Watch our Good Breeder Webinar with Ryan Boyko, CEO of Embark!
by Dr. Judi Stella, PhD - Head of Standards & Research at Good Dog
Are you a dog breeder with questions about canine genetics? Join Good Dog’s Head of Screening & Standards, Dr. Judi Stella, and Ryan Boyko, CEO of Embark, for a conversation about canine genetics and get answers to your most important questions. Learn more about Good Dog’s approach to health testing and community standards, why genetic testing is so critical to canine health, and how you can use genetic tests to help you have healthier dogs.
Good Dog is on a mission to build a better world for our dogs by educating the public and advocating for dog breeders. We are a young, tech-savvy company with an online community that educates the public, supports breeders, helps people connect directly with responsible breeders who have passed our screening and comply with our standards, and promotes responsible dog ownership.
Our goal is to use technology as a force for good and empower the good forces in the dog world with technology – to be a voice and platform for dog breeders, to counter the extremist propaganda, and change the conversation so the public realizes how critical it is to support and recognize breeders. We’re free for breeders (breeders also can’t pay to be listed) and provide support (legal, tech, breeding), and discounts on health testing.
Cat Matloub [00:06] We’ll go ahead and kick things off. First of all, a huge thank you to everyone for taking the time to join us today. Even though we’re working from home, I find myself to be even busier than before, so we’re really appreciative of your time. And we’re really excited to talk with you about such an important topic. I think you know the nicest thing about so many of us hunkering down at home is it’s actually a nice opportunity to take the time to learn about something that is so important to your dogs’ health and is so important to understand how to use it as a tool for having healthier dogs. We’ve got time while we’re at home with our dogs to do this type of testing, because it’s such an easy thing that you can do at home. Not to mention that Embark’s got a great discount going on for folks, but I’ll let them talk about that a little more.
For those new faces out there who might not know about Good Dog or about Embark, I’ll give a quick intro for Good Dog and then Ryan can give a quick intro about Embark, and then we’ll dive right into an exciting conversation between Judi Stella (Good Dog’s Head of Screening and Standards) and Ryan Boyko (the CEO of Embark).
For those of you who may not know: Good Dog is a relatively new company, launched last year. We’re an online community that really works to educate the public and to support good breeders. Not only are we free for breeders, but breeders actually cannot pay to be a part of our community; they just have to meet our community’s standards and pass our screening. Breeders who do pass our screening get a host of benefits, including: a personalized profile page that can essentially replace a website, if that’s what you want; free tech support; a certified Good Breeder badge; access to information and education like this; free services, like mediation services and legal support; and some free breeder tools as well, like weight loss management, customizable application, and some discounts on health testing with some awesome companies like Embark. That’s who we are! I’ll turn it over and let Ryan give a quick intro about Embark, and then we can dive into the main topic.
Ryan Boyko [02:41] We started in 2015 with the mission to end preventable disease in dogs. We see, of course, the biggest bang for your buck in doing that is working with breeders to help produce puppies who are healthier and healthier. We’ve been very excited to be growing our breeder community over the last couple of years and really seeing that taking off with greater and greater recognition amongst the breeder community about what we offer and how that differs from other offerings. I know some of the questions might get into that deeper, so I don’t want to spend too long on that. We offer a full panel of testing that lets us look across a dog’s entire genome in a way that is, in some ways, similar to a 23andMe company that allows us to both provide complete results for a dog today and also updates dogs’ results in the future. And also we’re able to drive research that, up until this point, has really been impossible to do. We now have hundreds of thousands of dogs in our system, and people can optionally choose to share information about their dog’s health and all that, and that all goes into our research. To me, we’re the only platform where you’re getting all of this information now, and also directly driving the ability to do more in the future.
Cat Matloub [04:24] Awesome, thank you! Judi, do you want to kick things off?
Judi Stella [04:32] Sure! I just want to give you guys a little bit of background about how we think about genetic testing and health testing in general, as part of our screening process. We do have a pretty substantial screening and standards process. When we review programs, we look at five key areas. One is breeding practices. Age of onset of breeding, retirement age, things like that. We look at physical health of the breeding dogs, as well as the puppies. Preventative care, as well as screening for genetic and health conditions. We look at mental health of the breeding dogs and puppies, and how that is attended to. The environment that the dogs are raised in. And also, how the breeder educates the puppy buyers and policies around that, and around supporting their puppies for the lifetime of the dog.
