Friday, February 06, 2009

How Accurate are Dog Breed DNA Tests?


Breed identification DNA tests are a sure-fire way to make money for those who sell them, including veterinarians who are paid for product endorsement.

But do they work?

No. In fact, the results shown here are common: a pure-breed dog comes back as being a vague pastiche of three or four breeds.

Breed DNA tests are not too different from Gypsy Fortune telling, Fortune Cookies, the I-Ching, Numerology and Tarot Card reading: If you give a vague-enough answer, the believers will rationalize whatever result you give them, pounding the square peg into the round hole.

This is especially true for mixed-breed dogs. The folks sending in their dog's DNA for testing here do not care what the answer is, so long as it answers the question. Even an obviously wrong answer gives them a story to tell when someone, inevitably, asks: "What kind of dog is that?"

So what's going on? It's pretty simple: there are hundreds of breeds of dogs, but the DNA tests only definitely ID's a few dozen. The gaps are "filled in" by claiming a dog is a cross of this and that.

But what about the Mars Veterinary WISDOM Panel™ MX that the vets are selling? Surely that veterinarian-administered test works well, right?

Ughh .... NO.

But don't take my word for it.

Mars Veterinary's own web site says the test is pure crap. Or, to be more precise, they say that if you actually KNOW what AKC breed of dog you have (because you have pedigree papers for your dog going back to the start of the registry more than 120 years ago) then they cannot help you.

But if you don't know what breed of dog you have, then they can positively tell you what you have.


Whiskey. Tango. Foxtrot.

You're kidding, right?

Nope. Read it yourself here:

Why can't this test detect purebreds?

The WISDOM Panel™ MX test was designed to determine the breed makeup of mixed-breed dogs. Its development involved the analyses of more than 19 million DNA markers from more than 13,000 purebred and mixed-breed dogs to best tell breeds in a mixed-breed dog apart.

In order to determine if a dog is a purebred, Mars Veterinary would ideally need DNA samples that cover all family lines for each breed of purebred dog. But since their focus was the development of a test capable of accurately determining the breeds in a mixed-breed dog, they did not focus on collecting such a catalogue of purebred dog DNA samples.

What's that mean? Not a damn thing! It's poppycock. It's typing by a monkey. It's stupid on stilts, with a side-order of bunko and larceny.

It's the get-out-of-jail card the company can point to when their DNA "test" is shown to be inaccurate, pure crap, and a complete fraud.

In fact, the question itself is paired with this question: "Is Mars Veterinary worried about lawsuits?" This is their answer: Our test is so worthless we cannot identify your breed of dog if it is pure-bred. Our test only "works" when you have no idea what breed of dog you have.

So, NO: dog breed DNA tests do not work.

Save your money. Or better yet, donate what you would have spent on this near-worthless test (about $150) to the local no-kill animal shelter.


psyquark said...

All genetic testing to determine breed is predicated on the existence genes that are specific to that breed.

If the test picks up the genes for "sludgy piss", its a Dalmatian!

If the test picks up "smashed in face" genes, its pug, boxer or bulldog (or otherwise related)!

If the test picks up "brain to big for skull", is it a Doberman? Nope, its that dog that won crufts!

Now lets look at distinctive traits (genetic) of the dog in question. It appears to be healthy with no obvious faults in construction so nothing there (I would hope that healthy body isn't exclusive to pits these days). It has a black and white coat (All border collies have that coat+pattern, few others do. Pits have many different coats). It has short hair (multiple breeds though mostly associated with boxer and other bulldog derivatives). It has a larger than average brain casing (again common to bulldog derivatives).

Given all that, is the results surprising? Her dog does not have any distinctive genetic faults to find. What genetic marker (read: mutation) is the test key off to select likely breed? Without inbreeding to exacerbate genetic mutations, especially intentional faults, there is nothing for the test to single out as special or unique to a breed. The accuracy of the test is based entirely on the breeding practices which you rail against on this blog!

That being said, I am confident that this company would pickup the breeds or breed groups of my mutt. She has the distinctive mask and stiff triangle ears of a husky type and she has the purple tongue and forward pointing ears of a Shar Pei. That combination of ears results in a look similar to a "crop" but without the exposure of the of the ear canal to the elements that cutting of the ears results in. My girl HAS mutations specific to breeds/groups, her dog does not and most healthy mutts won't either.

