NEW YORK (GenomeWeb) – A Cornell University team has performed a large, across-breed genome-wide association study in dogs, uncovering variants associated with everything from body size and fur traits to dog diseases such as epilepsy, cancer, and dysplasia.
As they reported in Nature Communications today, the researchers genotyped more than 4,200 dogs for the GWAS, focusing on a dozen common dog traits and diseases. The analyses — which included dogs from 150 breeds and 170 mixed breeds, as well as 350 village dogs from 32 countries — led to loci with ties to several dog diseases, in addition to quantitative trait loci associated to canine body size, fur length, and shedding.
The team also used simulation studies to take a look at the number of dogs, SNP array marker density, and GWAS design that would be needed to track down the full suite of modest and large effect loci involved with canine complex disease.
"These results not only validate the dog as a large animal model for the discovery of genes associated with multigenic traits of importance to human medicine, but also demonstrate how to improve canine association-mapping methods for future studies on the genetics underlying complex canine disease," senior author Adam Boyko, a biomedical sciences researcher from Cornell University, and his co-authors wrote.
For their analyses, the researchers genotyped 4,224 purebred, mixed breed, or village dogs at some 180,000 SNPs using a semi-custom array that was intended to boost the team's ability to assess complex canine traits.
A more commonly used canine array typically tracks around 173,000 markers, the team noted, and has previously been used to map large-effect alleles related to canine conditions such as squamous cell carcinoma or canine compulsive disorder, provided at least 100 cases and as many unaffected controls from a given dog breed are available. Generally speaking, larger groups of pooches have been needed to find genes and expression quantitative trait loci underlying dogs' physical traits.
Starting from more than 65 documented dog traits or clinical characteristics, the authors of the new analyses narrowed in on the 12 most frequently phenotyped traits for the GWAS.
For example, the researchers used data for dogs from 82 breeds — 113 cases and 633 controls — to track down two loci linked to elbow dysplasia. One locus had stronger effects on elbow dysplasia risk in Golden Retrievers and English Setters, the other showed closer ties to the trait in Labrador Retrievers and German Shepherds.
Another 921 dogs from 69 breeds were included in a search for canine hip dysplasia, which led to a genome-wide significant associated locus in and around the transcriptional co-repressor-coding gene CTBP2 on the canine chromosome 28.
The locus showed pronounced effects in Golden and Labrador Retrievers, the researchers noted, but was not significantly associated with hip dysplasia in German Shepherds.
Along with the across-breed analyses, the team did several within-breed association studies, including a lymphoma analysis in Golden Retrievers, a search for idiopathic epilepsy in Irish Wolfhounds, and a Boxer-centered analysis of granulomatous colitis.
Within individual breeds, the researchers were able to find significant associations with far fewer dogs. Similarly, they demonstrated that a set of 17 QTLs showed more significant ties to size in a breed-level analysis involving nearly 1,900 dogs than in an individual-level analysis of 2,072 purebred, mixed, and village dogs.
The team's subsequent simulations suggest that GWAS centered on a single breed have higher power to pick up causal loci using smaller sets of dogs, though increased sample sizes and SNP array densities can boost the power of across-breed analyses.
"Our modeling indicates that future studies consisting of 500 to 1,000 cases and 500 to 1,000 controls from numerous purebred and mixed-breed populations, with denser marker spacing … will substantially increase the number of loci known to affect canine complex diseases," the authors concluded, "many of which are homologous to human disorders."