So, the physical health aims of the breeding dogs: one of the things that we focused on is setting up and establishing health levels for testing, for what kind of screening processes and screening tests the breeders implement in their programs. This is a pretty involved, multi-step process. So we gathered as much data as we could that’s publicly available. We looked at chick requirements for each breed, and then we also researched the conditions to identify the most prevalent and severe conditions that are common in each breed. We looked at the prevalence of abnormal carriers and affected dogs for genetic testing, where that was available. OFA data, the scientific literature—we put all that information together, and then we developed these levels. We focused primarily on the most severe and the most prevalent conditions. Those are minimum requirements for getting on our platform. For every breed, the most prevalent and severe condition is the one that we require (or several, if there are several). And then we go up from there. So genetic testing is a really, really important part of this, where we have identified mutations and we can screen for them. It’s really important to incorporate that into our breeding programs. That is why we’re going to talk about that today, specifically.
Cat Matloub [06:51] Can I just ask one question? Can you speak briefly about how our levels and how the construct there relates to our mission?
Judi Stella [07:03] Our mission, as Ryan said with their mission as well, is we want to improve the health and wellbeing of dogs for their lifetime. We want to minimize the risk of these conditions being produced in populations. If we do do the screening tests, we will wind up reducing that risk and breeding healthier dogs.
Cat Matloub [07:24] Exactly. And the levels set forth a road map to help breeders get there, one step at a time.
Judi Stella [07:32] Just to start off, Ryan, if you want to tell us a little bit more about the Embark genetic testing panels? If you want to give us a little more information about those, how they work, how you would implement how breeders would go about getting a panel done?
Ryan Boyko [07:52] We’re offering a panel of breed-specific mutations to each breed. We, for most breeds, encapsulate either all of the known mutations or nearly all of them and all of the ones that are prevalent in that breed. We start with the premise that we don’t ever want to release something until we’re certain that we’re as accurate as possible within the standards of lab science. Generally, to us, it’s more than 99.99% accurate. We both have to be confident that the mutation actually does matter (a lot of times somebody might make a publication or might not even have a publication, and we need to see evidence that it actually matters before we do something) and then we need to be able to have it on our panel and demonstrate that level of accuracy. For most breeds, we have everything that people normally test for. We will send you a swab, you get the swab and send it back to us. (There’s a postage-paid envelope that we send you with it.) And then it enters our laboratory flow. All of the swabs actually get tested across the same panel that’s over 200,000 genetic markers. As we report results, we report them to you in a breed-specific way. We say, “These are the things that matter for your breed. Here are those results.” But we also do test every dog for everything, which lets us do several things. One is we’ve found many cases where a dog had a different paternity than expected. We can see the relatedness of dogs with our panel, as well as specific health mutations. We’ve been able to solve mysteries for breeders much more quickly than if you were doing one-off testing. But we also do test conditions that are not known yet in breeds, and we’ve had some cases where we’ve been able to detect—early on—that mutations not known to be in the breed are starting to appear in a breed, which lets us work with the breed to see whether or not those actually cause illness in the breed. The testing, as we think about it, is we’re giving you a complete panel for your breed, and we’re also serving as a sentinel for anything else that might be there. We’re also able to help you in relatedness. We’re able to help you when you’re looking at breeding pairs. What will the puppies of these two dogs be like (in terms of color, in terms of health traits, in terms of their coefficient of inbreeding)? We’re the only tool that can get you that accurate coefficient of inbreeding for what a prospective litter will have. We’ve been able to show that more highly inbred dogs within a breed live less long, have smaller litters, that kind of thing. We package all this information together. From the time we get the swab back from you, the large majority of people are getting results in under two weeks and everyone’s getting results in under three weeks.
Judi Stella [11:38] That’s really interesting, and you did mention that there are benefits to doing the panel, versus just doing the known or identified mutations within a breed. Can you give an example of any diseases within any breeds that you may have found, that you may have identified the mutation for, through looking at it at a population level like that?
