global SD from each other. The global mean and one SD for these two combined experiments were 1.003 and 0.039, respectively. The horizontal blue lines represent the average straight ratio, and error bars for each BAC show SD between two measurements in dye-swap experiments.
(B), (D), (F), (H), and (J) show the absolute difference in allele frequencies for the heterozygous SNPs on the beadchip for the corresponding chromosomal regions. The solid orange lines indicate the average absolute difference for all heterozygous SNPs for each of the twin pairs, and the dotted lines indicate the standard deviation (SD) for these. The green diamonds indicate SNPs deviating from the average with more than one SD. All genomic positions on the x axis are in million base pairs (Mb) and according to the hg18 build.

Statistical Analysis of Two Dye-Swap Experiments from Twin 291 versus 292
(A) Dye-swap-averaged log2 ratios of the data points from the long arm of chromosome 11 (11q) between twins 291 and 292. Boundaries of the inferred region of deletion are marked. A nonparametric Wilcoxon rank sum test comparing the values of the 120 data points within the region to the 359 data points outside the region yields a p value less than 2.2 × 10⁻¹⁶. In a series of n points, the number of contiguous regions of any length was n(n-1)/2, because a contiguous region was defined by its boundaries. The number of possible regions in the data shown in (A) was therefore 114,481. Thus, even the Bonferroni-corrected p value was on the order of 10⁻¹¹. This shows that the region of deletion was not merely the result of capitalizing on chance.
(B), (C), and (D) display the autocorrelation function (ACF) computed on the data shown in (A). (B) was computed with all data points from 11q, and (C) and (D) were computed on data points within and outside the deleted region, respectively. The ACFs show that the autocorrelation evident in (B) was due almost entirely to the deletion and was negligible after controlling for it.