Pet DNA Testing Company Mistakenly Identifies a Human as a Dog (theguardian.com) 74
"A pet company has twice sent back dog breed results for human swab samples," reports the Guardian, "prompting doubts surrounding the accuracy of dog breed tests."
On Wednesday, WBZ News reported its investigations team receiving dog breed results from the company DNA My Dog after one of its reporters sent in a swab sample — from her own cheek. According to the results from the Toronto-based company, WBZ News reporter Christina Hager is 40% Alaskan malamute, 35% shar-pei and 25% labrador.
Hager also sent her samples to two other pet genetic testing companies. The Melbourne, Australia- and Florida-based company Orivet reported that the sample "failed to provide the data necessary to perform the breed ID analysis". Meanwhile, Washington-based company Wisdom Panel said that the sample "didn't provide ... enough DNA to produce a reliable result"...
The global dog DNA test market, which was valued at $235m in 2022, is projected to grow to $723m by 2030, according to Zion Market Research. The industry's main players include DNA My Dog, Orivet and Wisdom Panel, among others.
But faulty results have cast doubt on the accuracy of the DNA tests.
Thanks to jd (Slashdot reader #1,658) for sharing the article.
Hager also sent her samples to two other pet genetic testing companies. The Melbourne, Australia- and Florida-based company Orivet reported that the sample "failed to provide the data necessary to perform the breed ID analysis". Meanwhile, Washington-based company Wisdom Panel said that the sample "didn't provide ... enough DNA to produce a reliable result"...
The global dog DNA test market, which was valued at $235m in 2022, is projected to grow to $723m by 2030, according to Zion Market Research. The industry's main players include DNA My Dog, Orivet and Wisdom Panel, among others.
But faulty results have cast doubt on the accuracy of the DNA tests.
Thanks to jd (Slashdot reader #1,658) for sharing the article.
Obviously just another scam (Score:2)
No surprise, really.
Re: Obviously just another scam (Score:5, Insightful)
For that first one, yeah, likely a scam. However:
Hager also sent her samples to two other pet genetic testing companies. The Melbourne, Australia- and Florida-based company Orivet reported that the sample "failed to provide the data necessary to perform the breed ID analysis". Meanwhile, Washington-based company Wisdom Panel said that the sample "didn't provide ... enough DNA to produce a reliable result"...
This is believable. They're probably using chips that are specifically made to detect specific genetic markers, and they just didn't find any of them because humans have few if any of them at all. This is pretty much the only practical way to do what they're being asked. Anything else is likely expensive to the point that nobody would bother.
Re: (Score:1)
One type of DNA test is basically just "see where the DNA sticks". Stick your targets on a surface, and see where your sample sticks best (closest similarity). Given that human genome has spent the vast majority of its time as the common ancestor between humans and dogs, it should be no surprise that there's significant similarities. The ancestry of many genes is older than the split between humans and dogs.
Re: (Score:2)
Basically one of these:
https://en.wikipedia.org/wiki/... [wikipedia.org]
These are cheap when built at a large scale. Every other method I'm aware of is a lot more expensive. I don't know the exact methodology here but for dogs I suspect they'll need to PCR first as I doubt you'll be able to convince a dog to spit in a tube.
I'm not sure what you mean about the similarity you're talking about, we share less than 25% of our genome.
Re: (Score:2)
How much of our genome we share depends on how you're measuring differences. You can measure per character, per mutation, per gene, all with or without counting non-coding DNA, all with or without counting irrelevant changes. By one measure, we share 84% DNA with dogs [slashdot.org].
Re: (Score:2)
It's more difficult than it sounds (Score:5, Insightful)
A good analogy are the old statements that "humans are 99% chimp", and similarly "humans are 90% banana". The genetic similarities between very different species are profound. To tell one breed of dog from another - where of course the chromosomes are the same - you need to look at certain highly variable regions of the genome. The problem though is that those same highly variable regions exist in our genome.
So what could they do differently? Well they could add a few more control reactions to their sequencing to try to rule out errant DNA. They were operating under the assumption that people were sending in only dog DNA, and now we see what happens when something else goes in. The real challenge though is what to do if you get a sample that has some of each - which could easily happen if a dog owner with a dirty house collects a dog sample in a cavalier manner and ends up sending in some of their DNA along with the DNA of their dog. It appears the company built their method without a terrible amount of concern for that either.
Re: (Score:2)
A better test would be to submit a consistent sample to all different testing companies, multiple times.
Test for consistency across companies, and inside one company.