Ryan Boyko [12:03] Yeah, so there are several examples in each direction. In one case, we’ve actually been able to compile a much larger data set of dogs with known DM results than has existed before. And we’ve been able to pretty thoroughly demonstrate that in a lot of those breeds, there really is a strong piece of evidence that it doesn’t matter. We actually updated the way we present results to people to show that. We offer veterinary consulting to our clients, and we’ve been able to work with people to first demonstrate that it really is not predictive of any kind of health condition. Our results change the way we talk to people. We’ve also found, for example, some PRAs (like RPGRIP1). We’ve seen that mutation in several more breeds than it had ever been identified in before. We are actively following up with breeders in those breeds who have those dogs, to see what the clinical significance is in those breeds. There’s one example each way—where in some cases, by amassing this data set, we can pretty clearly demonstrate that mutation’s not important for a lot of breeds. In other cases, it looks like mutations that weren’t known in other breeds may actually be important for some of those breeds.
Judi Stella [14:00] That’s interesting. Can you explain the technology that you use to do these screenings? There was a question that came in yesterday over email; people wanted to know more about the SNP technology and how it’s used and the accuracy of those results. Is it a single mutation versus if it’s a more complex mutation—how accurate are those SNPs?
Ryan Boyko [14:39] The way that technology works is it’s looking for a pattern in the genetic code, essentially. While you’re testing a single SNP, there’s probes basically, and those probes are 35 nucleotides (35 letters) attached to a specific spot in a dog’s genome. It tells you what the next letter is. You’re actually testing for a sequence to look a certain way. For example, even things that are not a single mutation difference, a single SNP difference, a single letter difference, you can actually tell with just as much accuracy (based on what that next letter in the sequence is). There are cases where even when something that’s not just a difference—where a T became an A in that genetic code—there are some that you can test correctly. There are others that you can’t test for in quite exactly the same way. What happens in those cases is we do different tests in our we do different tests in our lab for a whole lot of dogs—dogs that are known to have the mutation, dogs that are known not to have the mutation. We also run them on our array, and we can combine that data to figure out if there is a stretch of letters that we can test for that are extremely predictive of having or not having. Like I said before, when we’re testing the single letter changes (or the things that we can directly test, which are always the single letter changes), that accuracy is over 99.99%. When we’re testing these linkage tests, we don’t release one until we are certain that we’re over 99% accurate. We do identify in the results which ones come from linkage tests, which ones come from other tests. I will say that because of all of the checks that we can do along the way, we also find cases of people that swab the same dog twice. When we test an entire litter, we find things like that. We find things that, because of our technology (where we’re not looking at one SNP alone at a time), we’re actually able to catch mistakes that would be impossible to catch. All these swabs have bar codes on them. They’re followed the entire way through. It’s done by computers. We’ve never had a case of swapped swabs. We’ve looked, and we have a lot of QC, so I think there’s one important thing to keep in mind: what’s the overall accuracy? The accuracy of a test is both how accurate are the probes and how accurate is the actual physical test you’re running? These array tests are as high or higher than other tests on that. But there’s also: what other kind of errors can come in? What kind of human error can come in—both from the user and from the company? We’ve been able to minimize that in a way that is just not possible to do with a lot of these single-mutation tests. Overall, our error rate we feel very confident is lower than the other labs. But we do let users know when a result came from a linkage test. Over 99% accurate (in many cases, probably over 99.9% accurate) is what we claim and stand behind.
Judi Stella [18:51] That’s on their results, or is it on the webpage so that people can find that?
Ryan Boyko [18:57] It’s on the webpage. When you click into the condition, it’ll say “Linkage Test,” and it has a little question mark so you can learn more about the linkage test.
Judi Stella [19:10] Okay, and I think that leads right into: How does Embark ensure the integrity and the quality of the results? What kind of quality insurance or quality controls do you have in place at your company?