Pit bull NM said...

The Coppinger's book, "Dogs", takes an interesting look at why specific genes are difficult to ID in dog breeds: they cite the "founder effect" and selection trends within the show dog fancy as influences that periodically shift gene frequency.

From the book:

"No breed can be ancient in the sense of an unchanging gene sequence. Even within our modern breeds, the gene frequencies are constantly changing, sometimes due to natural selection, sometimes due to artificial selection, and sometimes just due to chance events. The breeders themselves change the gene frequencies in many ways--breeding many females to this year's grand champion creates a founders effect in the next generation, for example."

It'll be interesting to see if the technology ever gets up to speed on this one. That said, I do wonder if many folks are truly looking for insight into their dogs personality/lineage or if most are just doing it for a good laugh?...the world may never know.


AnnB said...

The Daily Dachshund has reported on someone whose purebred Dachshund was identified as a mastiff. Hmmm.

Barb said...

Oh, Oh, Oh... "stupid on stilts"!
My new favorite term! Unfortunately I'm sure I'll have several prime opportunities to use it this coming week.
Thank you for that! And, of course, for all the thought-provoking writing you do!

PBurns said...

A funny story to that term -- I invented it a few years back when I came across a photoshopped picture of a lion on stilts hunting giraffes. It was perfect.

Picture here >>

Still a keeper!


TC said...

Too bad. I have two cavalier king charles spaniels (rescues) and a "found playing in road" rescue. They could not be more different in temperement. I was hoping to find a test that would tell me what Tucker (named after the road she was playing in) was so I could find the best way to train and play with her. The Cavs don't require much. They just like to sit wherever you are. This one (who looks to be some sort of spitz/jack rat mix) will go off to rooms by herself, jump the fence or just run crazy around the house. She is not a food responder and she looks miserable and bored most of the time. I am clueless with her. I was taking her for walks but she goes crazy whenever she sees another dog and she slipped the leash a couple of times, so had to stop. I was banking on identification being the key to targeted training. Oh well, back to the expensive trainer.

Donna0404 said...

I had my shelter dog's dna tested. The shelter had her listed as a flat coated retriever mix. She does look like a black golden ret except she is only 48 lb. I have owned 2 goldens and a lab in the past, but never a flat coat. She seemed to me to have a lot of the golden traits I was familiar with, but has no interest in retrievig so I had her tested. I did not send a photo. She came back no primary or secondary findings. In the mix (approx 12% ea) golden retriever, labrador, bernest mountain dog and scottish terrier. I was suprised by the last 2, but to tell you the truth I don't doubt either one. She displays herding behaviors when we are at the dog park, and loves the winter weather and snow. She also loves to dig in the dirt and snow and races around like a crazy dog for short spurts,often under the tables and in other small areas my retrievers would never have attempted. There is still about 50% of her genetics unknown, and I am trueley considering having it done again with a company that has more breeds. (I went with Heritage who had just over 100 breeds.) Donna

ejeco said...

I wanted to see what breed my rescue dog was and was very impressed with the results from Mars. Pehaps since he was 50% of both breeds detected it made it easier than all the breeds that can be blended together in a dog that has come from mixed breeds over a longer period of time. He follows the personality traits of the siberian Husky while the Boxer explains his shorter coat and broader head and muzzle and larger ears which do not both stand as erect as the Husky.

Pet Geek said...

The cost of these inexact tests is brutal. Perhaps a year ago or more, a woman in the DFW area had to give up her small black and tan dog (20 lbs or less) because a vet-administered test declared it was part Rottweiler, and the apartment management did not allow Rotts or Rott mixes. All I could figure was that the test assigned "Rottweiler" to any black and tan dog. There are any number of small dog breeds with short hair and black and tan coloring: Miniature Pinscher, Manchester Terrier, Dachshund, and Chihuahua come to mind. Which was more likely, one of those or Rottweiler? But a dog was needlessly stripped from its familiar life and a family deprived of its pet because of that test result. Ridiculous and tragic.

equsnarnd said...

@Pet Geek
How exactly did the apt mgr find out?
And since when can't test results be challenged?
I think you have the makings of an urban legend.

Anonymous said...

Oh God, really? These tests are totally crap? I guess there is never ever going to be an authentic way of testing dog breeds. This is really disturbing.

Wisdom Panel Discount Code