My Photo


follow us in feedly

Enter your email address:

Delivered by FeedBurner

Best Of KC Dog Blog

Become a Fan

« Weekly Roundup - Week ending 5/20/12 | Main | Two more Ohio cities repeal their breed-specific laws »

May 24, 2012


H. Houlahan

I have seen nothing to suggest that the "breed ID" DNA-testing products are anything other than snake oil.


H -- my experience with the Mars Wisdom Panel seems pretty solid. I think a good many of the "Snake oil" ideas were based off of a few early tests and when testing known breeds whose markers were not a part of the database (one famous one is the one where the woman has a purebred Am. Staff and does a test that doesn't include Am. Staff in the database -- well, of course it comes back with jacked up results).

My experience is that a) the tests have gotten better over time and b) if the dog's breed is in the database they seem pretty accurate and c) the Mars Wisdom Panel has a much deeper database of breeds and is more consistent than the other tests.

Central Ohio Dog Blog

Here's the comment I left at the end of the survey, which sums up my thoughts: "That was hard, and was an interesting exercise for me to see what I look for in breed identification. My hypothesis is that from non-professionals (folks without animal medicine/physiology training) you will likely see them prioritize obvious traits like coat color, ear shape, and musculature and de-emphasize things like body size, shape, bone structure, etc. I'm also going to guess that most folks will reserve their guesses to more common breeds, such as german shepherd dog, labrador retriever, golden retriever, beagles and some other hounds, bully and shepherding breeds. I'll be really interested to see the article that comes out of it. "


I don't think I did well. It reminds me of a site i posted a while back where some dogs taken from websites devoted to that breed were featured and we were supposed to identify which one was the Pit Bull. None of us got the right dog on our first many guesses.


CODB - - I agree that most people will gravitate toward more popular breeds. In reality, most people, even in animal welfare, don't know what 1/3 or more of the breeds on the list really look like (for instance, I don't know small dogs well, or a lot of the very specific hunting breeds). The reality is, by gravitating to the more common breeds they'll be right more often, but it would be the first instinct since those are the dog breeds we know.

Jan, many of of the "find a pit bull" games mostly include pure bred dogs -- it's so much harder when mixed breeds are involved.

Central Ohio Dog Blog

But the "find the pit bull" games are hard enough, even with pure breeds! Seriously, I kinda felt like a second grader looking at a calculus test. I didn't even know where to begin!
And it also makes sense to gravitate towards popular breeds, since they're more popular and more likely to have mixed progeny out there, showing up in shelters. Sorting through that whole long list of 180 breeds felt a bit overwhelming, and I definitely googled a number of the breeds I didn't recognize.
I am really curious to see the results on July 1, and I'm even more curious to see the findings of the study.


I'm pretty sure the findings will show that we don't know what we're talking about when it comes to labeling dog breeds. And even further, will then cast doubt on pretty much all the breed-based dog-bite studies -- all of which rely on visual breed ID. If we're right 20% of the time, these "studies" aren't going to fair very well.

For the record, I got this sent to me by an AKC judge -- and it is my experience that the show breeders are far better at this that rescue/shelter workers because they're the ones that know what all these breeds really look like.


Brent, re: your response to H. Houlahan, you are right on the money. These tests have gotten progressively better as the whole plethora of single nucleotide polymorphisms ("Snips") that characterize specific breeds have been identified more fully. I don't recall the exact number, but I think a recent seminar I viewed mentioned about 30,000 dogs in the Mars database representing 99% of the AKC registered breeds.

Because many breeds share certain mutations (in hair color or leg length, etc), there will naturally be some overlap and even amazement. One specific case showed a 21 lb dog that for all intents and purposes looked like an overgrown Papillon....turns out that the dog had cocker spaniel (generating some of the parti type of color) along with Maltese and another breed that I don't recall immediately.

I will have to go see this sounds amazing.


I dunno, I've seen some pretty wonky results from these tests. I knew a pointer mix who came back as mostly Maltese.


Until these companies allow some independent testing, they are still snake oil in my opinion and should be treated as such.

As for the photo tests, they are bogus as well. Yeah, it's fun that Tom Skeldon couldn't ID the "pit bull. But you can't ID a dog's breed from a couple of still photos.

Yes, you DO need to see movement, behavior and all the other action characteristics that ALSO define breed. Because breed is defined NOT just by appearance, but by behavior.

