How to Audit a Dog DNA Test and Interpret the Results

Split the report into five different products

AKC explains that its traditional DNA profile is for identity and parentage, not breed identification or health. Commercial ancestry and health panels add different analyses. Compare the exact product, not the company name alone.

Treat breed ancestry as a model output

AKC’s veterinarian-authored heritage guidance says breed ancestry tests compare DNA segments with reference panels of known dogs. Clubs and registries define breeds; a company reference panel cannot certify pedigree or purity. Village-dog populations, closely related breeds, recently developed breeds and underrepresented regions can be particularly difficult to classify.

A 2024 JAVMA comparison sent six cheek swabs from each of 12 registered purebred dogs to six direct-to-consumer companies. Five tests always included the registered breed, while one often did not; predictions also differed across tests. The study is small and used purebred rather than unknown mixed-breed dogs, so it does not rank every current product. It does show that “DNA test” is not one uniform method and that company choice can affect the report.

Ask the seller:

• How many reference dogs represent each reported breed, country and working or show population?
• What quality and pedigree checks admit a dog to the reference panel?
• What minimum ancestry percentage is reported and how are low-confidence segments handled?
• Can the algorithm return unresolved or village-dog ancestry instead of forcing a named breed?
• Will the report change when the reference panel or algorithm changes, and are old versions preserved?

Read every health result through three validity gates

The AKC Canine Health Foundation and OFA whitepaper separates analytical validity—whether the lab measured the genotype accurately—from clinical validity—whether the variant relates to the stated condition in the relevant dogs—and clinical utility—whether acting on it improves a decision. A polished dashboard can hide a weak bridge at any gate.

For each result, record the gene and exact variant, inheritance model, direct-versus-linkage method, breeds in which the association was established, penetrance, age of onset, severity, modifiers, independent replication, and recommended confirmation. “One copy detected” can mean carrier, increased risk, dominant risk, uncertain significance or little relevance depending on the condition.

Breed relevance matters more than panel size

AKC’s DNA health testing guidance notes that many variants in a broad panel do not apply to every breed. A disease name may be caused by different variants in different populations, and a panel may test only one. Hundreds of negative markers therefore do not create a universal clean bill of health.

For a known breed, use the national parent club and OFA CHIC recommendations to identify priority screens. Many important evaluations are not cheek-swab tests: hips, elbows, patellas, eyes, heart, hearing and other phenotypes may require imaging or examination at defined ages. A CHIC number indicates completion and public availability of required tests; it does not promise that every result was normal or that the dog will remain healthy.

Start at the Pawpry breed profiles for context, then use the parent club’s current health statement and OFA record rather than a generic breed-risk card.

Audit the laboratory and chain of identity

1. Confirm the legal laboratory name and location, not just the retail brand.
2. Ask about relevant accreditation, proficiency testing, control materials, replicate rules and sample failure rates.
3. Ask whether the method detects the causal variant directly or uses a linked marker.
4. Check how the swab is barcoded, received, extracted, stored and matched to the dog.
5. For a decision with high consequences, ask whether a veterinarian-collected sample and an independent laboratory confirmation are appropriate.

Voluntary quality credentials can be useful signals but do not validate every ancestry algorithm or disease interpretation. OFA’s lab and test lists help with accepted submissions, yet OFA also cautions owners to confirm current availability and breed specificity with the laboratory.

Keep seller claims separate from performance evidence

Embark currently markets hundreds of breeds and varieties, more than 270 genetic health risks, traits, relatives and personalized recommendations. Those numbers describe the product’s claimed scope. They do not tell you the sensitivity of every variant, ancestry accuracy for every population, penetrance of each risk, or whether a recommendation improves health.

• Most accurate: ask for the comparator, sample composition, preregistered metric, current algorithm and independent replication.
• Screen for hundreds of risks: count only variants relevant and interpretable for this dog.
• Personalized care: identify which recommendation is genotype based, breed based or general dog care.
• Find relatives: check relationship confidence, database coverage, user verification and privacy settings.
• Stay ahead of disease: request evidence that the result changes a beneficial screening or treatment path.

Decide about data before collection

A dog’s sample can be connected with owner identity, location, pedigree, relatives and veterinary information. Read the current policy and consent screens for sample retention, research participation, commercial partnerships, de-identified data, account deletion, biological-sample destruction, law-enforcement requests, company sale and policy changes.

Screenshot the terms and settings on the purchase date. Determine whether relative matching and research are separate opt-ins. Ask what deletion removes: the online account, derived data, raw genotype, backups and the physical sample may have different rules. The privacy practices of human DNA services cannot simply be assumed to apply to canine services, but the decision still deserves informed consent.

Use the privacy worksheet in the GPS tracker audit to map account, location, retention and deletion questions. For how Pawpry separates source types and unresolved limits, see the editorial methodology.

Use a result without turning it into a diagnosis

1. Verify the dog and sample identifiers and check for contamination or collection problems.
2. Save the original PDF, raw genotype if offered, test version, date and policy snapshot.
3. Separate ancestry, traits and health pages.
4. For a health flag, locate the exact variant and laboratory interpretation; do not act on the disease name alone.
5. Ask the veterinarian whether confirmation, examination, imaging, laboratory monitoring or no action is appropriate.
6. Record symptoms and family history independently of the panel.
7. For breeding, involve the parent club, OFA resources, the veterinarian and a qualified genetics professional; do not remove a dog from a gene pool from a misunderstood result.

Equipment and behavior decisions remain individual. A reported breed percentage does not select a harness size, explain aggression, set an exercise dose or predict trainability. Measure the dog with the harness guide and evaluate behavior in context.

Decision rule

A worthwhile test answers a defined question with a transparent method and an action proportional to the evidence. If the goal is curiosity, ancestry uncertainty may be acceptable. If the result could change medical care or breeding, demand variant-level evidence, breed relevance, laboratory quality, confirmation and professional interpretation. The higher the consequence, the less room there is for a colorful but ambiguous dashboard.