Our method is similar to restriction fragment length polymorphism (RFLP). In RFLP analysis a genomic DNA sample is digested by restriction enzymes and the resulting fragments are separated by gel electrophoresis. Instead of using gel electrophoresis we use a microarray.
Restriction enzymes cut molecules of DNA at specific recognition sites, so cutting with a particular enzyme should always produce the same size and number of fragments. However, when a site is polymorphic, or existing as two alleles, the restriction enzyme will not recognize one allele and will not cut there, leading to a length polymorphism.
The ~60,000 SNP probes on the CGH+SNP microarray span variant AluI or RsaI restriction enzyme recognition sites and measure the copy number of the uncut allele at those loci (Figure 1). We measure the total copy number of the region encompassing the SNP site by neighboring CGH probes. The copy number of the cut allele can then be inferred from the total copy number and the copy number of the uncut allele.
A SNP copy-number call is made from the log ratio of the test sample (Cy5 signal) versus a genotyped internal reference (Cy3 signal). The copy number of each SNP in the reference sample is known. In order to determine the allele-specific copy number of the test sample, the log ratios of the SNP probes are adjusted by the copy number of the reference sample.
The reference-adjusted log ratios fall into three categories corresponding to the copy numbers of the uncut alleles in the sample, which correspond to the three possible diploid genotypes for the SNPs: AA, AB, or BB. Regions of copy-neutral LOH or UPD are then located by identifying genomic regions with a statistically significant scarcity of heterozygous SNP calls.