Forum Replies Created
That is intriguing. It would depend on which weight- and size-affecting genetic variants Nigel inherited. In a size prediction model we are working on releasing, Nigel is predicted to be calf-high in height relative to an average person, which is in line with his reported height but doesn’t reveal much about weight.
As the APBT call is quite low (5% is our threshold for likely real, based on mix simulations), there is probably another breed that we cannot call yet in Nigel’s ancestry. The other sub-threshold call that we didn’t report (almost certainly not real) is 1% American Cocker Spaniel. The algorithm is trying to find a match, and I think it’s come up short for that 10%.
That said, the breed standard weights for those tiny breeds are quite low, but from my experience as an owner of a hefty pet Papillon, and a daintier one who has put on some weight in his older age, sometimes the pet population of a breed has more variation in size and weight than the show population.
Michael — While “Shapes & Colors” survey doesn’t cover all color patterns, the “Physical Traits” survey has a question and option close to what Ollie expresses. The “How much white fur does DOG have?” question, option 6 is the right choice for very small amounts of white spotting (like socks or blazes).
Rebecca — Good point! For “Shapes & Colors”, we wanted to keep the coat pattern options in black & white to best generalize the patterns, as there can be a lot of variation in color and pattern combos, but we did miss out on tan points due to this choice.
Sometimes, when we perform genomic studies, we will combine information from multiple survey questions. If we see that a dog has selected black + tan + white in their colors question, and some level of white spotting in the “Physical Traits” survey, then we have a good idea that they are tri-color. It can also be useful to look at these as two separate traits — black & tan coloration and white spotting — because many coat traits can be controlled by multiple genes.
Kate — we are so glad you are enjoying it! Chandler’s unknown portion was comprised of a many tiny fractions of nonsensical breeds, below the threshold for accurate calls. We are indeed suspecting that when this happens, that it may be non-breed or village dog ancestry. One of our next big goals is to confidently call village dog ancestry. The big challenge is that village dogs are not really a breed per se, and they have massive genetic diversity across the world. So, instead of a reference panel, we may look for other genomic hallmarks and machine learning to detect it.
Lee — we love Boone! Right now, I am re-analyzing the MuttMix survey results using the v2 results, now that we can call American Pit Bull Terrier ancestry. Boone, so far, is the best-guessed dog in the survey. His iconic Basset ancestry really shined through. 🙂
Hi Kristen & Bre Anne,
The primary reason for breed results changing were changes to the algorithm we used. A couple other aspects of constructing the new breed reference panel might further explain the new results for Dexter & Dexter:
First, the individual 12 dogs sampled for each breed have changed. Previously, we used publicly available array data for these breeds, for which the genetic data is sparser (~200,000 locations on the DNA) than the data we gain through sequencing (millions of markers). Wherever possible, we substituted in dogs of the same breed that had deeper sequence data available.
Second, the number of genetic markers has increased. When we were limited by the publicly available genotyping data, we previously examined under 200,000 markers for breed ancestry. Now, we have identified over 600,000 informative markers that help distinguish between 101 breeds.
When I take a closer look at Kristen’s Dexter and the old results, I do see that the most confident breed portions came from JRT and ACD (followed by Wire Fox Terrier). I suspect that the non-JRT Terriers — including the current 2% Yorkshire Terrier call — are misidentified portions of Jack Russell Terrier. The Jack Russell Terrier is an especially challenging breed for both the old and new algorithm, because, as a breed, JRTs are exceptionally genetically diverse!
For Bre Anne’s Dexter, these results were certainly affected by the substitutions in the panel. Golden Retriever is a breed with a ton of deep-sequenced data available, so we were able to prioritize representing more diversity within the Golden Retriever breed panel.
Please let me know if I can clarify more with the new results!
Kristen, I’d lean towards “no” because the spots don’t extend out to the tongue. We had the Chow and Shar Pei tongues in mind, but also know many dogs who have distinctively spotty tongues.
I personally think this tongue trait in the survey, oddly, may help us explore the question of “why is there so much Chow in mixed breed dogs?!” (and is all that Chow real?)
We are working on a blog post soon that will explain more about how we tested out the new survey and the status of our current physical traits studies.
Thanks for the feedback, Jennifer!
It’s definitely a challenge to define these traits in a way that’s agnostic to breed-specific terminology. The hope is that, even if not all dogs answer exactly “correctly” with regards to coat color genetics, that across dogs these answers will be sufficient for finding genetic associations. We also encourage using the comment section for answers, because we can also mine that data for further clarification.
