Dog Breed Tests: Do They Really Work?

Otis with the Wisdom Panel dog-breed test in 2012. (Photo: Russell McLendon)

Editor's note: For an update to this article, see the results of a different DNA test, Embark, that Otis took five years later.

Recently, I introduced my quest to learn the lineage of my 2-year-old mutt, Otis (pictured above). Labeled a "husky mix" at the animal shelter, he has since grown to resemble a strange brew of canine breeds, inspiring people to guess everything from Akita and beagle to German shepherd and chihuahua.

I decided to clear things up by ordering a dog-breed DNA test, which involved swabbing skin cells from Otis' cheeks and mailing them to a lab for analysis. I originally ordered one last year, but its results were so unexpected that I recently splurged for a second opinion — and since other mutt owners likely share my obsessive curiosity, a co-worker suggested I write about what I learned.

The first breed test I ordered was from a company called BioPet Vet, which has since stopped selling them due to a patent infringement suit from Mars Veterinary, a division of U.S. candy giant Mars Inc. Aside from dog food and other pet products, Mars sells a breed test called Wisdom Panel — and since it dominated my Google search results, I figured I should give it a try. (Mars also seems to dominate the entire breed-test market now, thanks to its recent purchase of rival brand Canine Heritage.)

The main challenge with both tests was swabbing Otis' cheeks, which he tolerated only because there were two of us restraining/reassuring him. BioPet and Wisdom Panel each warned us to wait a while after he ate so food crumbs wouldn't contaminate his sample, and they also sent multiple cotton swabs to ensure accuracy. The kits included envelopes for stashing the DNA-laden swabs, and after that it was just a matter of filling out forms, sealing it all up and putting it in the mail.

Within a couple weeks of sending off each sample, we received the following results:

Test 1

I won't dwell on the BioPet Vet test since it's no longer commercially available, but I do want to share the results. Up to this point we still assumed Otis was part husky — or at least a similar-looking breed, like Akita — so imagine our surprise at this:

We were incredulous at first, but then we started to see it — his curled, bushy tail could be from a pug and Pekingese rather than a husky; his stocky frame and bow-legged gait are reminiscent of a bulldog; and we had already noted his beagley face and snout. Still, as most dogs can attest, once you start digging it can be hard to stop. When we heard Mars Veterinary had sued BioPet Vet for patent infringement, we couldn't resist trying again to see if we'd get the same results.

Test 2

Wisdom Panel's results were significantly different, but also more detailed. The only breed that showed up in both tests was pug, and Wisdom Panel found several new ones — namely Australian cattle dog and chow. It offered more information about Otis' family tree, and while it identified three of his ancestors as "mixed breed," it also listed five of those breeds along with a specific percentage for each.

Here are a couple screen shots from the PDF results we received:

Breeding between the lines

How could two breed tests yield such different results — especially since last year's patent dispute suggests they used similar methods? I emailed Mars Veterinary in hopes of learning more, but after an initial response from a PR representative, my followup emails were never answered. The question is addressed on Wisdom Panel's website, though, starting with this reassurance:

"We are often asked if it is possible to get different results on the same dog depending on which company performs the DNA analysis. The answer is yes, it is possible. But we're confident that Wisdom Panel Insights will provide you the most accurate results!"

Breed tests have been sold in the U.S. for about five years, but the technique dates back at least to a study published in 2004. It revealed a process that can match DNA sequences, aka "microsatellite genotypes," to specific dog breeds, reportedly with 99 percent accuracy in trials. This eventually led to the rise of commercial tests, which look for genetic markers in the cheek cells of customers' pets. Mars obtained an exclusive license to the technology, according to a report by the Veterinary Information Network, and later sued its rivals when it received a patent in 2010.

Any breed test's accuracy depends largely on the number of breed-related genetic markers in its database. Wisdom Panel claims to be the industry "gold standard" in this respect, citing 19 million marker analyses that let it identify 203 different breeds — "more than any other test on the market," according to its website.

This method makes sense on paper, says Joshua Akey, professor of genome sciences at the University of Washington. "In theory, it should have pretty good power to say the breed origin of a particular dog," he tells MNN. "You're essentially looking for genetic markers that have very different allele frequency. One allele might have a high frequency in great danes, for example, and zero percent in chihuahuas."

In the early days of dog-breed testing, companies were using just 30 or 40 genetic markers, he adds, which made them virtually useless if your dog wasn't from one of those breeds. Wisdom Panel's claim of 203 markers may indicate a leap forward, but Akey still cautions against overvaluing the results. "It does sound like the accuracy of these tests may be getting better," he says. "But these tests are still not perfect. There is some imprecision, and they won't give you a 100 percent match."

Otis mostly matches the profile of an outbred dog, whose genetic diversity tends to favor a moderate size and wolflike appearance. "Regardless of parental coloring, the coat color for mixed breed dogs is often a light-to-medium brown ... or black, frequently with white markings on the chest and elsewhere," the Wisdom Panel website says. "A brown coat with black across the top and sides is also quite common, especially in outbred dog populations." Outbred dogs typically weigh about 40 pounds, it adds, and stand "between 1 and 2 feet tall at the withers."

So after shelling out $80 per test and giving Otis a cotton-swab phobia, I'm slightly more confident about what kind(s) of dog he is. Was it worth it? Wisdom Panel certainly thinks so. Its slogan, "Because I love my dog," implies that not buying a test signals an unloved dog. But it also tries to manage expectations, warning that its test "is not designed to determine which disease traits — if any — might be present in a dog." Short-nosed dogs like pugs often have breathing issues, for example, but not all do; and even though Otis is apparently part pug, his snout is relatively long.

Aside from curiosity, part of the reason we wanted to learn Otis' ancestry was to anticipate breed-related health problems or exercise needs. We initially thought he was a husky mix — meaning he'd likely needs lots of activity — but we only bought the first test after we already suspected we were wrong. And as Akey points out, behavior and body size can usually tell us more about a dog's needs than a DNA-based breed test.

"It's interesting to explore the ancestry, but my intuition is that the things you're going to learn are things you already expected," he says. "My advice would be to not have unrealistic expectations about the new things you'll learn."