A high density SNP array for the domestic horse and extant Perissodactyla: utility for association mapping, genetic diversity, and phylogeny studies

Abstract

An equine SNP genotyping array was developed and evaluated on a panel of samples representing 14 domestic horse breeds and 18 evolutionarily related species. More than 54,000 polymorphic SNPs provided an average inter-SNP spacing of ∼43 kb. The mean minor allele frequency across domestic horse breeds was 0.23, and the number of polymorphic SNPs within breeds ranged from 43,287 to 52,085. Genome-wide linkage disequilibrium (LD) in most breeds declined rapidly over the first 50-100 kb and reached background levels within 1-2 Mb. The extent of LD and the level of inbreeding were highest in the Thoroughbred and lowest in the Mongolian and Quarter Horse. Multidimensional scaling (MDS) analyses demonstrated the tight grouping of individuals within most breeds, close proximity of related breeds, and less tight grouping in admixed breeds. The close relationship between the Przewalski's Horse and the domestic horse was demonstrated by pair-wise genetic distance and MDS. Genotyping of other Perissodactyla (zebras, asses, tapirs, and rhinoceros) was variably successful, with call rates and the number of polymorphic loci varying across taxa. Parsimony analysis placed the modern horse as sister taxa to Equus przewalski. The utility of the SNP array in genome-wide association was confirmed by mapping the known recessive chestnut coat color locus (MC1R) and defining a conserved haplotype of ∼750 kb across all breeds. These results demonstrate the high quality of this SNP genotyping resource, its usefulness in diverse genome analyses of the horse, and potential use in related species.

Figure 1. Decline in genome-wide linkage disequilibrium across and within breeds.

Genome-wide linkage disequilibrium (LD) was estimated both within a given breed, and across all breeds, by calculating r² values between all pairs of SNPs with inter-SNP distances of less than 4 Mb as described in Materials and Methods.

Figure 2. Distribution of pair-wise genetic distances.

a), within and between domestic horse breeds; b), between domestic horse breeds and between domestic horse breeds and Przewalski's Horse. Genetic distance (D) between pair-wise combinations of individuals was calculated as described in Materials and Methods.

Figure 3. Multidimensional scaling with 14 domestic horse breeds.

Metric multidimensional scaling analysis of pair-wise genetic distance was used as described in Materials and Methods to identify relationships between the 14 domestic horse breeds.

Figure 4. Phylogenetic tree of extant Hippomorpha.

Unrooted cladogram constructed via parsimony analysis, considering only the Hippomorpha, using 40,697 autosomal markers. Bootstrap support >50% calculated from 1000 replicates is shown.

Figure 5. Phylogenetic tree of domestic horse breeds and Przewalski's Horse.

Parsimony analysis across 46,244 autosomal loci in the Domestic and Przewalski's Horse. The tree is rooted with the domestic ass.

Figure 6. Allele association analysis for three known equine coat color loci.

a), mapping of the chestnut locus across breeds, based on inferred coat color phenotype; b), mapping of the black (agouti) locus across breeds, based on inferred black coat color; c), mapping of the gray locus across breeds. Phenotypes were inferred from the genotypes at 9 known coat color loci and known inheritance models as described in Materials and Methods. Unstructured case control association analyses using chi-square tests for allelic association were then performed on a pruned SNP set also as described in Materials and Methods. SNPs on each chromosome are labeled with a different color on the X axis as indicated.