( http://www.doggenetics.co.uk/ is a great resource for canine coat color nerds, or aspiring coat color nerds like me )
Light brown is indeed trying to capture “dilute liver”, so that would cover “lilac”, “isabella” dogs — yes, best exemplified by Weimaraners!
Pink is a funny one. I threw in there in case a dog’s coat is so sparse or white that the owner mostly sees their skin tone, or if they’re largely bald like a Crested. I’ll be surprised if anyone chooses it but I wanted to provide the option.
I would consider sable, grizzle, or agouti dogs as having black in their coats, but it is up to the owner to make that call. Across dogs, we will hopefully capture enough information to study it.
“None” is indeed a catch-all for none of the above patterns, including solid. I found that in the previous Physical Traits survey, many dogs with ticking or roan patterns answered “merle”. So, I felt it was necessary to have a question to break those patterns up. Some evidence suggests that roan and ticking have a shared genetic basis, and that Dalmatian spots are a modified form of ticking at another locus. This may help us decipher that relationship.
My dog falls into the “freckled but not ticked” category — she has several freckles on the legs, but not across her body. I suspect those who answer Q4 with ticking or roaning may also select this, but I’m curious to see dogs who have very few ticks — closer to freckles.
Oh, that’s very interesting and I’ll admit, we didn’t realize this possibility. I suppose comments on the answers may help pick these out.
Of course! These are important questions for ancestry testing in general, too.
Yes, the implication is accuracy. It depends on multiple features: number of markers, # of breeds, # of individual dogs in the panel, and algorithm. As Darwin’s Ark and Embark differ on all of these features, a direct comparison of what feature matters the most isn’t possible.
But! We have ways to test what features make a difference in accuracy. We perform computer simulations to make virtual mutts, where we know exactly the family tree and how much DNA they inherited from each breed. We can then vary the marker number, breed number, algorithm, etc. to see how accurate the results are compared to what they should be.
For grandparents, ancestry is usually estimated from the tested dog’s overall percentages, assuming a quarter from each grandparent (which is not completely true). If the pup is roughly 22-27% Golden, then it is a good guess that they have a Golden grandparent. Without testing the parents or grandparents directly, we cannot find their ancestry.
If one did test the grandparents and found 95%, that could still be interpreted as full-blooded Golden. A unique lineage from the UK, for instance, might be inherently 5% different from all the U.S. Golden Retrievers in the reference panel, and that lineage difference may be responsible for the 5%. Even show populations versus working dog populations may have essential differences despite being the same breed. There is no way to catalogue every Golden Retriever lineage but the more individual dogs, the better.
For Bee’s results, I would guess the biggest difference is the sheer number of purebred Golden Retrievers she is compared to in Embark (publicly available data plus all new Goldens in their database) versus the 12 purebred Goldens (publicly available data) she gets compared to here.
As Jessica mentioned, it may be the individual Goldens we have in the reference panel (12 dogs) don’t represent all lineages of Golden out there. For the new breed panel (v2) that I am constructing and testing right now, we have a different selection of 12 Golden Retrievers (who had deeper sequencing data), so the percentage calls on Bee might shift again when we re-analyze Bee’s DNA with v2 of the panel.
In addition to a wider panel of breeds, Embark may have more than 12 individual dogs / breed represented in their reference panel. For every new purebred Golden Retriever that genotypes through Embark, they might further expand the panel, thereby representing more variation across Goldens.
We also use a different type of machine learning algorithm than Embark, which may introduce some differences. Our method is very similar to the approach that the human genetics test 23andMe uses. In the long term, I’d like to do some comparisons between ancestry calling methods and see what works better on mixed breed dogs.
Another thing to note is that even siblings may inherit different proportions of their grandparent’s DNA. When genetic material is contributed from each parent, the DNA from both grandparents has been recombined and randomly assorted. So, the exact percentages from different breed ancestries might still vary a bit between siblings.
As Brittney mentioned, American Pit Bull Terrier is not currently in our panel but will be added soon. I am right now validating and finding the ideal parameters for a new breed reference panel that looks at more genetic markers and includes several new breeds, namely: American Pit Bull Terrier, Bearded Collie, Belgian Malinois, Chinook, English Shepherd, Entlebucher, Saluki, and Tibetan Mastiff.
When the panel is updated and your dog’s results are re-analyzed, you will receive an email notification.
Thanks for participating!
Yes, most of the eating questions come from Dr. Eleanor Raffan’s questionaire. Her lab focuses on metabolic diseases in dogs but they also published an interesting paper on food motivation and obesity in Labrador Retrievers — and how the genetic region associated with food motivation is more common in guide dogs. There can definitely be some interesting overlap with feeding and behavior!
