Why dog genetic testing isn’t so simple — Speaking of Chemistry

Meet Ultraviolet, or UV, as her friends call
her. We know UV because her human is C&EN’s audience
engagement editor, Dorea Reeser. And while we get to spend a lot of time with UV, we actually don’t know much about her past. Dorea says the shelter where she adopted UV really had no information. But what if a genetic test could finally give
us some answers? Dorea: Everyone wants to know what she is. And it seems like more than half of the people
in the dog park have done it, so it’s kind of this cool badge of honor to be able to
say what your dog is from these DNA tests. Tien: Over about the past decade, companies
like 23andMe have popularized genetic testing for the average person, giving them a glimpse into their ancestral past. Now people are projecting that same genealogical
curiosity onto their pets. They want to learn the genetic makeup of their beloved mutt, or they want to know if their purebred schnauzer is really as pure as the breeder
claims. Last year the animal genetics market pulled
in about $4.2 billion worldwide. But are these products worth the money people
are spending? How do the tests work, and how accurate are
the results? Well, we tried to find out, with UV’s help. We ordered three different brands of dog DNA
tests to see how the results stacked up against one another. We also wanted to see how consistent a test
was with itself, so we bought three of the midrange Wisdom Panel tests to see if they
all gave the same results. UV graciously let us collect the DNA samples
for all five tests and then we mailed them off. And we waited. For a few months. And that was our fault. It usually takes only a few weeks to get the
results back, but funny story: we set off Wisdom Panel’s quality-control alarms with
our three tests. UV’s tests, which we submitted under three different aliases, gave the exact same results. And although Wisdom Panel eventually sniffed
out that the samples were from the same dog, they had to make sure they hadn’t accidentally
tested the same sample twice. So they sent us two more tests to confirm
what they were seeing. And that whole process took about a couple months. So one question answered. Do the tests give reproducible results? Yes, at least if you’re using Wisdom Panel. On to the bigger question: Do they work? By the time we got UV’s official test results from all three tests, we were pretty pumped. We even asked some colleagues to help us unveil
the results. And those results were . . . weird. Dorea: I have so many more questions. I was so excited about opening these and now
I just, I’m so confused. Tien: DNA My Dog said she was mostly German
shepherd. Wisdom Panel said she had an eighth each of a bunch
of breeds like dachshund and Lhasa apso. Embark said she was a third poodle, with other
breeds mixed in. While there was some overlap between the tests,
the percentages were all over the map. Clearly we needed help making sense of these
results. So I called animal geneticist Jessica Hekman,
who’s at the Broad Institute. And she really seemed to take the results in stride. Jessica Hekman: It’s not too surprising to
me that if you take a dog who’s got a zillion different things in her, which this one clearly
does, and give her to a bunch of different groups, that . . . biology is just messy and
genetics is just messy, and we’re not there yet to be totally exact. Tien: Jessica says consumer tests are great at identifying dogs that are purebred or half and half. Once you get to a dog that’s got an eighth
of a breed in her DNA, things get dicey. To understand why, we need to understand how
these tests work. Animal DNA tests work the same way as human
tests. These tests sequence, or read, the DNA in
a dog genome. But the genome contains about 3 billion base
pairs, so reading all that genetic information would get pretty expensive. Instead, most commercial tests read only a
fraction of that information. Specifically, they look at breed biomarkers,
which are short sequences of base pairs in the genome at particular locations that have
been associated with a breed. While geneticists have identified a ton of
human biomarkers, there’s much less published data on dog biomarkers. To figure out what biomarkers to look for,
companies get samples from purebred dogs and build unique breed reference panels. But there’s a lot of genetic diversity,
even within a breed. For example, border collies bred for show
are different from those bred for herding, so it’s important that these reference panels
contain as many lineages as possible. To look for these biomarkers in your dog,
a lab takes a sample, extracts the DNA, chops up the DNA, and pours it over what’s called
a sequencing chip. And it’s dotted with lots of little strings of DNA, each one designed to match a specific biomarker
