Elinor Karlsson, PhD: Genetic Testing from a Scientist’s Perspective

Jessica Hekman: Hi friends. Elinor Karlsson is the director of the Vertebrate Genomics Group at the Broad Institute of MIT and Harvard. And incidentally, my boss. She has a deep knowledge about both population genetics and what we call Complex Trait Genomics, or the genetics of hard to pin down things like the risk of developing cancer or a behavior problem. Elinor talked with me about a recent paper. This paper looked for genetic variants associated with diseases in a large number of dogs, from many pure breeds and from mixes. She discusses what this paper found and some ways to interpret those findings. And she also talked a bit about a paper she published recently about the perils of over interpreting genetic testing results. This is a nerdy one, but Elinor is really great about explaining stuff in a way that people without a solid science background can understand.

JH: Elinor, thank you so much for coming on the podcast.

Elinor Karlsson: I’m excited to be here, Jessica.

JH: Yeah. So the first question that I always ask people is about their dogs. So I think you have three dogs of that breed, Domestic Shorthair. Am I right about that?

EK: They are. They’re a bit strange dogs. They’re quite small. And they have pointy ears, and they meow a lot. But you know, other than that, you’re totally on board.

JH: Particularly flexible dogs…

EK: Yeah, exactly. They like to jump and climb an awful lot. Yes. What Jessica is alluding to here is the fact that as a person who is well known for being a dog genomics expert and a dog genetics expert, who’s done this for many, many years now. I always tell people that my dirty secret is that I have never actually owned a dog. And I am a cat person, not a dog person. So I have three cats at home.

JH: And they’re very good cats who frequently come and visit our lab meetings. So the other thing I should reveal is Elinor is my boss, but a very nice boss. So.

EK: Yeah, one of the first things that I learned when I started the dog project that we’re doing, because there’s a lot of dog behavior stuff in it, is I started working a lot with people that work in animal behavior, which was new for me. And the first thing I learned was that cats are much more trainable that we give them credit for and we actually are training them constantly, we just don’t realize that we’re doing this. And so it’s been a lot of fun working with the behavior community, with me knowing the genetics and them knowing the behavior to actually understand what the intersection of those things is.

JH: You have not bad behavior chops at this point with your cats. You’ve trained them to do some pretty impressive stuff. Honestly.

EK: They are amazing cats. And it is an awful lot of fun once you realize what you’re doing.

JH: Yes. And it’s sad that people don’t have the video because the best part of having video calls with Elinor is her enormous cat wheel. And sometimes the cats actually get on it and start running on it while we’re having calls. So that…

EK: Yeah, I have one cat in particular, her name’s Lacey. And she is a cat wheel fanatic. So she uses it pretty constantly. I hear her on it all night. She really loves that cat wheel. So I have it in my living room, of course, because if I put it in the basement, they wouldn’t use it because, you know, they like sharing my space. And so it gets to be in the background of all my work meetings now that we’re all working from home.

JH: So three good dogs in Elinor’s house. Alright, so then, we wanted to talk specifically about this particular paper. I haven’t done a podcast interview before where we have covered a paper. But you and I had talked about this paper when it first came out. So the title of it is “Frequency and Distribution of 152 Genetic Disease Variants in over 100,000 Mixed Breed and Purebred Dogs.” And it came out in 2018. So two years ago, and what was interesting to you about this paper when it first came out?

4:28

EK: So Jessica might remember when this paper came out because there was a fair amount of yelling on my part, not about the paper itself, but about the way it was getting talked about. And so the reason this paper interests me is that there’s some really high quality science in there looking at some really interesting things about dog breeds and dog genetics and things like that. But at the same time, the way it was getting talked about had the potential to really scare dog owners. And suggests that we were learning something that we weren’t actually learning from this paper. And that worried me a lot. 

Specifically, the title itself that Jessica just mentioned, if you think back a couple minutes: “The Frequency and Distribution of…” And then they talk about 152 genetic disease variants. And those three words actually kind of got to the crux of why I was yelling so much when this paper came out. Was that when you know, a person who’s not a scientist hears the word “genetic disease variant” probably what they’re thinking about is a change in the DNA, if they kind of know what a variant is. So basically, you’re thinking about a change in the DNA that when a dog carries that change in their DNA, and the DNA they inherit from their parents, that they’re going to get a disease. That’s why it’s called a disease variant. And that sounds really scary. And it sounds like something that’s definitely a bad thing. But for most of these 152 genetic disease variants that they looked at in this paper, we don’t actually have very much information to suggest that they actually are disease variants. 

They are definitely changes that we see in the DNA of dogs. But the problem is that the dog genome, which is the DNA that kind of is the blueprint for making a dog… It’s 3 billion bases in humans. And in dogs, it’s 2.4 billion bases. So that’s that string of A’s, C’s, G’s and T’s that you see when people kind of write out DNA. And so 2.4 billion bases is an awful lot of letters. And there are changes everywhere. On average, there’s one change every… I think it’s about one change every 2000 bases, if you compare one dog to another dog from two different breeds. If you compare dogs within a breed, I think it’s about one in 1000… If you compare dogs from two different breeds, and then about one in 2000. If you compare dogs within a breed, although I’d have to check those numbers, and nobody quote me on them. But anyway, there’s a lot of changes.

So 2.4 billion letters of DNA, one change every thousand bases, that’s a lot of changes. Most of them don’t do anything at all. Only about 1.5% of the genome actually makes genes. We think that maybe 5% of it, or 6%, or maybe up to 8% of it is doing something. But that means that most of these changes in the DNA are changing parts of the DNA that don’t do anything at all. They’re just variants. And so just because there’s a difference doesn’t mean that it’s doing anything, let alone causing a disease. And so, when you call these things “genetic disease variants” you’re implying that there’s something bad. And that was my concern when I was reading this paper was that we don’t know that. We know that there are changes. We know a few of them might be disease causing, but for most of them, we don’t have a lot of information.

8:06

JH: So that bugged you about the paper. So how did they actually… So let’s take a step back and see how the paper actually works. So what were its methods? Like, how did it go finding these disease variants.

EK: So there’s been a lot of small scale genetic studies in dogs over the years. And I think this is probably where I get asked a lot about it because genetic disease variants are used in dog breeding. And so often people will have done a study where they looked at a disease in a particular breed, and identified a change in the DNA that seems to match up with which dogs are getting sick and which dogs aren’t getting sick. There’s two problems with that. The first is that the studies were often quite small. And so what we’ve learned since then, in genetics, is that you often need very big studies to find genetic disease variants with confidence. And the other problem was that they were doing them in one breed. And so, and they weren’t even often doing them in one breed. Often they did the studies in a single pedigree from a breed. And what we’ve also learned since then, is even when you have a change in the DNA that seems to cause a disease like a recessive disease mutation or something like that, it turns out that whether you get the disease or not, and how severe the disease is, gets influenced by all the other DNA that you carry.

