Figure 4.

Simple CNVs and Inference of Genotypes, Based on Discrete Log₂-Ratio Clustering

For two CNV-containing genomic regions that have similar estimated breakpoints across all individuals, probe-by-probe log₂ ratios are depicted in heatmaps (see scale bar) in the upper panel (with rows representing individuals and columns representing probes ordered by genomic position). Mean log₂ ratios of the probes within the CNV are provided in the lower panel. The mean log₂ ratios form discrete clusters, letting us infer CNV genotypes. For both loci, there is one cluster with strongly negative log₂ ratios, suggesting that these individuals have homozygous deletions for this DNA segment. For the CNV on chromosome 4 at 155.5 Mb (hg17), there are three mean log₂-ratio clusters, likely reflecting zero, one, and two copies of this DNA segment. For the CNV on chromosome 12 at 130.8 Mb there are four mean log₂-ratio clusters, likely reflecting states of zero, one, two, and three copies; therefore, this CNV would be considered to be multiallelic. Error bars represent the standard deviation (SD).