In the end, since all dog "breeds" descend from a few "types" of dogs, within the last 200 years, I would expect that DNA tests (even if accurate) would show many different and apparently surprising results.

So what?

If DNA tests are accepted by courts to save dogs, fine.
If they're being used as yet another stalking horse to deny that distinctive breeds actually do exist, and have value, then they serve some other, less laudable, agenda.


Cristy - -It makes a big difference what test you use. Most have very shallow breed pools which obviously impacts their validity. My experience from the Mars Wisdom Panel test is that its depth of breeds makes it far more reliable than other tests.

Emily, agree that judging breeds based on photos is not the best way. I have a foster dog now that looks pure Basinji - but its behavior makes us very much think she's a Smooth Coated Collie. Obviously very different behaviors involved.

I would hope that no one is trying to say that breeds don't exist. That would be disappointing and certainly not my intent. I would love for testing to allow us to get better insight into shelter dogs to be able to better match them with owners and look for disease issues -- and to note the inaccuracies of visual ID (which I think for mixed breed dogs is nearly impossible most of the time).

I'm sure you've seen this -- but I think it's a great article from Nat. Geo on "How to Build a Dog" - and talks about how few genetic markers dogs have because of they "evolved" via unnatural selection vs natural selection -- so markers with the greatest physical appearance or behavior changes were selected for breeding. It's an interesting read for folks who haven't seen it.

Dianne R.

I think people de-bunk the tests because what studies have shown is that when we guess a dog's breed by sight, 85% of the time we are wrong and the DNA is right. I took the test and please remind us in July when they post the DNA profiles. This should speak volumes to the nonsense of BSL.


jeeze. that was hard.


The Wisdom Panel is full of...something.

From one location (on pit bulls) on the the Wisdom Panel website:
“Wisdom Panel™ Insights is designed and intended to be used solely to identify the genetic history of a mixed-breed dog and no other purpose is authorized or permitted.

Wisdom Panel Insights is not intended to predict behavior in any particular dog. Each dog is unique and their physical and behavioral traits will be the result of multiple factors, including genetics, training, handling and environment.”

From another location (wisdom panel insights):
"Wisdom Panel® Insights™ ancestry report gives you the power to:
1. Create a more targeted and effective behavior training plan.
2. Plan exercise and play activities that cater to your dog's natural tendencies."


Pet peeve, so this is long. Sorry.

I'm, with Heather and Emily. Certainly if you want to ID a purebred, since the AKC gave something like 10,000 samples of each breed to the dog genome group, you should achieve some accuracy but it will likely never be 100% because science doesn't work that way. Even mitochondrial DNA tests are not 100% accurate.

This, of course, would be very easy to verify - just send in DNA from your purebred dog and see what comes back.

However, when trying to guess the breed of a mutt using either a visual or a DNA sample test, you have no way to confirm your accuracy. The only way you could be sure of the results would be to have scientifically randomly bred dogs going back generations, including multiple crosses, tested. This would require a huge number of dogs, an immense amount of record-keeping over generations and would, I'm afraid, ultimately be pointless.

The fact that there is no solid science behind MIXED breed testing is confirmed by Mars and other who do these tests. They state that the tests are not to be used as evidence in court cases. In other words, they don't stand behind the results and around here, none of the companies will send a rep to court to be examined. We've tried.

That's all you need to know.

It's a party game, and that's cool as long as everybody is aware of that.

I did the test just because I have always been fascinated by dog breeds and breed characteristics. I posted it on Facebook but said that it will take two inaccurate identifications to prove who knows what.

A mutt is a mutt and there's nothing wrong with that. The only reason to say Collie type is to indicate size/haircoat, etc. It has nothing to do with breed. Also, if these mutt tests were accurate, they would show all breeds in a dog's makeup, even a purebred, since they were all originally mutts too.

What they are doing is running against a grid, which you could do manually.

So, say there's a gene for brindle. That cuts out a bunch of breeds and brings it down to about 40. Then say there's a gene for dwarfism, then you might have a Dachshund, Basset hound, Corgi, etc. Then ear shape, tail type, blah blah blah.

DNA can't say how big you are, for example, which is why you get these Italian Greyhound x Tibetan Mastiff and other wacky results. Two rare breeds, neither out running around mating indiscriminately, yet somehow the Border collie-looking thing in front of you is related to these breeds.