And there was actually a really nice study published by some scientists at the Broad Institute last week about this in people, showing that even for a disease that you think is caused by a single change, everything else in your DNA actually can influence whether you get the disease and how severe it is. And so if you’ve done your study to find this disease variant in one breed or in one pedigree from one breed, you have no idea that that same change is going to cause the disease in any other breed or in a mixed breed dog.

And so, in this study, the strength of this study is the hundred thousand dogs. I mean, that’s fantastic. We don’t usually have datasets of that size in dogs and the authors did a really nice job of sharing the results that they got in the supplemental information from the paper, and it was really well done. And so it’s not the data itself that’s the problem. It’s the assumption that dogs that are carrying these changes in their DNA are carrying a disease variant that could make them sick. And I don’t think in most cases that we know that.

10:37

JH: So what you’re saying is, if we find these disease variants, and we link them to particular diseases in purebreds in a particular breed, and then we go try to look and see the same variants in a different breed or in a mixed breed dog. And we expect to see the same association with the disease in this different breed or mixed breed that you basically what you’re saying is, you’d have to go test it all over again. You can’t just test it in one breed or a few breeds and assume that it’s going to be the case everywhere.

EK: Yes. And ideally, what you would do is you would do a really, really big study, and anybody that knows me knows I’m always talking about really big data sets. This is why one of the projects that Jessica and I work on in our lab group is this project called Darwin’s Ark, where we try to get people to sign up their dogs, because we want really big data sets. Once you get to very large sample sizes, lots and lots of dogs in your studies, you can actually look not just for the one disease variant, but for the other things that are influencing it as well.

And so you might be able to figure out that Labrador retrievers tend to carry a particular version of the DNA in other places in the genome, particular versions of genes that influence the action of that first gene. And you might actually kind of be able to make a more complicated prediction that takes multiple different things into account. And so that’s kind of hopefully where we’ll get to eventually. And it’s also worth mentioning at this point that for many, many diseases, environment matters as well. And that’s never taken into account in these kinds of genetic studies. We know that environment is important, but we don’t know how it’s important.

12:15

JH: So in this study, basically, what they did was they took essentially like a genetic health test panel. So just like you might go send off your dog’s saliva to… In this case it was Genescoper, I think. And get back a list of, you know, “The dog is clear for these things, carrier for these things, at risk (meaning that they’re homozygous for) these other things.” And they basically did sort of that on dogs of a lot of different breeds and mixed breed dogs. And then they reported where they saw these variants coming up positive in these different groups, right?

And so there’s places where they’re sort of like collie eye anomaly in collies. It’s there, you know. It’s not a surprise. But then they also dug down into some other breeds where they’d say, you know, “We hadn’t expected to see this particular disease in these breeds. But we did see some cases of dogs being carriers.” So I guess it might be useful to talk about the difference between being a carrier and being what they call “at risk,” so homozygous versus heterozygous.

13:21

EK: Yeah, I think. So that’s a really good point, Jessica. And I think this is the part that… This kind of study would be really interesting if it included something that… People who aren’t doing this kind of science sometimes just assume they’ve done but they don’t actually do, because it’s really hard to do, which is actually go and look at the dogs and find out whether they’re sick or not. And so rather than just saying a dog is a carrier, or a dog has the disease based on the fact that they carry particular versions of their DNA, to actually go look at those dogs and find out if they’re sick or not. And except for a handful of cases, they haven’t done that at all in this study, partly because it’s just really hard to do, as you can imagine. You know, to go talk to a whole lot of dog owners and get veterinary medical records and things like that.

JH: 152 diseases and 100,000 dogs.

14:12

EK: Yes, it will be an awful lot of work. And hopefully we’ll get there. I mean, I think it’s something that we talk about a lot wanting to get there. And that’s what some of the bigger genetic projects that are out now are trying to do.

So to get back to your question, there seem to be two terms that get used a little bit interchangeably. I think both… I’m not sure in this paper. I didn’t look for this, but it often happens in the kind of dog genetics world as well. And so, as a genomics person who tends to not assume that a change in the DNA is having a particular effect, I tend to use the words homozygous and heterozygous.

And so say that you had a position in the DNA where our mysterious ancestral dog had an A. And then at some point in time, that position mutated in some dogs somewhere and a mutation is not necessarily bad. As I said, most of the DNA isn’t doing anything. So you can have mutations and it’ll be totally fine. But it just randomly happened to mutate to a G in some dog. And so after that any dog that was descended from that dog had the possibility of carrying a G. So if a dog inherited the A version, what I was calling the original version, from both of their parents, from both mom and dad, they would be homozygous A. They would have two copies of A. They’d be AA. And if a dog had a G that they’d inherited from their mom and a G that they inherited from their dad, they would be homozygous G, or GG. And then heterozygous just means that you’re an AG. So you’ve inherited A from one parent and G from the other parent.

So those are the terms we kind of use when we’re just talking about variation in DNA, and we’re not assuming anything about what those variants do. In the world of disease mapping or even just genetic testing, we use these words “carrier” and then “affected.”

16:15

JH: I think they actually said “at risk”, which I did like a little better than “affected.” Yeah. But it might be worth talking through what the difference is between those two terms and why it matters.

EK: So, so there’s different models for disease and one of the models for disease, meaning the way it gets inherited, would be a recessive disease model. So under a recessive disease model, you have a trait where, if we go back to my AG example, A is the ancestral version. It’s the healthy version. And so if you get two A’s, then you don’t have any of the version that causes the disease, and you’re not a carrier, and you’re not at risk of the disease. If you get two G’s, which are thought to cause the disease and is a recessive disease, then you will be at risk of actually getting the disease if you’re homozygous for the G. And so those would be the dogs that they’re calling “at risk” in this paper. Carrier, which is this middle category, are the heterozygous dogs. They’re the ones that have inherited an A from one parent and a G from the other parent. And it assumes that the G is disease causing, but also that having only one copy of the G isn’t enough to make you sick. This is why I don’t like the term “carrier.” Because that’s not something that we usually know.

Most diseases are not simple recessive diseases. In fact, I would say that the number of simple recessive diseases goes down the more we study simple, recessive diseases, because it turns out everything gets complicated when you study it. And so in most cases you can often look at individuals that are carrying one copy of the mutation that we in the past thought were unaffected. And it turns out often they will have symptoms of the disease but not be super affected. And so it just never got picked up by doctors, because it wasn’t enough for them to go in and complain about. But if you go look for it, you’ll actually see symptoms of the disease. So does that make them a carrier or not? And then the same thing in the ones that are at risk.

As you said, Jessica, one of the nice things they did in this paper was they did call it “at risk” and not “affected” because it at least is kind of encompassing that idea that just because you carry two copies of the disease causing variant, you’re not necessarily going to get the disease because environment matters. And also what the rest of your DNA looks like turns out to also influence your risk. So once again, once we start looking more deeply, what we find out is it turns out sometimes dogs that have two copies for some of these disease mutations are totally fine or minimally affected to the extent that they never get diagnosed with anything. And sometimes they’re very sick.

