Comparison between human and rhesus macaque CNVs and enrichment of genic CNVs in rhesus macaques. (a) Size distribution and summary statistics of the observed CNVs. The frequency distribution of CNV sizes. CNVs that are larger than 300 kb were omitted from the graph. The red and blue bins show the frequency distribution of the CNV sizes observed in this study and in Lee et al. [14], respectively. Note the overall increase in the frequency of detected CNVs in this study. (b) The size distributions of human and macaque CNVs are similar (that is, as the frequency increases, the size gets smaller). However, the human CNVs were identified with a higher resolution platform and so there are smaller human CNVs detected compared to rhesus macaque. (c) The overlap with genomic components, such as repeats and segmental duplications, are similar for human and macaque CNVs. However, rhesus macaque CNVs are much more likely than human CNVs to overlap with genes. (d) As observed in humans, most of the rhesus macaque CNVs that overlap with genes are gains or multi-allelic CNVs. Note the dramatic increase in the ratio of genic CNVs.

Enrichment analyses of primate CNV hotspots. (a) Chimpanzee and rhesus macaque CNV coordinates were converted to human reference build hg18 coordinates using the Galaxy Liftover tool. HC and HR CNVs are human CNVs that overlap one or more chimpanzee or macaque CNV(s), respectively, when using a 50% overlap criteria. For human CNVs that overlap both chimpanzee and macaque CNVs (HCR), critera were employed that required at least a 20% overlap between CNVs from any two species and a minimum of 50% overlap for overlapping CNVs from all three species. (b) We generated 1,000 random permutated CNV datasets containing CNVs of similar size distribution to the 12,146 human CNVs identified in Park et al. [15] and Conrad et al. [1]. The CNVs in each permuted human dataset were then assessed for their overlap with known chimpanzee and rhesus macaque CNVs. The x-axis represents the number of human permuted CNVs that overlap with chimpanzee CNVs (red distribution) or macaque CNVs (blue distribution) during these permutation iterations. The green distribution represents human CNVs that overlap with both chimpanzee and rhesus macaque (HCR) CNVs during these same permutation iterations. The y-axis represents the frequency of the permutated human CNVs that fall into each category. The dotted vertical lines indicate the actual number of overlaps with either chimpanzee CNVs (1,387), rhesus macaque CNVs (467) or both (170). (c) Based on the expectation of overlap with 1,000 random sets of intervals depicted in (b), the fold enrichment of human CNVs that overlap with chimpanzee (HC) or rhesus macaque (HR) CNVs is plotted, as is the fold enrichment of human CNVs that overlap with both chimpanzee and macaque CNVs (HCR). Error bars represent 1 standard deviation from the mean fold enrichment.

Primate CNV hotspots are more likely to be complex in nature. (a) Depiction of a typical primate CNV hotspot region, which is complex in nature, based on the presence of multiple overlapping CNVs with different breakpoints. Multiple members of a single gene family are present in this genomic region and may be contributing to some of the different CNV formations. CNVs were defined as complex based on whether they reciprocally overlapped other CNVs by less than 50% [1]. (b) The ratio of complex and non-complex CNVs in primate CNV hotspot and non-hotspot regions. Note the greater number of complex CNVs among HCR CNVs. (c) Enrichment analyses for the 1000 Genomes Project hotspots overlapping HCR CNVs. We iterated 1,000 intervals that mimic the size distribution of 55 human CNV hotspot regions [17], and constructed the expected overlap distribution of these intervals. Blue bars show the expected distribution and the dotted line marks the observed overlap. Here, we are showing that the observed overlap is much higher than expected by chance independent of the size of the intervals (P < 0.001, Kolmogorov-Smirnov test).

Primate CNV hotspots overlap functional regions. (a) CNVs were classified as intergenic, intronic, or exonic, based on whether any part of them overlapped with genes. Promoter regions were defined as the 2-kb region immediately upstream of the transcription start site of a gene. (b) Enrichment analysis for the HCR CNVs that overlap with exons of UCSC gene track. For calculating the significance independent of the size distribution, we generated 1,000 intervals that mimic the size distribution of the HCR CNVs and plotted the distribution of their genic overlap. The dotted red line indicates the number of observations, whereas the red-bins show the expected distribution (P < 0.001, Kolmogorov-Smirnov test). (c) The K values for all genes, genes that overlap with human CNVs, and genes that overlap with HCR CNVs. We plot here the density of K values, which is a measure of positive selection [18]. The genes that overlap with HCR CNVs have significantly lower K values (P < 0.001, Kolmogorov-Smirnov test), indicating that they are more likely to evolve under positive selection. (d) A cumulative fraction plot of conservation for primate hotspot and non-hotspot CNVs. Conservations scores were obtained using the phastCons on 17-species multiz track from the UCSC Genome Browser [31]. The D and P-values were calculated using a Kolmogorov-Smirnov test. Combined, these data indicate that most of the genes overlapping HCR CNVs are evolving under lineage-specific positive selection.

Impact of CNVs on gene regulation. (a) There are multiple ways in which CNVs can impact transcription by overlapping coding regions of the genes. (b) Blekhman et al. (2010) used RNA-seq data to determine whether specific genes are differentially expressed between human, chimpanzee, and macaque [32]. Based on their results, we plotted the proportion of human CNVs (H) and hotspot CNVs (HCR) that are differentially expressed between species. In particular, 3,423 of the Ensembl genes analyzed by Blekhman et al. (2010) overlap with human CNVs. Based on these data, here we plot the proportion of the genes that are differentially expressed between two or all three species, evolved under directional selection in the human lineage (Directional Human) or under stabilizing selection (that is, no expression differences between species). (c) At least three HCR CNVs overlap with regions with clear enhancer and/or promoter signals in the genome. To visualize the enhancer and promoter activity, we used the H3K4Me3 track generated by the ENCODE consortium [33] from the UCSC Genome Browser.