Those that can't even do the latter truly deserve to be closed down.
Re: (Score:2)
Re: (Score:3)
Yes. This is like people who put peanut butter or other substances on their COVID tests and then complained that they got odd results so the tests are crap.
Re: (Score:3)
But his entire post is just him making up an explanation unless he knows exactly how they're performing tests. Even if what he says is correct, it still proves that their results are garbage and wrong. If it could determine human DNA this badly, then it's probably getting dog DNA just as wrong and just guessing breeds.
The other companies could not determine a breed. This company was able to determine 3 breeds, all incorrectly.
Re:It's more difficult than it sounds (Score:5, Informative)
I don't work for any of these dog sequencing company, but I know more about Next-Generation Sequencing (NGS) than most people. I've seen more DNA sequencing formats and done more DNA assembly than most people as well. There are only so many ways they can do this type of DNA work at a scale that makes it possible for them to turn a profit at the prices they charge. If they had some completely novel new DNA sequencing technology they'd be selling it to research labs and hospitals, not selling it on the cheap for people to identify the parentage of their pets.
Re: (Score:2)
But his entire post is just him making up an explanation unless he knows exactly how they're performing tests. Even if what he says is correct, it still proves that their results are garbage and wrong. If it could determine human DNA this badly, then it's probably getting dog DNA just as wrong and just guessing breeds.
The other companies could not determine a breed. This company was able to determine 3 breeds, all incorrectly.
How do you know they determined the breeds incorrectly? Seems to me that this company is just going the cheapest route and analyzing only the breed portion of the DNA, since more sequencing costs more, and if someone is sending in human DNA they really don't care. In short, if some bits were flipped in her DNA and she was a dog, that's the breed mix she would be, based on the part of DNA that maps to dog breeds.
Re: (Score:2)
So instead of humans and dogs having a common ancestor, people are descended from not just wolves but multiple discrete breeds of dogs?
What happens on the farm stays on the farm, I guess.
Re: (Score:2)
Unless you're trying to say that wolves already had most of the genetic variances that make up the dog breeds before being domesticated and that those exact same variations are hidden but present in several layers of common ancestor but even that seems pretty strange. Dogs have had plenty of time to develop unique mutations more than being just selectively bred.
Re: It's more difficult than it sounds (Score:1)
All living species have overlapping DNA, whether accidental or common ancestry but especially comparing mammals to mammals, you share tons of DNA with animals. Like 99.9% with other primates and 96-99% with most other mammals.
These DNA testing companies donâ(TM)t do whole genome sequencing (which could be gigabytes per sample), they only take the âoeinterestingâ parts and then do some statistical matching with prior results. It is entirely possible that you have genes that if transplanted (eg
Re: (Score:2)
You're basically trying to claim that despite diverging long before, we developed genes that don't just code for similar traits but share genetically similar sequences to mutations of wolves through the millennia.
Re: (Score:2)
Or, put another way, you are calling them "interesting parts" because they are areas of notable mutation/variation. They would be the least likely sequences to be similar to human DNA.
Re: (Score:1)
It could be. It could be random chance, but wolves (dogs) have co-existed with humans for hundreds of thousands of years, both species have evolved traits that tune them to each other. Ever wonder why dogs barking wakes you up in flight-or-fight mode but other animals (deer, owl, frogs, cats, despite often being closer and louder) just annoy you if they wake you at all? It's genetically encoded that you are alerted to dogs barking similar to crying babies, because our ancestors that didn't ended up not surv
Re: (Score:2)
Convergent evolution doesn't create matching DNA sequences for a trait.
Re: (Score:2)
> people are descended from not just wolves but multiple discrete breeds of dogs?
No, it's just that these folks missed the obvious conclusion that the reporter was a werewolf!
hawk
Re: (Score:3)
Sure these test *results* are garbage. But it doesn't mean the testing *protocol* is garbage, since that testing protocol was designed to work on a different kind of sample.
In other words, think of any test result as being a statement of Bayesian probability: "Given that the sample is from a domestic dog, we can say with reasonable confidence that it is around 40% malamute." You could easily modify your test procedures to exclude human DNA samples, but why would you do that?
Re:It's more difficult than it sounds (Score:4, Informative)
It's not that hard, actually. 99% of DNA is the same, but how the DNA is arranged is pretty unique per species.
Dogs have 78 chromosomes - 38 from each parent plus the two sex chromosomes. Humans have 46 chromosomes - we get 23 from each of our parents (1 sex chromosome from each).
Surely a basic DNA test would at least check the number of chromosomes matches up.
