View Full Version : Tell me about Doggy DNA Ids

Dec. 20, 2010, 10:18 AM
This came up a little bit in my stray thread but no one really spoke up about it much. How accurate are the labs about giving you an idea of what "flavor" your stray really is? Right now, I have SIX!!!! And I'd like to know what we are dealing with as far as breeds are concerned. TIA.

Dec. 20, 2010, 10:21 AM
We had our dog DNA'ed and was pretty surprised by the results! With that said my little funny looking dog is 20% Great Dane 20% Irish Setter, 20% poodle 20% Shitzu and some Pomerainian in there as well.
I can't see the Great Dane, but see every single other breed in him....

Tom King
Dec. 20, 2010, 11:14 AM
It's very accurate. People that don't know what they are talking about are willing to speak up and say that it's useless, but anyone that does know what they are talking about says it's accurate.

Only a few genes on a few of the 78 chromosomes determine what size the dog is and what it looks like. Most of the other chromosomes just make it a dog.

The majority of different breeds have evolved over a couple of hundred years at least, so even with one gene out of 40,000 mutating, breeds still end up with their own distinct DNA footprint.

Some people who speak against the test say that the databases are not large enough. Actually, it doesn't take a very large database since breeds typically have only anywhere from 2.2 to 4.4 chromosomes available per location.

Also, the people hired to do the initial research were hired by a philantropist, so the typical carismatic leader of a research team, whose most important job is raising money, was not needed. So the position of a spokesperson was not even thought of, and you ended up with a group of very good scientists who were not necessarily the best communicators.

Dec. 20, 2010, 11:15 AM
My vet put several of her dogs in and she knew the breeding and they were all wrong.

Tom King
Dec. 20, 2010, 03:17 PM
Interesting. Your vet's dogs' DNA matched current breed specific data bases, such as AKC's, and Mars found 0 out of 78 matching chromosomes?

Auventera Two
Dec. 20, 2010, 03:40 PM
A few years ago there was a thread on a dog forum I used to read about this. Apparently some people on that thread had sent in results of dogs they knew for a fact the breeding on, and they came back as weird mixes, not remotely similar to what the dog actually was. If I recall correctly, there were far more dissatisifed dog owners than satisfied ones. Maybe the tests have improved in recent years?

Dec. 20, 2010, 03:49 PM
There are various places to send your DNA for your dog, and varying degrees of capability in those places. I think there was an article in Whole Dog Journal about sending the same DNA to three different places and getting different results each time.:confused:

Dec. 20, 2010, 03:58 PM
I was hoping the results would be reliable. So there is not 1 that is good? Does UC Davis have a dog ID lab?

Dec. 20, 2010, 04:36 PM
I got 2 kits for Xmas last year.

I have a purebred beagle and a beagle/Shepherd mix.

The results confirmed that.

The way the results came back could confuse people, though (at least with the company I used). They list the DNA results in order, and so it may come back that somewhere in your purebred doggie's DNA there is a smidgen of something else. Which isn't exactly surprising since many breeders crossbred at some point. The AKC hasn't been around forever.

I don't know what sort of quality control there is so I guess it is possible for customer samples to get mixed. It's also possible for a person to collect the sample improperly which will affect the results.

I thought it was a cool gift.

Dec. 20, 2010, 04:37 PM

Tom King
Dec. 20, 2010, 04:53 PM
A lady came into our Vet's to get puppy shots while we were there once. She had them in a covered box. The receptionist asked her what she had. I was assuming the question was about boys and girls. The lady said,"She had two Jack Russells, a Boston Terrier, and a Cocker Spaniel." I looked and sure enough that's what they looked like-all in the same litter.

There are more self proclaimed experts than people who know what they are talking about. AKC even says that their registry is "98 percent accurate". The only way you can tell for sure is if every mating dog has their DNA on record, and then check the offspring. AKC's database is not the largest one because they started keeping it only for frequently used Studs when it became obvious that some people were selling dogs as a particular breed because of the way they looked. Apparently there was a lot of selective breeding going on from different directions and puppy mills filling out paperwork to fit appearances.

