Polygenic Risk Scores in Dogs: What the Next Generation of Genetic Testing Actually Tells You

Beyond the Single-Gene Model

For the first generation of consumer canine DNA tests, the model was simple. A disease was caused by a mutation in one gene, the test detected that mutation, and the result was a clear clear/carrier/affected classification. That model works for a small number of conditions — MDR1 drug sensitivity, von Willebrand disease type 1, certain forms of progressive retinal atrophy. For the majority of diseases that actually cause suffering in modern dogs, it does not.

Hip dysplasia, elbow dysplasia, most cancers, most behavioural conditions, epilepsy in most breeds, atopic dermatitis, and a long list of others are polygenic — determined by the interaction of many genetic variants, each contributing a small effect, often modified by environmental factors. A single-gene test for these diseases either does not exist or, where it does, captures only a small fraction of the genetic risk.

Polygenic risk scoring (PRS) is the next-generation response. This is how it works, why breeders are starting to encounter it, and where the current limits genuinely are.

What a Polygenic Risk Score Actually Is

A polygenic risk score is a weighted sum of the effects of many genetic variants across the genome on the probability of a specific outcome. In practice:

1. A reference population of dogs is genotyped at thousands to millions of variant positions.
2. The relationship between those variants and a trait (e.g., hip dysplasia score) is measured across the reference population.
3. Variants that associate with the trait are assigned effect sizes.
4. For an individual dog, their genotypes at the relevant positions are combined into a single score, weighted by each variant’s effect.

The output is a score, usually expressed as a percentile or a standardised number, that positions the individual relative to the reference population. A dog in the 90th percentile for hip dysplasia PRS carries more risk-associated variants than 90% of the reference population.

Why This Matters for Canine Breeding

Traditional breeding against a polygenic disease like hip dysplasia relied on phenotype selection — hip scoring the breeding dogs themselves. This works, but slowly, because an individual’s hip score reflects both genetic and environmental contributions, and because each dog’s phenotype says relatively little about the genetic quality of untested relatives.

Polygenic risk scores provide a second, independent signal. A dog whose own hip score is average but whose PRS is in the bottom decile for risk may be a better breeding choice than a dog with an excellent hip score but a PRS in the top decile. The score captures genetic risk the phenotype alone cannot see.

Livestock breeding has used genomic selection of this kind for over a decade — dairy cattle bulls are now selected primarily on genomic estimated breeding values, and the rate of genetic progress has accelerated substantially. Canine breeding programs, particularly in guide dog and working-dog populations, are beginning to adopt similar approaches.

Current Availability

As of 2026, polygenic risk scoring for dogs is available for a limited but growing set of traits:

If a commercial lab offers you a “polygenic” score for a trait, the right question is: what is the reference population, how large is it, and what has been published on validation in your breed? Answers should be specific, not marketing copy.

The Limits That Matter

Polygenic scores are not destiny. They are probabilities over populations, applied to individuals with substantial uncertainty. Key limits to understand before acting on them:

Breed transferability. A PRS developed in one breed often performs poorly when applied to another. A hip dysplasia score trained in Labradors may have weak predictive value in Bernese Mountain Dogs, even for the same disease.

Within-breed precision. Within the breed it was trained on, a PRS typically explains a modest fraction of total phenotypic variance — perhaps 10–30% for well-studied traits. That is useful for selection, but it is not diagnostic.

Environmental interaction. Polygenic risk interacts with environment. Hip dysplasia expression depends on nutrition, weight, exercise pattern, growth rate. A high genetic risk expressed in a lean, well-managed dog may not reach phenotype.

Pleiotropy. Variants that reduce risk for one trait sometimes increase risk for another. Aggressive selection on a single PRS can produce unintended consequences. This is one of the reasons genetic diversity in closed registries matters so much.

Reference population drift. As breeds change through selection, the relevance of older reference populations decreases. Scores require ongoing validation and updating.

How a Sensible Breeder Uses a PRS Today

For most companion and sport breeders in 2026, PRS is a supplementary tool, not a replacement for current practice. The practical approach I recommend:

• Continue full phenotype health testing on breeding candidates (OFA, elbow, eyes, breed-specific DNA)
• Where PRS is available and well-validated for your breed, include it as one additional data input on both sire and dam
• Do not select on PRS alone — use it to tie-break between candidates of otherwise similar quality, and to flag outliers in either direction
• Recognise that PRS thresholds for “good” versus “bad” will shift as reference populations grow and breeding decisions shift allele frequencies
• Interpret results in the context of the rest of the pedigree — PRS is a per-individual score, not a line assessment

For working-dog programs operating at scale, formal integration of PRS into breeding value calculations is now viable and genuinely improves the rate of genetic gain. Research groups at institutions including the National Human Genome Research Institute are increasingly applying these approaches to companion animal genomics.

Where This Is Going

Five years from now, I expect routine canine breeding panels to include polygenic scores for a dozen or more traits across the major breeds. Reference populations will grow, scores will refine, and the conversation about what a breeder “knows” about a dog’s genetic quality will shift meaningfully. Single-gene thinking is already incomplete for most of the health conditions that matter. The next generation of canine genetic testing will not abandon single-gene tests — they remain essential for the conditions they address — but it will layer over the top a probabilistic view of the genome as a whole. Understanding how to read that view, and how to integrate it with phenotype testing and pedigree analysis, is now a core competency for serious breeders.