The digital DNA index estimates the value determined by flow cytometry in 488 neuroblastomas. DNA index was determined by flow cytometry as part of the clinical evaluation of children with neuroblastoma. For 349 out of 488 (72%) samples, the digital DNA index (average CN of all SNPs divided by two) matched the flow cytometry value. Discordant samples are shown that are not detected to be significantly aneuploid by either method.

Corrected LRR and CN of a chromosomal segment from a primary neuroblastoma. The top two plots show BAF and LRR. The third plot shows the LRR after correction for aneuploidy. The change in LRR across the segment indicates a change in CN. The relatively constant BAF of ∼0.33 restricts the possible CNs to multiples of three. After correcting for aneuploidy, the LRR values are most consistent with a region of three copies on the left and a region of six copies on the right. Algorithm output is shown in the bottom plot.

Aneuploidy affects chip-wide normalization. Data from three chromosome arms from a near-triploid neuroblastoma sample (DNA index = 1.43) assayed on the SNP array (A) as well as on a BAC-based aCGH platform (B). The leftmost chromosome shows decreased LRR and aCGH intensity ratio; the middle chromosome shows “normal” baseline LRR and aCGH intensity ratio; the rightmost chromosome shows increased LRR and aCGH intensity ratio. These intensity values imply CN = 1 for chromosome 3, which is inconsistent with the presence of heterozygous SNPs (BAF of 0.5). The LRR values would also imply CN = 2 for chromosome 1, which is inconsistent with the allelic imbalance seen in the BAF plot.

Correcting the LRR for aneuploidy improves CN determination. Data from three chromosome arms from a primitive neuroectodermal tumor assayed on the SNP array and analyzed by FISH. The top two rows of plots show BAF and LRR. The third row shows the LRR after correction for aneuploidy. Considering the corrected LRR values in conjunction with the BAF leads to the CN determinations plotted in the fourth row, which reflect the number of copies seen in the FISH images at the bottom. (A) The chromosome initially appears to be a CN loss (homozygous and negative LRR), but is present in two copies by FISH; this is consistent with the corrected LRR near 0. (B,C) Similarly, the corrected LRR reflects the correct CN of three and four copies, respectively.

B allele frequency (BAF) and log R ratio (LRR) across a single chromosome of a neuroblastoma cell line. The annotated chromosome regions (A–G) are plotted in the two-dimensional scatterplot of LRR and BAF. The regions labeled A are CN losses and fall in the light red zone (the nonzero BAF represents the presence of a minority of cells without the loss in the sample). The regions labeled B and C represent LOH without CN loss and fall in the blue zone (region B denotes LOH with CN gain). Region D is a four-copy gain with BAF ≈ 0.25 and increased LRR. The regions labeled E are made up of heterozygous SNPs present in two copies. The regions labeled F represent three-copy gains with BAF ≈ 0.33 and increased LRR. Region G denotes an amplification where the very high LRR is sufficient to distinguish it. (Bottom plot) CN as determined by the algorithm that detects the losses, the three- and four-copy gains, and the amplification.

Copy number (CN) determination using B allele frequency (BAF) and Log R ratio (LRR) across a single chromosome of a primary neuroblastoma. (A) A chromosome is displayed, from the short arm on the left to the long arm on the right. (Top plot) BAF values range from 0 to 1: areas of homozygosity have BAF of 0 or 1; normal diploid regions have BAF of 0, 0.5, or 1; areas of allelic imbalance show intermediate values; homozygous deletions have no detectable signal so the calculated BAF appears as noise. (Bottom plot) LRR values of 0 represent two copies with lower values in areas of loss and higher values in areas of gain. (B) Illustrates how CN is determined using LRR as a function of BAF. Each SNP window has a median LRR and median BAF, which fall in a colored zone in the plot; CN is then calculated based on the BAF for (green zone) gains and (red zone) losses. Gains can further be characterized by their number of minor alleles (NOMA). The yellow lines outline call zones for NOMA 1 (lowest LRR) to NOMA 4 (highest LRR). (Blue zone) Homozygous SNPs whose LRR is not consistent with loss have CN of 2 or higher with LOH; those CN are calculated based on the LRR. CN for amplifications (AMP) is also based on the LRR, while homozygous deletions (HD) have CN of 0. SNPs that fall in the gray zone are undetermined; CN is determined by interpolation.