A lot of the time people "know for a fact" that a dog is a particular breed because of the way it looks and what someone told them.

Desert Topaz
Dec. 20, 2010, 06:04 PM
I just got my hound mix DNA tested. Didn't get anything really definitive back, which doesn't surprise us terribly. The company we used had "primary," "secondary" and "In the Mix" results. No primary for her, Border Collie as a secondary and Rottweiler and Italian Greyhound in the mix. The most guessed breeds for her are Blue Tick Hound and German Shorthaired Pointer and now we know she's neither of them. At this point, we figure she's a mix of breeds they don't have in their database, like Black Mouth Cur and Mountain Cur, which are my farrier's guesses.

Dec. 20, 2010, 10:39 PM
I've never done it so can't say anything for the accuracy, but do know one rescue group who had a vet go into an animal shelter and DNA test a dog because it was an owner turn in and as a side note the owner told them the dog was half wolf... which of course they can't adopt out. Rescue group leader really wanted to save the dog and could tell the dog wasn't half wolf, but had to get proof before they'd let her take the dog. Dog ended up coming back husky/malamute (which is apparently exactly what he looked like).

I could see how purebred dogs could come back as mixes though.. many breeds are often based on a couple of breeds. Based on how long the breed has been around some of the other breeds could possible pop up. I have a pembroke corgi... "Pembroke Welsh Corgis are thought to be descendents of: Swedish Vallhund Dogs, Schipperke, Pomeranian, and other Spitz-type dogs such as the German Shepard and Akita". Of which I'd own maybe one of those dogs, but I'd totally have a herd of corgis if hubby would let me :D

Dec. 20, 2010, 10:46 PM
That is priceless!!! Ok.....I'm done. I'm just gonna guess......it seems that is what they are doing!!!!

So....Mollie is a pitty, brown something that LOVES the pond.
Duece is a HUGE black pitty .....maybe something else ?????
Penny is a medium haired black dog with dalmatian legs (really!!!!)..... sweet baby!!!!
Dash is a medium haired black sausage doggy with a large dose of hound...oh yeah that noise is a hound!!!!
Tres is a 6 month old black bull dog, great dane, mastiff looking thing that showed up with a purebred Rhodie...yep a purebred Rhodie.
Chewy is a HUGE retreiver looking long haired bull dog mastiff thingy.

Please......... no more!!!

Dec. 21, 2010, 01:09 AM
I consider them about as reliable as a pet psychic.

Dec. 21, 2010, 09:41 AM
My hubby just bought me a kit for a Christmas present.

Then I looked online and found some pretty terrible reviews so I am skeptical. The company also filed for Chap 11.

I'm not putting any hope in it, because on their website, none of the breeds that my dog is presumed to be, are on their list. Sigh.

Tom King
Dec. 21, 2010, 09:52 AM
I guess at some point of mix, they can't figure it out, and at that point it doesn't really matter anyway, but as far as verifying parentage for some number of generations there is no reason the throw it all out.

Also, if the mix looks like a cross between a Bassett and a Beagle and the test comes back Bassett, Beagle, with other stuff like Italian Greyhound, it doesn't mean the whole test is worthless.

Dec. 21, 2010, 03:07 PM
I consider them about as reliable as a pet psychic.

Very true.

One of my best friends has her DVM and a PhD in genetics; she laughs at those tests, especially the "cheek swab" ones. They're completely worthless.

Tom King
Dec. 21, 2010, 04:09 PM
Every DNA test I've ever seen for dogs was by cheek swab. We probably have one of the most tested packs of breeding dogs anywhere. They've been tested for all sorts of things including either the genes for curly coat and shorthair (which is actually controlled by a head furnishings gene).

Am I now supposed to believe that no DNA cheek swab test is valid and all the results should be thrown out?