It's why certain genetic diseases in humans can't be found in dogs exactly - the DNA that is problematic would exist in a different chromosome on a dog.
CBC Marketplace did such a test nearly a year ago... and yes, they even submitted human DNA as well. Quite a few of the tested companies did detect it as "non dog DNA".
https://www.cbc.ca/news/busine... [www.cbc.ca]
Re:It's more difficult than it sounds (Score:4, Informative)
It's not that hard, actually. 99% of DNA is the same, but how the DNA is arranged is pretty unique per species.
Except that they aren't doing full genome sequencing (which is vastly more complicated and vastly more expensive). They are sequencing only specific regions of the genome. It would be similar to comparing the Bible to the Koran based on how many times they use the word "Thou". In the end you'll know they're both books and they're different but you won't know the chapter counts or the year of publication.
Surely a basic DNA test would at least check the number of chromosomes matches up.
Not necessarily, and for more than one reason.
It's why certain genetic diseases in humans can't be found in dogs exactly - the DNA that is problematic would exist in a different chromosome on a dog.
That doesn't apply here though. Sequencing technologies are not biased towards or against particular chromosomes, and the chromosomes are not sorted out before sequencing. The whole sample goes in and primers bind to anything they have affinity to. Sequencing then proceeds regardless of whether it starts on chromosome 4, 16, 21, or some other chromosome entirely - as long as the start and end are on the same chromosome. And if you're looking at variable regions within genes, the likelihood of those starting and ending on the same chromosome is exceptionally high.
CBC Marketplace did such a test nearly a year ago... and yes, they even submitted human DNA as well. Quite a few of the tested companies did detect it as "non dog DNA".
Which may just mean that the other company had included some additional tests to look for "non dog DNA", and this company did not. That's a smart control that this company should have thought of, although depending on the scenario it might only tell you about contamination, not complete substitution.
Re:It's more difficult than it sounds (Score:5, Funny)
BTW I checked into it and humans are only 17~60% banana depending on the criteria used:
https://lab.dessimoz.org/blog/... [dessimoz.org]
Re: (Score:2)
Re: (Score:2)
Are you happy to see me or are you 17~60% banana?
Re: (Score:2)
The real challenge though is what to do if you get a sample that has some of each - which could easily happen if a dog owner with a dirty house collects a dog sample in a cavalier manner
You can't rule out one strand vs another because it's already fragments at this point in the process, right? But the quantity of human or non-dog detections vs all other genetic material classified would give you a relative confidence level or contamination level, which you would send back with the results.
I think Christina Hager might be my ex... (Score:1)
Re: (Score:2)
Re: (Score:2)
Are you a furry?
On the Internet, nobody knows you are a dog (Score:3)
So... (Score:1)
Man, that's (Score:1)
..ruff
Preposterous! (Score:2)
"According to the results from the Toronto-based company, WBZ News reporter Christina Hager is 40% Alaskan malamute, 35% shar-pei and 25% labrador."
She's obviously 32% Poodle, 28% Golden Retriever and 40% Shih Tzu.
Ruff! Woof *BARK* WOOF! (Score:2)
Ruf. Ruf ruff! WOOF! RrrrrrrrRUFF! *angry paw shake* RUFF RUFF ruh RUFf, WOOOF *more paw shaking*
It would seem the dog took offense at being mis-specied.as human.
Man. *anything* today creates outrage! Even the dogs are riled up and angry!
Obviously some people have disposable income (Score:2)
"The global dog DNA test market, which was valued at $235m in 2022, is projected to grow to $723m by 2030, according to Zion Market Research. The industry's main players include DNA My Dog, Orivet and Wisdom Panel, among others."
As a dog owner, I was just thinking about the amount of good things this money could be used for if it wasn't being spent on people wanting to know what breeds their mutts belonged to.
Re: (Score:3)
Finding Your Roots (Score:3)
I just want to hear Dr. Henry Louis Gates, Jr. tell my dog [gannett-cdn.com] that he is a fourth cousin of Lassie.
That's sounds about right (Score:1)
I know some dogs who think^H^H^H^H^Hknow they are people.
I hope they don't offer (Score:2)
It might have been correct (Score:2)
Maybe this was a human that identifies as a dog.
Also, there's that 2002 movie "Bark!" https://en.wikipedia.org/wiki/... [wikipedia.org]!
Or they discovered your mom's terrible secret? (Score:2)
Dog DNA companies (Score:2)
The best one that I'm aware of is run out of Cornell University, Embark, I had them run the test on my shelter dog and the results made logical sense. They also have a health screening additional test.