Ryan Boyko [19:22] So we have a lot. Myself and my cofounder come from a scientific background. We built this in order to provide a tool that is better than anything out there. There’s no point in making a company unless you can do that. It’s been in our DNA since the beginning. We start with the array technology, which as I said, is greater than 99.99% accurate. We designed multiple different ways of looking at each mutation. Your DNA code goes in two directions, right? We can look from both directions, and we can put multiple probes that are looking at each mutation. For some conditions, we have up to seven that we’re looking at. Each one of those is 99.99% accurate or more, on average. And then we force them all to agree with each other, or else we will give you a “No Result” for that. We’ll send you a new swab to test. At the heart of what we’re doing is a process that started out as highly accurate and then we built redundancy. Because of the way this technology works, it’s also nearly entirely hands-off on humans. From the beginning, you activate the swab to your dog, and you tell us a little bit about the dog (at minimum: the sex of the dog). And you send it to us, and then from that point on, it’s all about the barcode. The barcode goes directly into the computer system, so there’s no human manually importing and exporting things that can lead to a lot of these lab errors that happen. The lab itself is a CLIA certified lab that runs human samples as well. It’s very highly regulated.
And then the data: There’s QC that happens at the lab. They send data to us. There’s QC that happens on our end. Like I said, we’ve never had a swab swap, but we test the sex of the dog so it matches the sex that we’re seeing, as one of many kinds of tests that we run, to make sure there’s nothing fishy going on. We also require the samples to be of a certain quality. We’re looking at over 200,000 markers, and if you’re missing more than 1.5% of them, then we throw out your data and send you a new swab for free. We have several steps of QC in how we design and look at all these processes and samples. If any sample shows an inconsistent result between probes or anything that looks off, there is a human step where we look at the raw data from that swab. We can either resolve it with confidence, or we can’t. And if we can’t, we send you a swab. In those cases, a lot of the time, they’re rare mutations—not of the thing we’re looking at, but it could be next to it. We’ll actually send those to our ND lab, so that we can try to figure out what’s going on. If we’re alerted to any results that might be inconsistent (they might be inconsistent with a different lab or maybe inconsistent with appearance), we send out free swabs, test them with multiple different methods with our ND lab. Over 90% of the time, we’ve confirmed that our results are right. A lot of the time, we’ve found that these things happen because of missed paternity, but there are other reasons, too. We proactively alert any user. In fact, we did have a case in December where we saw one result that we did give incorrectly. We’ve done over 20 million individual tests overall, so it happens—rarely, but it does. And then what we did, even though it was over the holidays, we immediately got a bunch of our team together to go over every similar result that we had given. We proactively found two more cases. Fortunately, none of them resulted in any affected dogs being born or anything, but we proactively alerted the owners of those other cases, and we gave all of the owners both refunds and free swabs for every puppy born of any of those three dogs. From start to end, we do our best to always give accurate results. We are highly, highly accurate. We also stand by making sure we do right by our customers, if something rare does happen.
Judi Stella [25:24] Okay, great! I’m going to switch it up a little bit and just start talking about how we incorporate genetic testing into overall health practices. How would you recommend genetic testing be incorporated into overall health—for breeding dogs, as well as for all the puppies being born?
Ryan Boyko [25:44] That’s a great question! The main thing is knowing your breed. We are very much a tool. We see ourselves as a tool—that hasn’t existed before—to be able to help breeders do what they’re already doing even better. Breeders know their breed in a way I’ll never claim to know. We look at 200 breeds. You’ve spent decades on one breed. I think there are some things that are obviously true, that if you have a disease that highly impacts the dog and it is prevalent in your breed, then I think you have a responsibility to test and to make sure you’re avoiding carrier to carrier or carrier to affected matings for that, right? There is, of course, a range of diseases and a range of frequencies, but I think what we know is that most breeds have one or several conditions that fall in the This Is Really Important bucket. To ethically breed dogs, you should take that information into account and avoid creating dogs that are affected with severe conditions. There’s also a growing awareness of the danger of highly inbred dogs in general. Embark has been able to collect more data than available before and make publications that have demonstrated that you actually can see a large reduction in life expectancy with more inbred dogs. You get smaller litter sizes with more inbred dogs. There’s a holistic process. There’s an art to this. I don’t claim to be an artist in this or in any other way, but I think breeders have a duty to use the tools that are available to (at a minimum) avoid breeding dogs that have knowing genetic conditions that will affect that dog profoundly. The next step up is to say: You really should be avoiding breeding highly inbred dogs, relative to the average level of inbreeding in your breed. That’s where Embark can come in, because even with pedigrees, you don’t get as accurate of an answer as you can get with genetics.
