Identification of copy number variation hotspots in human populations

Abstract

Copy number variants (CNVs) in the human genome contribute to both Mendelian and complex traits as well as to genomic plasticity in evolution. The investigation of mutational rates of CNVs is critical to understanding genomic instability and the etiology of the copy number variation (CNV)-related traits. However, the evaluation of the CNV mutation rate at the genome level poses an insurmountable practical challenge that requires large samples and accurate typing. In this study, we show that an approximate estimation of the CNV mutation rate could be achieved by using the phylogeny information of flanking SNPs. This allows a genome-wide comparison of mutation rates between CNVs with the use of vast, readily available data of SNP genotyping. A total of 4187 CNV regions (CNVRs) previously identified in HapMap populations were investigated in this study. We showed that the mutation rates for the majority of these CNVRs are at the order of 10⁻⁵ per generation, consistent with experimental observations at individual loci. Notably, the mutation rates of 104 (2.5%) CNVRs were estimated at the order of 10⁻³ per generation; therefore, they were identified as potential hotspots. Additional analyses revealed that genome architecture at CNV loci has a potential role in inciting mutational hotspots in the human genome. Interestingly, 49 (47%) CNV hotspots include human genes, some of which are known to be functional CNV loci (e.g., CNVs of C4 and β-defensin causing autoimmune diseases and CNVs of HYDIN with implication in control of cerebral cortex size), implicating the important role of CNV in human health and evolution, especially in common and complex diseases.

Figure 1

Distribution of the Estimates of M in Three HapMap Populations The distribution of the estimates of M for 3147 polymorphic CNVRs in YRI (A), 2360 polymorphic CNVRs in CEU (B), and 1615 polymorphic CNVRs in CHB+JPT (C) is plotted. The range of M corresponding to a given mutation rate is indicated in the upper part of each plot (black for ∼10⁻⁵ per generation and red for ∼10⁻³ per generation). The range of M for a given mutation rate indicates the 2.5–97.5 percentile, based on 1000 coalescent simulations, which are based upon properly selected demographic models for each population.

Figure 2

Comparison of CNVRs with Potential Hotspots and Nonhotspots The percentage of CNVRs overlapping with SDs (A), the recombination rate (cM/kb) between a CNVR and its flanking SNPs (B), the CNVR size (C), the genetic diversity of the CNVR (D), the percentage of NAHR-mediated CNVRs (E), and the percentage of VNTR-mediated CNVRs (F) is illustrated for CNVRs with nonhotspots and potential hotspots. Statistically significant differences (p < 0.001) between the hotspot group and the nonhotspot group are indicated by asterisks.