Characterization of single-nucleotide variation in Indian-origin rhesus macaques (Macaca mulatta)

Abstract

Background: Rhesus macaques are the most widely utilized nonhuman primate model in biomedical research. Previous efforts have validated fewer than 900 single nucleotide polymorphisms (SNPs) in this species, which limits opportunities for genetic studies related to health and disease. Extensive information about SNPs and other genetic variation in rhesus macaques would facilitate valuable genetic analyses, as well as provide markers for genome-wide linkage analysis and the genetic management of captive breeding colonies.

Results: We used the available rhesus macaque draft genome sequence, new sequence data from unrelated individuals and existing published sequence data to create a genome-wide SNP resource for Indian-origin rhesus monkeys. The original reference animal and two additional Indian-origin individuals were resequenced to low coverage using SOLiD™ sequencing. We then used three strategies to validate SNPs: comparison of potential SNPs found in the same individual using two different sequencing chemistries, and comparison of potential SNPs in different individuals identified with either the same or different sequencing chemistries. Our approach validated approximately 3 million SNPs distributed across the genome. Preliminary analysis of SNP annotations suggests that a substantial number of these macaque SNPs may have functional effects. More than 700 non-synonymous SNPs were scored by Polyphen-2 as either possibly or probably damaging to protein function and these variants now constitute potential models for studying functional genetic variation relevant to human physiology and disease.

Conclusions: Resequencing of a small number of animals identified greater than 3 million SNPs. This provides a significant new information resource for rhesus macaques, an important research animal. The data also suggests that overall genetic variation is high in this species. We identified many potentially damaging non-synonymous coding SNPs, providing new opportunities to identify rhesus models for human disease.

Figure 1

Validation by category, direct comparison examples. Only the largest data sets are shown. (A) Validation in a single animal with multiple chemistries. All data for the reference animal 17573 are displayed by chemistry. SNPs falling into the overlap region are considered to be validated. (B) Validation in multiple animals with a single chemistry and with multiple chemistries. SOLiD data from re-sequencing efforts were compared to the unrelated animals r1766 and r02120. A large number (603,463) of additional SNPs not identified in the reference animal were identified and validated by comparing r02120and r1766. We observed roughly one-third the number of validated SNPs from r02120 that were obtained using the sequence data from r1766, due to reduced uniquely mapping sequence coverage in r02120 (Table 1).

Figure 2

Genomic distribution of SNPs. (A) The total number of SNPs identified in each chromosome. (B) The second bar graph shows the relative concentrations of SNPs validated in SNPs/Mb by chromosome. Only chromosome X (**) displays a SNP concentration that deviates more than one standard deviation unit (195.3 SNPs/Mb) from the mean (1057.1 SNPs/Mb).

Figure 3

SNP annotation. Genomic context of annotated validated SNPs. All annotations were obtained from Ensemble build 57 (released March 2010). The number of SNPs that fit into each category represented by a bar is listed over the respective bar. The bar for "Additional transcripts" represents only those SNPs where the specific annotation placement information was different between different transcripts for the same SNP.

Figure 4

Low stringency vs. high stringency SNP call validation effectiveness. Validation efficiencies cannot be determined using existing validated SNP data sets, as only 777 SNPs are currently available for rhesus macaque in dbSNP, most of which are polymorphic between subspecies (Chinese to Indian) rather than within subspecies (ie: Indian to Indian). We observed improved validation efficiency using low-stringency SNP calls rather than high-stringency. Both high and low stringency SNP calls were obtained for the reference animal (17573) mate-pair sequence data. The percentage of total SNPs validated in the low-stringency SNP set was slightly less (33.8%) than that observed in the high stringency SNP set (45.7%). In absolute numbers, however, there were 1.8X more SNPs validated from the low stringency SNP calls compared to the high stringency SNP calls.