And actually in this paper there was a nice little anecdote of exactly this. So they had a small number of dogs where they actually identified dogs in their data set that had disease mutations and followed them up to see if they actually had the disease. And there was one example in there. I’m just scanning through the paper right now to look for it.

JH: I’m trying to remember. There was MDR1 where the dog was… Are you looking for the heterozygous one? Because there was a MDR1 heterozygous dog, I think.

19:35

EK: No, it was… Sorry, I was just looking through this. Guys I really, for all of our listeners, should have done this ahead of time. No, it was this one. It was the one this dog carried two copies. So they should have been at risk and severely affected according to what we knew about this mutation of a gene of a variant that had been linked to something called spino-cerebral ataxia. And so when they talked to the…

JH: Spinocerebellar ataxia, I think.

EK: See, this is why you’re the veterinarian.

JH: Hi, I’m a vet!

(laughter)

EK: I am not! Thank you! So when they went in to talk to the owner, the guy was like, “Oh, yeah. Here’s some videos of my dog.” And it turns out this dog has this kind of uncoordinated gait that is connected to this disease. But it’s pretty minor to the extent where the owner had never even asked the vet about it. He just thought his dog was clumsy. 

JH: And that the disease… The affected form of the disease included, like seizures and severe sort of problems focusing and things like that.

20:42

EK: Yeah, I think it’s one of those diseases where you see videos of the dog, and they just, they can’t even walk properly. It’s really quite sad. And so it might be that this mutation isn’t fully penetrant. And so some dogs carry it and you never even know that they have the disease, and nobody’s ever actually gone to look for it. Or it could be that they originally identified these mutations, this particular mutation, in Parson russell and Jack Russell terriers. That was the original study. And now they’re finding out that other dogs carry it, which is not surprising at all, because all the breeds are descended from a common dog population. So they almost always… Almost always these variants are shared between different breeds. And so it might just be that once you’re not a Parson Russell or Jack Russell terrier, this mutation just doesn’t have the same severity of effect or something like that.

KH: And that would be, again, because of the different, what we call, genetic background.

EK: Yes. And so you can imagine that the reason why dog breeds exist is that you have these populations of dogs where the dogs are all much more closely related to each other than random dogs. So they share a lot of their DNA. So maybe there’s something in the DNA of the family of dogs that makes up the Jack Russell terriers, that means that a Jack Russell terrier is more likely to have another gene that makes this particular mutation more penetrant. Sorry. You know, have a bigger effect than it would in another dog. I am totally making this up at this point in time. It’s just kind of the different possibilities. Once you think through, you know, they have this one observation of this dog that had the mutations but didn’t really have a clinical disease. And these are the different reasons that I could think of for why that might be true.

JH: Yeah, well, and it calls back exactly to that paper that you talked about that came out of the Broad (which is where Elinor and I both work) last week. I read that paper yesterday, and I was so excited to read it because I thought this is going to have so much relevance to genetic testing and dogs. Showing that you can find a variant that seems to have a really large effect, like we’re talking about here for spinocerebellar ataxia, but then it only seems to really have that effect in certain populations.

And it turns out that the reason why is that there’s a lot, a large number of other genes that have to look a certain way, have certain alleles, in order to sort of either turn it on, if you want to think of it that way. Or allow it to have its effect or not compensate for it, or however you want to think of that. And so that in some populations, what Elinor’s hypothesizing is that in Jack and Parson Russell terriers, there’s a lot of genes that are sort of all set to be a particular way to allow this variant to take effect, or to not prevent the variant from taking effect. But when you are in a different breed… I mean, this is what a breed is, right? Is that a lot of stuff is all sort of fixed to look a particular way or be very likely to look a particular way.

23:38

EK: Yeah, and I think this… Actually, Jessica, that was very nicely explained. Thank you. It’s always hard to figure out how to actually… To kind of explain these probability kinds of ideas simply.

But I think this is where I get really concerned about how genetic testing gets used in dog breeding. And this is something that I’ve been concerned about since I first got into doing genetic studies in dogs was that as we find out more and more about genetics, what we find out is that no mutation acts in isolation. You know, if you think of our bodies, or even each cell in our body, as being a really complicated machine of intricate parts, you can imagine that nothing is acting on its own. And so the more we look at these things, the more variants we’re going to find that contribute to some degree of risk of getting a disease. And I got really concerned that if dog breeders went into this with the idea that a mutation was a “bad” mutation, and any dog that carried this variant was a “bad” dog to breed, that they would end up taking a lot of dogs out of a breeding population and creating problems with inbreeding just because they were shrinking the number of dogs that they had breeding in their breed. 

And so I’m not a veterinarian and I’m not a dog breeder, so I don’t have the background to have recommendations about this kind of thing. But it’s something that I’ve always worried about as being on the genetics end of things that as we publish these papers and tell people that things are disease causing variants or disease associated variants, how does that information get used by the dog breeders who are very genuinely trying to make their dogs healthier in a very complicated situation?

25:23

JH: Yeah. And I think there’s another really great example of that in this paper, because they did cover degenerative myelopathy, or DM, which I know is an issue that you covered in your own paper. And one of the things I want to say before I give you a chance to talk about that, because I think you definitely should talk about it, is that in this paper they said that all of the variants that they were going to be talking about were what’s known as Mendelian. And Mendelian is, the term is basically just what Elinor has been saying. The concept is that, you know, if you have… That it’s just the one gene that is controlling the trait in question entirely. So you can just look at, is the individual homozygous for, you know, one or the other version? Or are they heterozygous, and based on that you would 100% be able to predict the trait, in other words, whether they get the disease or not.

And what Elinor is saying is that we, there’s a lot of diseases that we think of as Mendelian, as very simple, that turn out to be a lot more complex. And so I was really surprised when this paper mentioned that they were considering DM, the test that has been found for DM, to be for a Mendelian disease. And I know Elinor will be happy to talk about that. And I’m also trying to think, if I remember exactly what they found in here, but I think really the main thing, which is that they refer to DM as Mendelian at all, which I found surprising.

26:50

EK: Yeah, no, and I’ve actually personally been trying to avoid using the word Mendelian ever since I started realizing how…

JH: Sorry!

(laughter)

EK: No, it’s actually a really good point to make is that it is a word we used to use a lot. And I personally have stopped using it as much just because I have started to realize how much more complicated even diseases that we think of as being simple inheritance actually are. And so I think degenerative myelopathy is a really, it’s a really nice example of the complexities of genetic testing.

So I wrote an article. It’s actually an editorial. So it was an opinion piece with two veterinarians a year ago, year and a half ago, talking about pet genetic testing and our concerns. And it was focused less on the concerns around dog breeders, and more about the question of how genetic testing gets used in a clinical setting. And the reason why we focused on that was that we didn’t think it was getting talked about enough, and that genetic tests in dogs are becoming more and more available. But there wasn’t a lot of guidance for anyone, including veterinarians, about how the information should be used. And we use the DM mutation, the degenerative myelopathy mutation, as one of our examples in that paper, because as I said, it’s a really nice example of both the potential, but also the difficulties, of genetic testing and interpreting it.