Judi Stella [28:40] You mentioned DM a little bit earlier, and we have had a couple of questions about DM in the chat. It is a pretty severe disease, but like you mentioned, the mutation’s been identified in a lot of breeds, but there’s very few in which it’s been a clinically significant disease. One of the things that we do when we are looking at DM and whether or not we include it in our health levels is to see whether it’s been confirmed through examination of spinal cord tissue from the University of Minnesota. Where it’s been clinically identified in the spinal cord tissue, that’s where we identify it also and include it in our levels. Can you speak more to that? It does seem to be one that is controversial, or at least people have a lot of questions about it.
Ryan Boyko [29:34] This is where I’m going to put in one more plug really quickly for if you do choose to test your dog with Embark. You can access our veterinary geneticist and talk to them about your specific breed and questions, because this does highlight areas where, when you’re trying to give a comprehensive answer across many breeds, it’s very hard to give a very clear answer without being able to talk to somebody about specific breeds and challenges. DM is a great example where, as you said, the only way to definitively diagnose it is on necropsy. There’s only been a small number of breeds where that has actually been done. We have now tens of thousands of dogs who are just looking at that one gene. If you didn’t know what breed they are, you would say they are genetically at risk for DM. But we know. We send health surveys. We also talk to other scientists and review literature all the time. We know that it’s not predictive in many of these breeds. That is an example of where you don’t want to overcorrect in breeding. You don’t want to say, “Just because this gene has been associated with a condition in some breed that’s not even closely related to my breed, it’s definitely a bad thing that I should breed away from.” We actually listen to our customers, and some of the change in how we presented results was just to be more clear, and some was to help them be able to show to their puppy buyers that this is not something they should be worried about. It really depends on the breed, but in general, if your breed is not known to have DM, then you don’t want to be removing dogs from your breeding pool and causing other problems. The more dogs removed from your breeding pool, the more other problems you’re likely to cause. Only remove dogs from your breeding pool for a good reason.
Judi Stella [32:10] That’s good. I’m going to use ocular diseases as an example here. I just want to see if we can discuss how we incorporate or use genetic testing in compliment with phenotypic testing. For instance, PRA the mutation has been identified and yet in some breeds, it’s still recommended that they get ocular exams. Why would that be, if the mutation’s already been identified? And can you just talk about how we incorporate genetics as well as the phenotypic testing in some cases?
Ryan Boyko [32:45] There are a couple of answers. One: many of the kinds of PRA are not what’s called 100% penetrant, so even within a breed where having that mutation predicts a higher risk of getting that disease, in some of these cases, it’s not 100% risk. It’s always going to be substantially higher, but it might not be 100%. No genetic test that anybody does is a diagnostic test. It’s a test that can help a veterinarian that’s actually looking at your dog diagnose a disease potentially, and it can help you prepare for things and help you avoid breeding dogs that get diseases, but it’s not a diagnosis in and of itself. You need clinical manifestations to make a diagnosis. It’s impossible to predict with certainty which dogs can get it. It can be impossible to predict when they will get it. When you’re thinking about “How do I care for this dog right now?” it’s important to have the information about what the dogs eyes look like right now. A genetic test is not going to tell you that at exactly two years of age, they’re going to develop this in exactly this format. Some of it is because the tests we do have now tell you that this dog is at a higher risk, but they don’t tell you they have it. Some of it, especially in certain breeds, could be that there are forms of PRA that we just don’t know their genetic cause of yet. There may be other reasons why the dog’s eyes are going bad. Some of them are not even genetic—with kinds of glaucoma, where things that happen over the dog’s lifetime lead to it even without a genetic risk factor.
Judi Stella [34:47] I want to make sure we leave a few minutes for questions at the end. Is that okay, Cat? Ryan?
Cat Matloub [34:56] That’s perfect.
Judi Stella [34:58] I’ll just ask you a couple more, then let’s leave some time at the end for questions. How would you recommend breeders incorporate genetic testing into their programs and when they’re making breeding decisions as well? Are there instances where carriers or affected individuals should or could be bred? I think you touched on that a little bit, but if you could just expand on it.