AKC's parentage DNA testing is also done by cheek swab. I've never heard anyone who knew what they are talking about say that the tests are worthless simply because the sample was taken by cheek swab.

Yes, I believe the line is not drawn at a good enough place to say that they just can't tell about breeds in a mix that are apparently a long way away from any kind of purebred, but to me it still doesn't mean that the tests are worthless.

Dec. 21, 2010, 04:20 PM
If you have a mutt, why not love him for the personality he presents, the love he gives, the protection he does willingly? Why does one have to know the cross he is? Some of the best dogs I have had are crosses and mutts, I love them for what they give and don't worry about what they may be.

Instead of spending money on a test that may or may not be accurate or satisfactory, why not spend it on a rescue group as a donation or to an animal shelter or have it go toward neutering and spaying??

Dec. 21, 2010, 05:17 PM
AKC's parentage DNA testing is also done by cheek swab. I've never heard anyone who knew what they are talking about say that the tests are worthless simply because the sample was taken by cheek swab.

I'm assuming that the AKC parentage test is used to confirm that the parents are who the breeder claims? That is, you claim a litter is by Sire X and the test confirms that it is indeed by Sire X? Because that is completely different from testing to determine breed.

There are a lot of things that make a dog "breed true", and of course all of those things are genetic. There is not, however, a "Labrador gene" or a "Great Dane gene", just a lot of different genes which, together, make the dog a Labrador or a Great Dane. You can see why this would make using genetics to determine the breeds present in a mutt a bit of a challenge.

Tom King
Dec. 21, 2010, 05:54 PM
Yes, it's some different but not "completely different". At some point in the mix for the ID stuff it's too much for the software to figure out, but that doesn't mean that the whole process is flawed. While there is not one gene that detemines a breed, each purebred breed is comprised of some combination of Chromosomes that is unique to that breed. These have pretty much been determined and documented. The genes that actually control the way the animal looks and its size are few in number and only on a few of the 78 chromosomes.

Genetic diversity varies from one breed to the next from about 2.2 to 4.4 chromosomes on average available at each location of the 78 that dogs have. A breed can be very old and still have a small genetic diversity like the 2.2. You can have a large population with a small genetic diversity, just as you can have a large genetic diversity in a small population like the 4.4. But any breed has their own version of that genetic diversity and it is by looking at those chromosomes within the breed specfic range that you can determine the breed by. If you look at those numbers, you will see that there are not many to choose from for any one breed. As an example if a breed had chromosome A, B, c, or D available at location 1 and the dog had an X at location 1, it would not be that breed.

It's not even a hard thing to do these days. They've been isolating single genes that do determine one specific trait for some years now and that's a much harder thing to do.

TBs are 5.4 but it came from some number of hundred mares brought to the first three stallions, and those mares were later taken to a few other stallions, so it's quite a bit different than any breed of dogs that were selected at some point from a smaller specific number of animals.

Soon after AKC first started keeping the DNA database, I heard the person in charge of it say that he could guess pretty close which breed it was just by looking at the chart.

The problem with the ID stuff is that they don't know at what point to tell you that they can't tell. I haven't bothered to read any disclaimers that they offer. Maybe there are some. I would think there is some probability factor when numbers for any possible chromosomes gets below some number to be included in a breed specific map.

But for those of us using DNA testing to find sources and to eliminate them for health problems, it's useful stuff. And not only useful for dogs, but since they have mapped the whole canine genome, it's useful stuff for us humans too to look at by comparison. For instance a retrogene of FGF4 was found to be the culprit for dwarfism in dogs, since they knew where to look for the short legged gene, and it was found to happen the same way with Humans.