There are types, some ancient, including mastiffs, sighthounds, the Chinese dogs and the pariah types that pop up in all breeds.

Understand I'm not saying it's impossible because anything is possible.

I'm saying that until these guys will stand behind their results and until there is a shred of scientific evidence that these tests on MIXED breeds dogs work, I'll play but I won't invest.

Paula G From Indiana

I'm sure I didn't do well. I picked out the closest breed I could. I too don't think highly of these breed DNA tests, every one I've heard of gives absurd results. I'll take your word on it Brent, that this one will be more accurate. This is exactly why I (along with many of you) get so frustrated with dog bites being identified as done by "pit bulls" and banning any dog with even a fraction of "pit bull" in it. I'm sure I'm way off the mark on these samples and I sure wouldn't want to have the job of officially plastering a breed mix name to a dog of unknown breeding.



But you and I cannot run a dog's looks through a grid -- because dominant and recessive genes don't work that way. We can only judge by what we SEE, but there are recessive genes in there that we cannot see that have to be accounted for.

I agree that there are a lot of questions that still linger on this topic -- it is pretty new science at this point. For instance, if someone has a Great Dane, does the dog show up at 100% Great Dane? Or 50% Great Dane and 25% each Mastiff and Greyhound? I'm not even sure what the "right" answer should be on something like this.

No doubt that it's a work in progress, but given how far they've come in the science in the past decade it's definitely better than "snake oil", but not absolute either. But given the choice between science and someone just looking and guessing, even 'experts' looking and guessing, I'd bet on science being more accurate. And if we accept Emily's point of view that "breed" is more than just looks, but also behaviors -- then it seems like the information could be far more useful than as a party game...


I took the test and I didn't think it was that difficult. I didn't overthink my answers - I just called most of them the first common dog breed that came to mind. I think there were about three dogs for which I picked the answer "no predominant breed", or whatever that choice was. I doubt that there are very many Ibizan Hound mixes or Toy Manchester Terrier mixes in the shelter. A couple of the dogs looked suspiciously like foreign breeds that weren't listed as possible answers. I saw what looked like a Portuguese Pointer and another dog looked like a young Berger Picard on the test.

Tests like this bother me because I don't believe for a minute every mixed breed out there is a cross between a purebred 13 inch Beagle and a purebred Cocker Spaniel, for example. I think a lot of the shelter dogs and strays in urban areas (and probably rural areas as well) are mixes of mixes of mixes of mixes of mixes. Very few of the dogs in shelters that are labeled as "pit bulls" bear even a remote resemblance to the registered American Pit Bull Terriers shown in conformation at ADBA shows, which probably explains why I flunk those "Identify the pit bull" tests every single time.

IMO, the only way to test DNA breed identification for accuracy is to take known crosses of purebred dogs, or better yet, breed them intentionally so you know exactly what the two breeds are (or however many breeds are in the pedigree) and then see if the DNA test matches what you already know. I kind of wondered if the people that put the survey together didn't use dogs of known parentage - that would make a lot more sense to the scientist in me, but they wouldn't let the survey-takers know that they knew which breeds were in the dogs because that would introduce bias. If they told me "We intentionally crossed two purebred dogs to get these puppies and we want you to tell us what breeds you think the parents were" I would be far more inclined to respond, "Looks like a cross between a Flat Coated Retriever and a Belgian Malinois". If it's a shelter dog I'm going to respond, "Labrador/GSD cross".

Personally I don't see how the DNA identification is useful at all, other than to possibly prove a dog is not a "pit bull" and you're trying to save someone's dog. If you want a breed that looks a certain way and has certain traits and breed characteristics go buy a puppy from a breeder and skip all this horsepuckey. If a person gets a dog from a shelter does it really matter what's in the woodpile as long as they love it, they train it, they're happy with it, and it has a good home?

There's a young man that is taking obedience classes at our training club and has been to some of our pit bull club meetings. He has a very sweet dog he adopted from KC Pet Project. She kind of looks like a smooth Border Collie/pit bull mix. He is having her DNA tested and said he wanted to know what she was. I told him to be sure and let me know - I'm interested.


Oh, and I guess it's possible that DNA tests could be useful when it comes to identifying breed when there's a dog bite/attack victim, as someone stated. Assuming the test is accurate, of course.


I agree with KMK obviously.