Other feeding questions we developed alongside (human) food allergy researchers, as we received funding from the Food Allergy Science Initiative to examine food allergy genetics in dogs.
Thank you for yours and your dog’s participation!
Many of our survey questions also come from published and validated surveys, particularly:
The Dog Personality Questionnaire (DPQ)
from Jones, A. C. Development and Validation of a Dog Personality Questionnaire. (University of Texas, 2008).
The Canine Health-related Quality of Life Survey (CHQLS)
from Lavan, R. P. Development and validation of a survey for quality of life assessment by owners of healthy dogs. Vet. J. 197, 578–582 (2013).
The Dog Impulsivity Assessment Scale (DIAS)
from Wright, H. F., Mills, D. S. & Pollux, P. M. J. Development and Validation of a Psychometric Tool forAssessing Impulsivity in the Domestic Dog (Canis familiaris). Int. J. Comp. Psychol. 24, (2011).
The Canine Cognitive Dysfunction Rating scale (CCDR)
from Salvin, H. E., McGreevy, P. D., Sachdev, P. S. & Valenzuela, M. J. The canine cognitive dysfunction rating scale (CCDR): a data-driven and ecologically relevant assessment tool. Vet. J. 188, 331–336 (2011).
The Certified Dog Trainer Exam (CDT)
from the International Association of Canine Professionals
The Dog Obesity Risk and Appetite questionnaire (DORA)
from Raffan, E., Smith, S. P., O’Rahilly, S. & Wardle, J. Development, factor structure and application of the Dog Obesity Risk and Appetite (DORA) questionnaire. PeerJ 3, e1278 (2015).
Hope this helps!
The genetics behind dog behavior remain an open field, and genes involved in behavior aren’t as well-profiled as, say, coat color genetics… but that’s what makes behavioral genetics very exciting, too!
I might suggest reading about this study on Labrador Retrievers for a good example of behavioral genetics:
The original scientific article is here:
In short, Labradors with deletions in this POMC gene have higher rates of obesity (health and behavior-related) *and* Labradors with this variant are overrepresented in service dogs (behavior-related). The hypothesis behind this being that POMC influences food motivation, and that dogs who are more food motivated will work harder for a reward (great for training!)
Those 2-3% portions of your dog’s DNA might in a region of the DNA that influences those traits — body shape and personality — though there’s still much to discover.
Scientists use an approach called “ancestry mapping” to find those regions that influence traits based on ancestry. We will be using this approach to study behavior in a few ways; for example, studying retrieving behavior by looking at Poodles, Retrievers, and a variety of “-doodles”.
Say there is a part of the DNA that influences whether a dog fetches. A 4/5 Poodle & 1/5 Golden Retriever might be a fetching fanatic because they have retriever ancestry at that place in their DNA; likewise, a 1/5 Poodle, 4/5 Labrador might not be into fetching, just because they don’t have retriever DNA at that spot. This is what we hope to find out using ancestry mapping!
As for DNA being static — yes, and no. The genetic code of DNA is very stable, but many traits are not dominant-recessive and depend on many parts of DNA as well as environmental factors.
Great question! This gets at a very important aspect of the genetics of health: breeding for improved health requires genetic diversity from which to selectively breed, but genetic diversity alone, in an individual dog, doesn’t determine health.
Our current measurement of genetic diversity isn’t a perfect proxy for “inbreeding”. Right now, it measures the differences between the maternal and paternal DNA, from which the dog has inherited one copy from each parent.
We are currently developing a better measurement, like COI. Here is a discussion about COI, and it brings up some points about COI being too low or too high:
I also want to emphasize that geneticists and dog professionals may use the word “inbreeding” differently. Here, it’s a number that describes genetic variation and all individuals have some level of it. For dog professionals, inbreeding refers to specific mating or family tree structures. Of course, the breeding leads to changes in the number, but the relationship isn’t 1-to-1.
Thank you for the question,
Thank you for notifying us about the missing field for New Guinea Singing Dog in the feedback section — I’ll see that this gets updated, as it will be important for recruiting enough dogs to construct and update the breed reference panel.
Right now, we have 5 published samples from NGSDs, as well as 4 other NGSDs enrolled in Darwin’s Dogs (in addition to your dogs).
We aim to have 12 dogs to complete the panel for NGSD, even if they’re descended from a single pair, so that we can sufficiently cover any genetic variation that established throughout the population’s history. Your dogs will help us reach this goal, so thank you very much for your participation.