Dec. 21, 2010, 06:53 PM
From what I understand (I own a mutt, and I completely get the temptation to know what her background is) the canine DNA tests are a fun gift, but should not be taken seriously when applied to mutts because that's just not how DNA works. You need to have something to compare it to; at best, if Pounddog Pickle's DNA comes back similar to Westminster Bubba's 18-generation AKC-registered DNA, you know 1/2 the bloodlines. And that's assuming the AKC breeder was ethical and didn't just register his/her dogs any old way, which has been known to happen. Naughty AKC.

In the same vein, DNA tests cannot prove that a dog is or is not part wolf - which I think is HYSTERICAL; with what they charge for wolf hybrids, that inability to prove the ancestry is like a license to print money.

The last time I looked into this was a few years ago, so Tom might be more up to date. However, I'm curious as to A) how precise you can really be with narrowing down genes for various traits - sure, if a dog has genes for, say, a long coat, pricked ears, a deep chest, etc., that suggests a collie in the background. But those aren't unique to the collie, and with the number of possible ancestors in any given dog, they could come from a combo - a prick-eared terrier, a long-coated cocker, a deep-chested mastiff. And B) would a test sophisticated enough to sort out those genes be available cheaply in kits sold at Petsmart?

Tom King
Dec. 21, 2010, 07:30 PM
To answer your questions:
A) genes that specify a particular trait such as coat type are not breed specific. There are a number of different coat types but any breed that has a coat like another breed has the same gene controlling it even though the rest of the genes on that chromosome might be entirely different. The whole canine genone has been mapped. If you want to come up with a test to look for a particular gene for almost any trait, all you have to do is come up with the money to fund setting up the test. Some things like health problems are controlled by more than one gene, but progress is being made every day.

B) No.

On the breed identification stuff, they aren't looking for specific genes. They are looking at the ID of each chromosome. I would think there would be some probability factor when the number of possibilities for some combinations might cross in the mixes.

Dec. 22, 2010, 01:28 PM
the best company is the Wisdom Panel Insights- fast, long list of breeds, results seem to be fairly reliable. They aren't testing for chromosomes or specific genes, they are testing for patterns of minor genetic polymorphisms that they have determined are specific to different breeds- SNPs. Nothing wrong with the cheek swab method of collecting DNA unless the customer messes up the collection process.
The reason people often think the tests are unreliable is because they really only give usable information if your dog is a mix of two purebreds. Many mutts haven't a trace of purebred in them, or if they do it's several generations back. Also people have been shown to be AWFUL at guessing dog breeds based on dog appearance or behavior, so when the test doesn't confirm their guess they decide to claim the test is bad.

Dec. 22, 2010, 03:39 PM
There's some discussion about the tests here....

I would trust the guess of an educated dog person over the results of the test, esp after seeing how the one dog's test came back. He's AT LEAST half golden, probably more. His results? A mix of Vizela, German Shorthair pointer, and BORZOI. Seriously? I've dealt with Vizela's and GSP's and let me tell you, there's not a TRACE of those dogs in him. And honestly, how many Borzoi's do you see running around loose?

I wouldn't waste my money. If someone offered me the chance to test some for free, sure why not. but I'd never pay for the chance to laugh at crappy results. If you're really set on doing one, I would do the Wisdom Panel over any other ones though.

Vet Tech

Dec. 22, 2010, 04:30 PM
I did the Wisdom Panel on my two mutts just to see how it all worked. It is a cheek swab test.

My dog who visibly appears to be a lab/pit/dane/border collie? sort of mix came back as moderate Beauceron and Chow with no primary breeds. He does have a spot on his tongue, so I can see a tiny bit of Chow in there somewhere, but Beauceron is incredibly random and a dog I've never laid eyes on here before (and I see thousands of dogs a year)...hmm.

My Aussie/retreiver/who knows mix (we met mom, who by all appearances I would call aussie) came back as unknown for primary, with a distant springer spaniel.

It was fun, but they definitely aren't there yet to pinpoint every breed and with pitties being what they are, I don't know how it would ever pick up on them.

I don't think I can post photos here but I would be happy to share shots of my Beauceron/Chow and my distant springer mix for you to see. :)