Ryan Boyko [35:24] Again, this all happens in the context of a breed, so giving an overarching answer might sound like I’m waffling a little bit, but it’s because each breed has a different answer. Overall, the goal is to slowly reduce the prevalence of known genetic diseases in general. So if you have a breed where half of the dogs are carriers for a condition (maybe we just discovered the gene involved, maybe it’s a late-onset condition and it was hard to breed it out until you knew the gene) or, for some breeders, if all of your dogs have it, that doesn’t necessarily mean you need to immediately stop breeding any dog who is a carrier or possibly even affected, depending on what it is. The key is to breed them to a dog that is clear. Assuming that it’s a recessive disorder, like most of these are, then if you breed one of the dogs that’s clear, then none of the puppies will be affected. They might be carriers, but they won’t be affected. Some of it is a little bit of this analysis. There’s conditions where only 2% of the dogs in a breed are carriers and it’s almost gone already. Maybe you make a decision and say, “No. I really don’t want to breed any carriers. We’re only losing a tiny bit of diversity, and we’ll be done with this disease soon, functionally!” Still, we’ll have to test for a while to make sure it doesn’t pop up or we missed it, but that might be reasonable. But if you have a condition where lots of dogs are carriers for it, in no way would you want to remove all those carriers in one or two generations. You want to do it slowly, so you’re not losing all this diversity.
Judi Stella [37:15] And you think that’s the same for affected as well? Especially if you have a small gene pool, you might want to breed the affected to the clear?
Ryan Boyko [37:27] Right, yeah.
Judi Stella [37:32] Last question I have for you, and then we’ll turn to some of the callers’ questions: What is on the horizon for canine genetics in the next couple of years, and for Embark as well, specifically?
Ryan Boyko [37:44] We hope they’re intertwined! Although there are a lot of great scientists doing a lot of great work, which is great to see. We’re working with some of them. One thing that we want to really bring out this year and make better known is what are the impacts of inbreeding, and how can you best incorporate that? I think we’re putting out increasing amounts of content to help with that. We’re changing some of the way we present some of the results. We’ve expanded what we call The Matchmaker, so now it’s up to about 40 breeds. Last year, we had been piloting it in about five, so now it’s up to 40 and we’re going to keep expanding it. You have privacy control, so don’t worry. If you test, you don’t have to be in it. But for people who want to be in it, they can see other dogs of their breed of the opposite sex and be able to see how inbred the puppies of these dogs would be and what that looks like compared to the average dog in the breed. That’s a tool where you can evaluate. Maybe there’s three or four dogs you’re thinking about as potential sires, and this is one piece of information. Getting a better sense of how we reduce bad inbreeding is one place we can make strides. Other things are: more discovery. We are very actively looking at a bunch of disorders ourselves, we’re doing some with partners (academically, some with dog food companies and others). I think that’s really going to start opening the door to more complex diseases. We’re going to start to be able to say, “What is the risk that this dog might develop this condition?” where it’s not just one gene. So much of what we can test for right now is just one gene, but if you look on the human side, most of the things we test for aren’t “one gene causes a recessive disorder that’s almost guaranteed to have it if you get it.” You’re looking at several places in the genome, and they add up to a risk or not. I think we’re going to be able to move more into that. You’re never going to make progress on things like cancer and hip dysplasia until you can get into that. You need a lot to be able to make that research work.
Judi Stella [40:38] One question is: Can we talk about how we collect the samples? Is blood testing more accurate than Embark’s test using saliva? But you don’t use saliva—you use cheek swabs, right?