Brent, my point is that for purebred dogs, these tests are relatively accurate - no scientist will every say a test is 100% accurate, which is another red flag with a couple of the smaller DNA my mutt companies around here.

Who says they've made great progress in identifying mutts? Companies that sell these tests to pet owners, so right there I'm not likely to believe it without proof.

The thing is, 'breeds' are really ethnic groups. For example DNA can't tell you what my ethnic background is, since it's all Caucasian for many generations back. You can't tell if I'm Irish, Scottish, English, Norwegian, Danish, German, etc, etc, etc. You can tell I'm human and by elimination of certain racial markers, that it's unlikely that I'm Asian or African in recent origin (we all come from Africa originally).

So, with no way to accurately measure the results, except as KMK points out and I mentioned, by deliberately breeding multiple generations of mongrels and keeping what amounts to pedigree data, there is no point in doing the test.

If these people did have deliberately bred mongrels spanning multiple breeds and generations that were bred under controlled conditions, then it would make some sense but in that case, the DNA wouldn't be needed anyway. The study would have to be double-blind, ie, the researchers would have to not know and the participants would have to not know the makeup of the dogs' ancestry.

If anyone is ready to do a study of that nature (and magnitude), I'll be first in line to read the results. To date, there is nothing, and I mean nothing, that I can find in the peer-reviewed literature to support the DNA test for 'breeds'.

For purebreds, as I stated above, there is more accuracy, about the same as doing a parentage DNA test, which is also not 100% accurate because science doesn't work that way.

Ultimately though, all dogs have virtually identical DNA, just like humans, and the fact that you have a dog is more important than what 'breed' he may be.


True Selma -- but dogs are VERY different from humans. Humans have evolved over time via small changes and morphs in DNA -- ie natural selection. However, dogs have changed via unnatural selection -- ie, for a very large part of the dog population, they were selected for breeding based on their genes -- and these genes were not small changes in DNA, but in fact, the largest, most visual ones people could select for. This makes the genetic markers much more clear, than say, being able to determine that I'm 1/2 German.

This would no doubt make testing mongrel dogs from many generations of mongrel dogs nearly impossible because they would have evolved form natural selection too -- which is why a significant number of the tests come back with a certain percentage of the dog being "unable to be identified".


Actually, dogs aren't very different from humans, they are among our closest relatives, DNAwise.

If you let dogs out and let them interbreed, within a few generations (five to seven, which is about 3 to 4 years) most of the dogs will revert to a pariah type - think ACD as a good generic type - and all the funny hair and odd shapes will be gone.

There isn't a lot of difference between them, except for some superficial qualities. That's what the science says to date.

Again, if there is some peer-reviewed evidence out there supporting the DNA testing of mutts, I really, really, want to see it. Obviously, for BSL it would be wonderful news since most of the so-called 'pit bulls' around here can't possibly be related to what are, in Canada, extremely rare breeds (APBT, AST, SBT). More rare than my breed, for example. So that would be great news for dog owners who are willing to go to court.

Once we have some actual evidence, which we don't yet, then maybe the DNA My Mutt cos will actually testify as experts and stand behind their results, which they don't now.


PS If you have a breathalyzer test, the results stand up in court and the company stands behind the results.

If you have a radar test, ditto.

If I go for a diagnostic test, the results are considered accurate but of course not 100% accurate and the companies that manufacture the equipment, and the doctors who interpret the results, stand behind that.


"Validation testing has resulted in an average accuracy of 90% in first-generation crossbred dogs of known parentage. The breeds in this validation study represent 45% of American Kennel Club (AKC) registrations. Accuracy was determined by average positive predictive value across breeds studied."

Q re: Can ACOs, etc, use the test to determine breeds...
"Wisdom Panel™ Professional is designed and intended to be used solely to identify the genetic history of a mixed-breed dog and no other purpose is authorized or permitted."

Translation, this test will not meet judicial standards.

DNA my dog:

"Can you confirm that my dog is pure bred? No. The DNA Breed Identification test is designed for the sole purpose of identifying breeds found in the genetic composition of mixed breed dogs. If only one breed is detected, it could be that there there is DNA present from another breed, but in amounts too small to be detected."


I think it's possible that DNA assessments could be useful when it comes to determining type when there's a dog bite/attack sufferer, as someone mentioned. Supposing the analyse is precise, of course.




Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment

Your Information

(Name and email address are required. Email address will not be displayed with the comment.)