So degenerative myelopathy, for anyone who doesn’t know, is basically the dog version of ALS. And so it’s this terrible disease that if the dog starts developing it, they will progressively lose the ability to walk and control their limbs and everything else until they die. And there’s no other outcome. You die from this disease. It’s a terrible disease. And several years ago some dog geneticists did a really nice study where they identified a change in a gene called SOD-1 that was associated with this disease, meaning that they found this mutation that they saw more often in dogs that were affected with degenerative myelopathy than dogs that weren’t. It was a great study, and it was really well done.

But of course very quickly this genetic variant started getting incorporated into genetic tests for dogs. And often what gets missed is this phenomenon, which is basically that even though dogs that carry this particular change in SOD-1 are more likely to get degenerative myelopathy, the variant itself is very common in dogs. And so millions and millions of dogs are running around homozygous for the SOD-1 mutation. They will never get this disease. And so when a person goes into the, you know, into a veterinarian with a genetic test result and they’re like, “Oh my God. My dog has tested positive for these mutations and it means that they’re going to get this disease called degenerative myelopathy.” That’s not true. Most dogs who carry these mutations will never, ever get this disease. And what I think is even worse is we don’t even actually know what the numbers are. Like, I can’t tell you how many dogs that carry these genetic variants in SOD-1 will go on to develop degenerative myelopathy. And to me, that feels like the first thing you should answer before offering something as a genetic test, is what does this actually mean for the people, for the dogs themselves? Are they going to get sick? Because that’s what people want to know.

And so this is kind of why we were writing this article, was just there was nothing wrong with the original research. It was a really nice genetic study that associated this mutation in SOD-1 with this disease. But that doesn’t make it a useful genetic test. And if you have a breed where many, many dogs are carrying this variant, you may not have an easy option of just removing all of those dogs from a breeding population. And so it would be really useful to know what actual influence on the risk of disease this variant has.

I mean, you can do that. So the studies that need to happen, and these are really hard, but they’re totally doable, is that what you actually want to do is you want to go out, it’s called a prospective study, and you want to get genetic information for, you know, 10s of thousands of dogs. And then you want to find out how many of them go on to get the disease. And that will actually, and then you can look at their genetics, and you can figure out how well does having this SOD-1 mutation actually predict which dogs eventually go on to get this disease? But to do that you have to have all of the other dogs. You have to have all the dogs that are carrying SOD-1, but never get the disease, you have to have in your study as well. And so studies that are focused on populations where you have a high risk of disease, they don’t see all the dogs that are carrying these variants, but never get the disease.

31:53

JH: Yeah. You don’t want to do your SOD-1 study only in Bernese Mountain Dogs, because you’re going to get a pretty skewed view.

EK: Yeah, exactly. You want to have as diverse a population as possible.

JH: Yeah, it’s, it can be really hard to know how to interpret these tests. And there’s not a good answer out there right now for owners or for breeders. And it’s scary. I mean, one of the authors on that paper that you wrote had actually heard a story at the clinic where she worked at the hospital where she worked of an owner choosing to euthanize a dog who had some signs that they might have DM, and tested homozygous for this mutation, but elected not to go do more tests to find out if the dog actually had this particular disease just because of the genetic test results.

EK: Yeah, I always like to tell people there’s nothing magic about genetics. You should be just as skeptical about genetic tests as you would be of any other clinical test out there. And as I…

JH: Don’t tell them that! How are we going to get them to give us money if you tell them that?

EK: Oh I know. Because eventually one day… Well, you know, I always think of it. And Jessica, you know more about this than I do. But I think in veterinary medicine if you ordered a blood panel on some things in blood, I don’t even know…

JH: (laugher) Like white blood cell count?

EK: (laughter) Yeah, white blood cell count. That’s a good one. Then when the veterinarian got back the information on what a dog’s white blood cell count was, they would also have a normal range. It would tell you, this is the range you would see in normal dogs. We don’t have that for these genetic tests. Nobody knows what the range is that is seen in normal dogs. And so I think… So this paper is actually a really interesting paper, because I don’t think it’s very good in terms of being able to tell you how common diseases are in mixed breed dogs, because we don’t actually know which of these variants are causing diseases. But it’s actually a really nice study for showing the normal frequency of these variants in a population of 100,000 dogs. Like maybe this is our normal range for these tests. And then we can somehow use that as a way to figure out what’s abnormal. So these are the kinds of studies we need to be doing. We just need to pair it up with veterinary records. So we know who’s actually getting sick.

34:05

JH: And then there was a second story that I… There’s a sort of a second problem with encountering a variance that is associated with a disease. See how I’m not saying disease variants anymore?

EK: Thank you.

(laughter)

JH: Encountering a variant associated with a disease in one breed and then extrapolating it to another breed or to a mixed breed. And they actually covered this story pretty nicely with von Willebrand’s disease. Do you remember this bit from the paper? Do you want me to summarize it?

EK: I think if I remember correctly, this is the one phenotype, correct me if there are others in there, where they actually had a population of dogs, not just cases. They were looking at a whole set of dogs and figuring out what the matchup was between carrying the, what was thought to be the disease variant, and what their actual clinical values were.

JH: Right. And so there’s two variants known to be associated with von Willebrand’s: von Willebrands 1 and 2. And I believe, although I may have it backwards, I believe that 1 is the one that’s really solid and that we know like, this is the causal variant. And I’ll leave it to you Elinor to explain what I mean by causal variant. But with the other one, we didn’t know whether it was the causal variant or not. And they found evidence that it was not. Because it was useful in one breed, but not in another breed. And so maybe you could sort of explain a bit about what causal variant means. And then about, you could even talk about linkage a little bit, maybe, but you don’t have to use that word.

EK: Stray into all the complicated stuff.

JH: Yes. 

EK: So. So yes. So a causal variant is the word that we use when we’ve done lots and lots and lots of studies on something, and actually shown that a particular change in the DNA is causing the trait or disease that we’re interested in. So there’s a causal link. It doesn’t have to be causing the entire thing, but it has to be that carrying this particular change in your DNA for some mechanistic reason is actually changing, directly changing your risk of disease. And so for example…

JH: It’s breaking this protein. The protein is a little robot in your cell that has a job to do. This mutation has broken the robot and it cannot do its job anymore, and you get sick.