in the genome. If your dog’s DNA sample has that biomarker, it’ll stick to the chip in a particular spot. Then the computer tallies up all the matches
and uses a proprietary algorithm to compare the results with the reference panel. And voilà! You have breed percentages. OK, now we know the tests identify breed biomarkers
with tiny chips and computer wizardry. But why is it hard for them to identify small fractions of a breed in a dog, as Jessica says? Jessica Hekman: You can think of it as having
these little windows where the whole world is dark and you get this tiny little window
to look at this little piece of information. And so these windows, you’re hopping around
and getting these bits. Tien: If a dog has is a single breed or has
a mix of two different breeds in her, then getting little bits of information while overlooking other portions of the genome isn’t a big deal. But for a dog like UV, whose DNA may include
as many as a dozen breeds all mixed together, looking at a sampling of the genome means
you may miss a lot of crucial information. Jessica Hekman: So visually you can imagine
that then if you could actually illuminate the whole thing, you’d see so much more. You’d be like, who knew all of these interesting
things that I missed because they happened to be between these two windows. Tien: So UV being a supermutt was a curveball
for all of the tests. But is there a way to tell which one was the
closest to being right? Jessica tells us there are two big factors
in a test’s accuracy: One is the number of breed biomarkers a test uses. More is better, she says. Wisdom Panel looks at about 1,800 breed biomarkers,
and Embark looks at about 200,000. DNA My Dog doesn’t disclose this number. The company did, however, claim that you need
only about 50 biomarkers for breed identification. We asked Jessica about that, and she says
the firm was likely misinterpreting a 2010 paper that points out about 50 regions of
interest in the genome associated with phenotypic traits, such as coat type, in different dog
breeds. But that doesn’t mean that these are the only locations in which breeds differ from each other. She says while it’s certainly possible to
identify a purebred dog with less than 100 biomarkers, you don’t know how
mixed a dog is going in. So again, more biomarkers are better. The other factor that can improve a test’s
accuracy is how many breeds are in its reference panel. DNA My Dog has about 100 breeds in its profile,
while Embark has more than 250 and Wisdom Panel has more than 350. So Dorea, what do you think? Which test do you trust the most? Dorea: I think I’d go with Embark since
they test for the most biomarkers and have the fanciest packaging. Tien: Yeah, I agree. But it’s not a cheap test. Dorea: No, it’s definitely not. It’s about $100 more than Wisdom Panel,
and their results weren’t that different. DNA My Dog was definitely an outlier. Tien: So now when people ask you what kind
of dog is UV, what would you say? Dorea: I would say she’s a dachshund-poodle-Lhasa
apso-German-shepherd-husky mix. OK, that was a mouthful. I think I’d go with supermutt. I like to think she has superpowers. Tien: Knowing what you know now, would you
spend your own money on a dog DNA test? Dorea: Absolutely. The results were a little confusing at first,
but it was really fun, and it’s interesting to have more pieces to UV’s puzzle. Tien: Have you ever tried a dog DNA test for your pet? Let us know about your experience in the comments. Thanks for watching.

You may also like...

3 Responses

  1. Chelsea Conlin says:

    This makes a lot of sense. We've had two of our dogs tested over the years (Wisdom Panel). Azu came back as a Lab/Boxer/German Short-haired pointer/Irish Setter mix, which made a lot of sense given her appearance, behavior, and later medical issues. Too bad I can't share photos here! Hazel, who is probably a mix of a LOT of different breeds, came back as 25% Chihuahua, 12.5% Mini Poodle, 12.5% Siberian Husky, and 50% "mixed-breed groups". Her results weren't really that surprising, either, as she's very weird looking, lol.

  2. monifern says:

    This makes sense. I used Wisdom Panel for both of my dogs. For one, I knew she was half white German Shepherd (or white Swiss shepherd). Her test results came back 50-50 GSD and Weimaraner. She looks 90% like a Weim and no one believes the GSD part, but I saw the mom, so I know it's true. My other dog is a super mutt. She looks sort of like a yellow Lab, but with a pinkish nose and amber eyes. Her test results were very random, but the only breed the was above the detection threshold was VIszla, which is actually the only breed in the test results that she even slightly resembles.

  3. rmt3589 says:

    I plan on getting a husky, so that will be interesting.

Leave a Reply

Your email address will not be published. Required fields are marked *