Most Similar To (Score:2)
This shotgun sequencing probably does a statistical match for the closest known breed-identifying sequences.
The others probably have a minimum confidence level to report 'no result' and this one didn't.
Also, yo mama jokes are obligatory here, so pick up the pace.
Garbage in, garbage out (Score:2)
You submit a DNA sample to a service that tries to match samples to dog DNA, you're probably going to get a match against dog DNA.
Try 23andMe next time.
There's a logical scientific kinky explanation (Score:1)
Alpaca DNA testing is going to be big (Score:2)
you never know if you mistakenly bought a lama that is spitting all over the place.
Re: (Score:2)
The Dalai really doesn't go in for that.
beauty (Score:2)
The real story here is the break through in being able to test for beauty with DNA. Sorry lady, test says you're a dog.
Are we sure? (Score:2)
"WBZ News’ latest report comes after its investigations team sent in a sample from New Hampshire pet owner Michelle Leininger’s own cheek to DNA My Dog last year. In turn, the results declared Leininger 40% border collie, 32% cane corso and 28% bulldog."
How do we know for sure she isn't actually a dog?
Re: (Score:2)
Re: (Score:1)
Have you ever wondered? (Score:3)
"Your puppy is 18 years old..." (Score:2)
"...So it's time to humanely put her down."
Whack that company with a rolled-up newspaper.
Play Stupid games... (Score:2)
I guarantee the Venn Diagram with people paying for genetic testing on dogs and people whining about how many forces of life are persecuting them and/or making them miserable approaches 100% overlap.
It's a damn dog, it eats, it shits, it licks your palm. Any random dog is about as good or useful to you as any other dog. And the best dogs are *generic dogs" not highly bred ones, only fatuous morons want DNA testings As such, you deserve bad service and random results, because it'
Fond memories (Score:2)
This reminds me of the good ol' days when Rob Cockerham sent in spray-painted trinkets [cockeyed.com] to Cash 4 Gold. (warning expired cert).
He also gave out his Safeway card number to see what would happen if you spent $100s of thousands on one card. Not sure how that played out.
I miss the old Internet that didn't even care about SSL, and where pranks involved making companies looked silly instead of sucker-punching people.
Anyway, I hope they keep sending weird shit to companies. We need this.
Who's a good reporter? (Score:4, Funny)
Who's a good reporter?
You are! You are!
Inclusion (Score:2)
Perhaps it was a dog that identifies as human?
The answer you get depends on the question you ask (Score:2)
Why should a company specifically that markets the product for dog testing be expected to put in a lot of effort to filter out non-dog samples?
While validating your inputs is generally a good idea, you can balance that against consequences from not validating which is this case is nothing.
Same thing happened to me (Score:2)
MAGAtard! (Score:1)
These tests are reliable (Score:1)
Re: (Score:2)
We're at the point now where even dollar store pregnancy tests are looking for hCG hormone specifically and not pH.
mistakenly (Score:2)
What does the word mistakenly add to the headline here? Because if a human had been legitimately identified as a dog, it would be big enough news that we'd have all Heard it already...
Perils of machine learning (Score:2)
I'll bet the company is using supervised machine learning in producing its analyses. Supervised ML doesn't classify what something is. It classifies what something is closest to from a range of possibilities. This misunderstanding is rampant in scientific research.
40% Alaskan malamute? (Score:2)
And 100% Loup-Garou.
Here's the answer: (Score:2)
Wolf. Your dog is 100% wolf. It's DNA is expressed differently though.
proves nothing (Score:2)
Think of this kind of testing as taking a vector in a moderately large dimensional space and finding to which of n subspaces it has a maximal projection, where the n subspaces correspond to breeds. If you put in human DNA, you will just find to what breed it's most similar. This really is just a case of gigo.
Oblig... (Score:2)
What bullshit (Score:2)
WBZ News reporter Christina Hager is 40% Alaskan malamute, 35% shar-pei and 25% labrador.
Well, labradors have only been around since the 1800s, so they're hardly distinct.
I've never understood... (Score:2)
the appeal of all these DNA online testing sites. Sure, it's helpful to know if you might be predisposed to a certain medical issue but beyond that, I don't see why anyone even cares.
Also, I really wouldn't want to send my DNA to some random website along with other identifying information. Seems like a wonderful database the government would like to have or maybe your health insurance company. Do you trust these websites not to sell their databases for more cash? I sure as hell don't.
In the "of course!" department of duh (Score:2)