36:20

EK: Yes, or alternatively, we’re now very concerned about what we call regulatory mutations, which is basically it changes the thing that tells the robot when to go out and do its job. So it does it right some of the time, but in some types of tissues it totally fails. And so you have some disease of your liver or something like that, or the heart. So in order to designate something a causal variant, you would need an awful lot more information than we have on almost any dog disease variant, so called disease variant, out there, because you’d actually have to do studies in like cells in a dish to figure out whether if you put that mutation in it caused those cells to actually change in their signaling or something like that. But for some diseases, particularly in humans, we have identified causal variants for diseases. That’s not what in most cases, there are a few exceptions. And it will be really fun or at least interesting at some point in time to go through all of these 152 and all the data behind them. Jessica and I have thought about doing this at some point in time, but it is dauntingly…

JH: We’ll make a grad student do it. I’m thinking Kathleen.

EK: We could probably make a grad student do three before they quit, out of the 152. 

(laughter) 

EK: But so mostly what they’re doing is they’re not actually testing the, they haven’t found the causal variant. They found something nearby. And that’s the linkage thing that Jessica was talking about. And so it turns out that DNA, and some people listening to this might know this, DNA gets packaged into things called chromosomes. So dogs have their DNA packaged into 39 different chromosomes. And those chromosomes get inherited as units between each generation. And so it’s not just that a single base of DNA gets randomly, you know: this puppy gets this one, this puppy gets this one. It’s more that the chromosome gets randomly flipped: this puppy gets this version of chromosome one from mom, and this puppy gets the other version of chromosome one from mom, because mom has two chromosome ones.

Anyway, so because DNA is packaged into these long strings, called chromosomes, the things that are near that causal variant will get carried along with it on the same string of DNA between generations. And so usually what these studies have found, because of the way studies have been done in the past, is they’re not testing the causal variant, they’re testing something nearby that’s in linkage with that causal variant. That works really well, as long as you’re looking in the breed that the mutation, the disease variant, the so-called disease variant, was found in. That works really well. But if you go into a different breed then those blocks… There has been enough time that the strings of DNA have kind of gotten shuffled. And so that variant that was a really nice marker for the disease, whatever the causal variant is, in breed one no longer marks the disease causing variant when you go into breed two. And so even though it was a perfectly good genetic test in breed one, it may not work at all when you go into a different breed.

39:13

JH: Yes, and so we have this assumption, right? That because we found it in one breed, that it’s going to be, or even in two breeds which might be related, that it’s going to be the same in other breeds. But just because you see the variant in another breed or a mixed breed dog doesn’t mean that it… It doesn’t mean that the causal variant which it was next to is there at all. So it might actually mean nothing. Sort of like having ripped a page out of a book, and the interesting piece of information is still in the book, which is on the shelf, and you just have the page. Yeah, I don’t know if that’s a good example. But… 

EK: We really should think of a good analogy for that one.

JH: That would be a good analogy to have. Yeah.

(laughter)

EK: I’m trying to figure out, but then you put the page into a different book, but why would you do that?

JH: Yeah. It’s if you ripped all the pages out and shuffled them… anyways. Um, but (laughter) so given all of that, though, they did find some interesting things. And there were some cool numbers in there. And I was just skimming through the paper to see if I could find those actual numbers that you and I talked about ahead of time. And of course, when you’re actually talking on the podcast, you miss it.

Oh, no. I just found it. So yeah. So they were looking at… So they looked at these 152 different variants which may or may not be associated more or less closely with diseases. And they looked at them in mixed breed dogs and in purebred dogs. And so they found… So this first statement I thought was really interesting. They found that mixed breed dogs were 1.6 times more likely than purebreds to have a disease variant in the heterozygous state. And I was saying to Elinor, that that made me stop and be like, “Wait, what? Like the mixed breed dog shouldn’t be having more.” But then they go on to say that the purebred dogs are three times as likely, is that where it was? Were even more likely to have these variants. Oh, 2.7 times more likely than mixed breed dogs to have the homozygous state. So basically it is actually sort of what you would expect that the individual little variants are floating around in the mixed breed dogs, but they’re not paired up with each other. And the purebred dogs are more likely to have them smushed together.

41:33

EK: When I was first reading this paper that sentence made me laugh, because as somebody who’s worked in dog genetics for a while, I was like, “Well, of course. Why would you expect anything else?” Because the one thing that we know about purebred dogs is that within a breed the dogs are more related to each other. So they’re more likely to share DNA with each other. And when you mix breeds, you get more genetic diversity. And so the mother and the father look less like each other because they’re either two different breeds or they’re mixed breed dogs. And so looking at a particular position in the DNA in a mixed breed dog, you would always expect that a mixed breed dog is more likely to be heterozygous at a given position in their DNA than a purebred dog, who will always be expected to be more likely to be homozygous at a particular position in their DNA because they’re a purebred dog, independent of anything to do with disease variants or anything like that. It’s just like it’s a matter of figuring out what your expectation is on something is the kind of key point here and dog breeds are interesting populations.

42:42

JH: Yeah, and remembering that we don’t actually expect mixed breed dogs to have no disease variants at all. They may be in there. But they’re not pairing up with each other.

EK: Yeah, I think one of the really interesting and common misconceptions of dogs is that the breeds are much more different than they actually are. As far as we can tell dog breeds, for the most part, are kind of less than a few hundred years old. There were probably informal dog breeds or landrace breeds or things like that around before that. But this kind of idea of closing a population of dogs and only breeding within that population is really something that’s only been around for a couple of hundred years. And so all of the dogs before that were kind of like this big dog population. They’re just dogs. There’s a lot of dogs that… Most dogs in the world today are still just dogs. And I find it really interesting that I can’t find a word to explain what a “just a dog” is. It’s a dog that doesn’t have any breed ancestry. It’s not a mixed breed dog. It’s not a purebred dog. It’s a dog.

JH: We do not have a word for it. We’ve called them non-breed dogs I think, but that’s really awkward.

EK: Yeah, we call them village dogs. But that implies a whole lot of things about their lifestyle which may not necessarily be true. It’s actually really interesting how 80% of the dogs on this planet, we don’t actually have a word to explain what they are. But anyway, so if you imagine that there’s this population of “just dogs” before we started creating breeds and the breeds are created from dogs sampled out of that population. And so you make golden retrievers out of this set of dogs that come out of that general population. But that set of dogs shared DNA between them. They were one big population of dogs the same way people share DNA with each other, you know. And so, even though you might see that a particular genetic change went to high frequency in a breed, because of everything that goes along with creating a dog breed that doesn’t mean that that variant isn’t going to be seen in other breeds, just at much lower frequency. And in fact, there’s very few variants that we ever see that are restricted to a single dog breed. Virtually all of them are shared. There’s a handful of examples of mutations that have happened since the breed was created, and those might be restricted to a single breed. But the breeds have only been around for a couple of hundred years, which is probably like, I don’t know, 50 to 100 generations in dog generation times, and that’s just not a lot of time for new mutations to arise in a population.

45:17

JH: And so I think it’s important to just tie this back to what we were talking about before, which is that what Elinor is saying is that although you will see those individual variants that we’ve associated through scientific studies with diseases spread across a bunch of different breeds and in purebreds, it seems to be that there’s this whole background that is probably important for a lot of them to actually manifest. And so what’s happening in these breeds that we’re creating is that we are pushing them in particular directions to look particular ways and act particular ways and have particular coats. And what we aren’t seeing is that underneath, we’re aligning a lot of their DNA to look particular ways as well. That’s just what happens when you start trying to fix traits. And we call that genetic background, and it’s that genetic background that is probably necessary for some of these diseases, for some of these disease variants that we’ve identified to actually cause the disease.

46:20

EK: Yeah, we actually published a paper several years ago that had a really nice example of this. This was a study we did on bone cancer, osteosarcoma, and we looked for genetic changes that increased risk of osteosarcoma in three breeds. We looked in greyhounds and rottweilers and Irish wolfhounds. Three breeds that are all at risk of developing osteosarcoma. And in the greyhounds we compared the cases to the controls and we got this beautiful association, meaning that there was a difference between the dogs that were getting the cancer and the dogs that weren’t, at a position near this gene called CDKN2A. Which, yes, it took me a while to learn how to say but I got there eventually. Which is a gene that’s known to influence cancer risk in humans as well. So that was a beautiful, nice association that made a lot of sense in terms of the disease we were looking at. That this particular change in the DNA, or something near it, could be increasing the risk of these greyhounds getting osteosarcoma. And so we said, “This is great.”

And then, as I said, we know that variants are shared between breeds. So we went to look in the rottweilers and the Irish wolfhounds to find out whether they also had an association at this locus where the cases and the controls look different. And we couldn’t see anything in either of these breeds. It was totally flat. No signal of association at this gene in either of those breeds. But then, and this was back when I was doing my PhD and it was probably very late at night. And it suddenly occurred to me to go look at this gene. And it turns out the reason we didn’t see any association at these genes in either of those breeds was that they were both essentially fixed for the risk factor that we saw in the greyhounds. Which means not that every single rottweiler or Irish wolfhound is going to get osteosarcoma, because as we’ve been talking about through this whole podcast genetics is really complicated. But it means that all of them are carrying a variant that probably puts them at increased risk, which is why more dogs within those breeds are getting this cancer than dogs that are not within these breeds.

And it was just a really interesting example to me of nobody intentionally fixed this variant in these breeds. It’s just something that goes along with creating a dog breed. You’re reducing genetic diversity. In order to get what you are actually looking for, which is a particular look and a particular behavior, and whatever it is that characterizes this breed. You’re getting there by reducing the amount of genetic diversity in your population by having the dogs be genetically more related to one another. And sometimes you can just accidentally fix things that are not what you’re looking for, and might increase the risk of disease.

49:03

JH: And I think you found something similar in another study where you were looking at canine compulsive disorder in dobermans.

EK: I’m trying to remember. So in dobermans we found a really nice… Oh, that’s right. Yeah. So we found…

JH: You talk about this in this talk that I have stolen slides from so I know you talk about it.

(laughter)

EK: Yes, I’ve done a lot of diseases at this point in time. I’ve got to keep them all straight. Now, I’m trying to remember what the name of that gene is. And I’m not sure I’m going to be able to reel it off off the top of my head.

JH: There was the one that you found that had the association was a cadherin. CDH-13? CDH- something. That was one of these genes for a protein that it ties neural synapses into place.

EK: CNC, CNDDA2 something like that.

JH: Oh it was CTNNA

EK: Yes.

JH: 2

EK: 2

JH: Yeah, that was a calcium. It’s a calcium channel.

EK: Yeah. Anyway, everybody can go look up the paper on this if they really want to know the names of the genes.

JH: No one cares. (laughter)

EK: But basically it was actually really interesting. So we found this one gene CDH7, I believe.

50:09

JH: I got the number wrong, but I knew it was a cadherin.

EK: Yeah, but was it 2? Anyway, that is associated… There’s a change in this gene that was associated with risk of developing compulsive disorder in the doberman pinscher. This really a particular form of it called flank sucking, where they kind of turn around and chew on themselves. But even though it was a beautiful association we actually did the calculations and figured out that probably this one locus was controlling maybe 5, or 8% of the risk of getting this disease in the brain. So it’s really only the tip of the iceberg.

JH: And when she says locus she means that particular variant that she found.

EK: Thank you. Yes, sorry, that particular variant. It’s amazing how many different words we have to say nearly exactly the same thing. It gets very confusing even to me sometimes. And what we found was that there was… So the next thing I did, I was like, “Okay, so we did this study on osteosarcoma where we found this weird pattern where the rottweilers and the Irish wolfhounds had this fixed region at a region that was associated in the greyhounds for compulsive disorder.” At this point in time we only had the one breed: dobermans. And so what I did was I just went and looked at the dobermans as a population and said, “Which parts of the DNA in dobermans look the same in all dobermans?” And I looked and I looked for regions where all dobermans look the same. And the second biggest region of the genome where all dobermans look the same as each other was this gene called CDNA2, or something like that. And it turned out that it was actually a protein that if you looked in the neurons CDH2 is this… Do I have the name of the gene right? CDH7. Anyway, the cadherin is involved in anchoring the presynaptic and postsynaptic membrane together so basically connects these two halves of the synapse.

JH: So that brain signals can travel.

EK: It’s the glue behind, between it. But so the other gene that I found was actually the protein that was involved in the anchoring of that protein that’s connecting the synapse so they actually directly interact with each other. And we went on to kind of expand that study and look at a few more breeds. And it turned out that in the other breeds we looked for, we actually did see an association at that gene. They weren’t fixed. And so we could actually see the association, but in the dobermans because they’re all fixed we couldn’t find a difference between cases and controls because they all look the same.

And this was actually one of the things that led me to start this Darwin’s Ark Project where we enroll all dogs, all breeds, all mixed breed dogs, because I realized that studies where you were looking only in one breed were always going to have this problem that you might be blind to things where the risk factor was carried by all dogs in that breed.

52:51

JH: It was a beautiful segue, because my next question was going to be what is the future of these studies?

EK: You know my biggest worry, strangely enough, with the way that the genetic testing is getting talked about in dogs is that it implies that we know a lot more than we know right now. It implies that we have the answers. And I don’t want to imply that we’re not going to have the answers. We’re going to get there. There’s a path forward, and I am very confident we’re going to get to a place where genetic testing is an integral part of dog breeding. But we’re not there yet. And I get worried that when papers get published implying that we know a lot of things that we don’t quite know yet, that it implies to people that we’re done and we can stop now. We don’t need to do any more research, because this is an answered question.

JH: Just get the test done on your dog. And you’ll know.

EK: Yeah, exactly, you’ll know the answer. So why would we need to do any more research? So I’m here to tell you that we don’t actually really know anything yet. Well, we know a few things we just don’t know nearly as much as we’d like to know.

JH: We know just enough that we deserve funding to learn more.

EK: Yes. And so in my opinion the future of dog genetics, and all genetics we’ve learned this in humans now, is that in order to understand how genetics influences disease risk, we have to have very, very, very large sample sizes. In humans, they have a study now looking for… They had a beautiful study where they found genes that were associated with risk of developing schizophrenia. This is really important work because we have no idea what causes schizophrenia. And the idea, obviously, this being humans is nobody is proposing that we’re breeding humans for schizophrenia or not schizophrenia. But if we could find genes that are associated with the risk of developing schizophrenia, we might be able to understand the biological causes and develop drugs that are more effective against the disease. And so that’s kind of the goal at the end of the day. And so they had this beautiful study where they found hundreds of different genetic variants in the genome that are risk factors for developing schizophrenia. But in order to do that they had to have something like, was it?

JH: It was 200,000…

EK: 200,000 people.

JH: Yeah. And there have been other studies that I mean, they’re over 500,000 in some of these studies now. I feel like I saw one that was pushing a million people in it recently.

EK: Yeah. That’s the goal.

JH: And I was like, “Oh, I need to tell Elinor to update.”

EK: Yes.

JH: Because the schizophrenia at 200,000 is old news, dude. So say again, how many variants for schizophrenia that study found.

EK: Oh, it must be at this point… I think was like 122 or something like that.

JH: Yeah, right? 122. And that’s crazy, because we think it’s going to be one or two, and you’ll be able to test for it.

55:41

EK: Yeah, it’s surprising to me how far… I mean if I had to get one message to people in the dog breeding world, it would be, “Remember that genetics is really, really complicated.” And that it’s. The more we look, the more we find. And so I have no reason to think that cancer in dogs is going to be easier than schizophrenia. We might end up finding dozens or hundreds of risk factors for cancer. But when we find them, we are going to be able to both understand the disease better, which might help us treat it better, and potentially identify which dogs are at high risk. And even from a dog breeding standpoint, we might be able to get to a point where we can identify which dogs are okay to breed together and which dogs are going to create puppies that are going to be at really high risk of getting this disease. I would love to get to that point. And we’re not there yet. So I think we basically need to follow the path that’s now set by the people in human genomics, which is that once they realized that they couldn’t get anywhere without having really, really large studies, they all started working together. 

And so that’s kind of the goal of our project, Darwin’s Ark. Where we’ve set the whole thing up as being an open data project. We don’t own the data. We’re not selling the data. It’s data that we collect as a resource for both our research and any other scientist out there to come in. And so if a dog participates in our project and we get the sequence data, we’re not holding that and sitting on it in our lab in order for us to do something for our projects. That’s now a resource for the community.

And the Dog Aging Project, which is another big dog genetics project that’s gotten started recently, they’re a lot more than genetics but genetics is part of it, is exactly the same idea. Is that this is something that we’re building as a resource for other scientists. And so some of their data sharing policies are slightly different than ours because it’s a different design of project. But ultimately, all of that data will be shared. It’s a publicly funded project. And it’s really beautifully done. And we’re working very closely together with them to make sure that all of this data can kind of get pooled into much bigger data sets.

But of course the other part, and I always try to make this point really clearly because it always surprises people, is that everybody thinks of genetics as being the complicated part of our research. And it’s not. The genetics, honestly, getting a DNA sample, sequencing it, looking at the DNA data on our computer. That’s not the hard part. The hard part is the phenotypes. It’s the diseases. It’s knowing which dogs get sick and which dogs don’t get sick. Which dogs have behavior problems, which dogs are the most amazing dogs ever and have never even thought about having a behavior problem. And there’s only one set of resources for us to get that information, and that is through the help of the dogs’ owners themselves. And this is the other reason that we set up our project as being an entirely open data project is that it’s not our data. It’s data that everybody’s actually participating to contribute. And so it felt only right to make sure that it was a shared resource that was available to everybody as well.

58:49

JH: Yeah, I love looking to the future with these projects because I think there’s going to be some really amazing things that we can do. And I also wanted to give a shout out to the Working Dog Project, which you didn’t get a chance to mention yet, but I just interviewed Eldin Leighton about the International Working Dog Registry, which is and so that hopefully that podcast episode will air right before this one. That’s the plan currently, but we’ll see what happens. So if not, it’ll be right after. But so we’re collaborating with him. So do you maybe want to talk about the Working Dog Project just a little bit too?

59:29

EK: Yeah. So the Working Dog Project is one of those… So I did not realize I would be doing a working dog project until a woman named Anita Migday from an organization called the Theriogenology Foundation. I met with her at the Broad Institute one day. Somebody said, “Oh, you should talk to Elinor.” And she asked me possibly what is the most dangerous question to ask any scientist which is, “What project would you do if you didn’t have to worry about funding?” And I said, “Well, I would do working dogs because I feel like we know where we need to get to with working dogs. And we know that behavior in dogs has a genetic component. And so we should be able to use genetics to breed more effective working dogs. But getting there is going to be really complicated and expensive, because behavior is complicated and influenced by many genes and the environment. And we’re going to need really, really big sample sizes.” And thus the Working Dog Project was born.

And not very long after that I reached out to Eldin Leighton, who was one of the first people in the dog genetics world that I actually met way back when I was doing my Ph.D. I traveled down to Seeing Eye with my advisor at the time, Kerstin Lindblad-Toh, and met Eldin right at the beginning of me getting into dog genetics. And so a lot of the way I thought about things had been influenced by what I’d heard from him at the time with all of the genetic, genomically informed breeding that he was doing there. And so when I got involved in this project with working dogs and trying to figure out how to breed better working dogs, the first thing I did was reach out to Eldin and say, “Hey Eldin, are you doing anything in working dogs?” And he said, “Well, actually, I have this thing I really want to do called IWDR.”

And so we paired up at that point in time. And so the goal of the Working Dog Project is that in the very long term we want to be able to do for dog behavior what we’ve done with other traits and diseases. Is really understand how genetics influences behavior. You know, as a scientist, I’m really intrigued by that question of, you know, how do you actually change DNA in order and get a change in behavior? Like, that seems amazing and complicated and beautiful, and it’d be really cool to figure that out. But along the way to get into that kind of, you know, what would be a really interesting science, we’re actually doing an awful lot of work to figure out how to breed better working dogs. And by better here we mean dogs that are able to do the jobs that they need to do but are also happy and healthy. And will live for a long time, because that’s just as important to being an effective working dog is their ability to do the job that they need to do.

1:02:15

JH: Yeah, and so to tie it all up with a ribbon, the hope is that we will be working with him to develop better genetic tests which would allow breeders to make some decisions about what kind of risk their dogs will have for hopefully diseases and possibly even wanted or unwanted behaviors. So we talked with him… In my interview with him we talked a lot about estimated breeding values, genomic estimated breeding values. And so I think that, my hope is that that’s going to be the future of genetic testing, not just looking at these individual variants, right? But looking at a whole lot of them.

EK: Yeah, I think that what Eldin does, and the approaches that he and other animal geneticists are using and have been using to be fair for decades basically, is because they all know that no genetic variant acts in isolation. That the genetic background matters. And so rather than looking at a particular disease variant which is what we’ve been, kind of how we’ve been kind of thinking about things in dog breeding, they look at the entire genome of the dog and making a decision about whether this is the dog, the individual that they want to breed or not. And so it actually is a really nice tie in and I’m really excited to be working with him on it because I think those approaches are the ones that are really going to be powerful for helping people breed dogs that are healthier, but not actually end up with some of the problems, you know, the concerns that you get with, you know, removing diversity from the population. Like how do you actually balance out retaining diversity in our population and reducing the risk of diseases? And that’s not something you can do by any other method then the statistical kind of genomic breeding approaches.

1:04:07

JH: Yeah, I’m really looking forward to that future getting here. So we’ve been talking for a while. So we should probably wrap up. What… Where would people go if they wanted to learn more about you or some of your research? Do you have anything on the internet that people could look at by any chance?

EK: Well, you know, our Darwin’s Ark project does have a website…

JH: It does!

EK: That they can go to: darwinsark.org. So if they’re interested in finding out about what we’re doing with people’s pet dogs and also with working dogs, that information is on there.

As a scientist I’m trying to learn how to be better at marketing because it’s not my first instinct, but if people are interested we would love to have them sign up their dogs. And we ask you lots and lots of questions about your dogs and you have the option of sequencing your dog through Darwin’s Ark. It costs $149. But that’s at cost. It was something that originally I really, really just wanted to sequence everybody’s dogs and not charge them money for it. But then a lot more people signed up than I thought would sign up, and we didn’t have enough money. So we figured out a way to get people to, if they chose, support the sequencing of their dogs at cost. You can participate without doing the genetics. That’s not a requirement. If you do the genetics, we will share the data with you. And we will share the data. We’ll put it into our open resource for any scientists to use.

As you might be able to expect from some of the things I’ve been talking about on this podcast, the one thing we’re not doing right now is reporting on the disease variants in our data. We’re doing a different kind of sequencing than the companies do. It gets us a lot more information on each dog because that’s what we need to do research. But we don’t report on the disease variants themselves, although we do share the data with people who are interested in doing something with it themselves. And I’m sure that Jessica could talk more about all of that, or has at some point in time.

JH: Yeah, I certainly can. Well, people can certainly come get on the Functional Breeding Facebook group and ask all kinds of questions and should ask all kinds of questions about this podcast. And I own a copy of Elinor’s brains, so I can just…

EK: Exactly. (laughter) Well, I mean, it’s, as you can tell from me trying to talk about it, it’s something that we’ve discussed an awful lot, because I definitely don’t feel comfortable reporting on these variants, disease variants, to dog owners because we have absolutely no idea what they tell you about any individual dog’s risk of developing the disease. But for some of them that have been used over time by dog breeders within a particular breed where they really understand what that variant means within a particular breed, there is some value to those genetic tests. And so figuring out how to actually communicate this information in a way that is safe and not damaging is kind of one of those things that we struggle with to figure out how to do.

JH: That’s been the other really hard part, right? Not just collecting… So that genetics is the easy part. Collecting the phenotypes is hard. And then the very hardest part is figuring out how to give the information back to people in such a way that they, A) understand it without being overwhelmed by it, and then B) don’t jump to conclusions that are going to be problematic for anybody.

EK: Yeah, we have absolutely no interest in hiding anything from anybody. And that is shown by the fact that it’s an open data project. And we will give you all the raw data. It’s just that most people don’t know what to do with that raw data. And so I mean… It’s what? I think about 10 to 15 million genetic variant calls for each dog. It’s a lot of data. And so we have to distill it down. We have to package it up in some way. And in that packaging you can inadvertently introduce messaging. And so you have to be really careful about the way you talk about what it is that you’re finding, especially given that we don’t understand a lot of this stuff yet ourselves.

1:08:05

JH: Yeah, the amount of time that we spent talking with the website developers about like, (laughter) how will we explain to people how we did this very complex analysis? Make it graphical. 

(laughter) 

EK: It’s, there’s many iterations that go back and forth. It’s been eye opening for me. We also said that the editorial that we wrote on pet genetic testing (which we wrote for a general audience, not for necessarily a scientist audience) is an open access article. And I think we have a link to it from our website as well.

JH: I can put it in the show notes.

EK: Yeah. And we did try to… As I said that’s focused more on the use of genetic testing in clinical veterinary medicine because, partly because I was really shocked to find out that it was being used in clinical veterinary medicine. I… As a geneticist I just didn’t feel like the information was there to be able to do that except for a tiny, tiny handful of cases. And so to find out that it had been adopted as something that was a thing was really scary to me. And so that’s what we ended up focusing on in the editorial. And also, we talk a little bit about the problems with how these genetic variants were discovered in the first place and things like that.

And, you know, as many people do these days I’ve got my own Twitter feed as well, where I tweet about science and possibly about other things as well.

1:09:26

JH: So what’s your Twitter handle?

EK: I think it’s @eenork which is very odd. I can’t remember how I came up with it.

JH: It is what it is. And I’ve always wondered how you came up with that.

(laughter)

EK: I have no idea. It was probably the middle of the night and Elinor was taken and I somehow ended up with some weird combination of letters and now it kind of stuck. So…

JH: I like it.

EK: At least it’s short.

(laughter)

JH: It is short. I always wondered if it was like… I was like, “Is Elinor actually her middle name and her first name is another E name or something? No…”

EK: I just wanted to confuse people. That was really what was behind it.

JH: Well, well done.

EK: Thank you.

JH: All right, well, thanks so much for coming on. And this has been a fantastic interview. I actually learned some things. I thought I knew everything in your brain, but I didn’t. So…

EK: You know, I’m always happy to help that brain dump process. It’s (laughter) you know, running a research group is a really interesting thing, because based on the people that are in your group, you can take your research in new directions. And so as Jessica knows, I’ve always been very excited about both the fact that she has a veterinary background, but that you’ve got an interest in the whole dog breeding side of things, because that is part of our work. And it’s not something that I have a lot of expertise in. And so it’s great to have opportunities to be able to talk to people that are struggling with how to use genetics in dog breeding.

JH: Yes, there are definitely synergistics to be able to connect to the research that we’re doing with dog breeders. So hopefully, we’ll keep finding ways of doing that.

EK: Yes, and muddling our way through trying to explain things.

JH: Yes. Hopefully people come out of this interview less confused and not more confused. But they should certainly feel free to reach out and ask questions if they’re more confused, and we’re happy to talk about it. All right. Well, thanks again, Elinor.

EK: Thank you very much.

Thanks so much for listening. The Functional Breeding Podcast is a product of the Functional Dog Collaborative and was produced by Sarah Espinosa Socal. Come join us at the Functional Breeding Facebook group to talk about this episode or about responsible breeding practices in general. To learn more about the Functional Dog Collaborative, check out www.functionalbreeding.org. Enjoy your dogs.

Copyright © 2021 Functional Dog Collaborative. All rights reserved.