Ryan Boyko [40:54] We use cheek swabs. So we’re using the cells that are sloughed off, basically. I should specify. We have a sponge that soaks up saliva. The saliva itself (the proteins in the saliva that you think of as the sticky saliva) are not what we’re testing. There are cheek cells that are suspended in your dog’s saliva that just come out of your dog’s cheek all the time. Every cell in your dog’s body has the exact same genes in it. Other than their eggs and sperm, every other cell has the exact same genetic material in it. It actually has no impact—whether you’re looking at white blood cells that you’re getting from blood or you’re looking at cheek cells that you’re getting from saliva. They literally are the exact same genes. There is one difference with blood and saliva, which is that in blood you’re going to get a lot higher volume of DNA. So that means that with a blood draw, theoretically, you will be less likely to not get enough DNA. What I can say is that with our swabs, we see about a 1% rate of not getting enough DNA, which is actually not much higher, even with blood. People might mishandle it, there’s reasons why that might not work. Those probes, those 35-inch base pairs that are pulling things from your dog, those only match to dog DNA. If you were to let a human do our thing, there would be no data. If you ran a cat, if you ran saliva from two dogs, it would fail QA. It doesn’t look like a dog. There’s a lot of technical reasons why, but the short answer is that saliva is exactly as accurate as blood. You get the exact same DNA. And we can pull out the dog DNA from the saliva. Bacteria isn’t going to match up. And even if you have two dogs’ saliva, it will cause a failure, because it’s very easy to see across 200,000 genes that this is not from one dog.
Judi Stella [43:14] There’s another question that’s interesting to me as well. Is Embark going to be able to offer advanced color testing for merle gene variants? I think someone does it in the Czech Republic.
Ryan Boyko [43:26] To be clear, there are a few limitations of using an array. It has tons of advantages. There are a few limitations. With merle, that highlights one of them. We can see whether a dog has the non-merle genotype or a merle genotype, but there are multiple merle genotypes, and they don’t all have the exact same impact. Unfortunately, it is a complete limit of the array technology we use that we won’t be able to use that array to do it. For a lot of breeds and a lot of breeders, what we can give is enough information. For breeders who really want to know what exact merle genotype I have, right now we don’t offer tests. At some point, we may expand into other kinds of lab tests. We’ve been prioritizing trying to get the most benefit to the most dogs as fast as possible. We’ve been optimizing this array and what we can do with it. We’re focused on that right now. Some day we might be able to do it, but it won’t be with the array. It would have to be a separate test.
Judi Stella [44:53] There is a question about quality control. How do you handle errors if they do happen? Because at hundreds of thousands of dogs, even at 99.99%, that would still mean that with hundreds of dogs, there could still be a mistake somewhere.
Ryan Boyko [45:11] I’ve seen that question on the side while I was answering the other questions. I tried to hit on it some. There are a lot of these QC steps that do catch things. That is one big way we’ve been able to do so: to catch them before they’ve even gotten to a user. With that said, we know of about a dozen total times, where out of those hundreds of thousands of dogs we’ve tested, an incorrect result has gotten to a user. Only for about half of those did we find out because somebody wrote us. When somebody writes us and says, “I don’t think this is right. It doesn’t match the previous lab. This doesn’t match the parents’ results,” we do everything we can to get to the bottom of the discrepancy as fast as possible. We send out swabs to them usually, and test them in our ND lab with other ways of testing. In the one case I had mentioned in December, we actually identified where (like I said, our process is nearly completely automated, but if you get a result where a couple of those probes disagree, a person looks at it). What we found is that the underlying technology was giving the right answer. Scientists knew the right answer. But the way they were inputting it into the website led to a couple of cases of human error. As soon as we were able to figure it out (which took a few days), we were able to identify every other test that we’ve ever given where the same process had happened. I’m not aware of any other lab that can do it. With other labs, if you just get a single result. If it says, “Here, I got two copies of this allele or not,” they don’t have the raw data stored in a way that they can look. But we have the raw data, so we’re able to look back at every similar dog that went through a similar process, and we were able to find one other error in the same gene and one error in a different gene. We had actually looked at all the genes that were called with this process. So we proactively identified those and notified the owners of those two. That’s what we do. As soon as we discover a potential error, we try to get to the bottom of it. Over 90% of the time when somebody has said that we have a discrepant result with another lab, we’ve actually wound up being correct. But then in those other cases, you do get into situations. What’s important for us is that when we find something that didn’t work, we get to the root cause, try to fix it and prevent it for the future.
Judi Stella [48:30] We have a couple of questions about OFA. Can you import directly into OFA and does OFA recognize Embark’s genetic results?
Ryan Boyko [48:44] Yeah, OFA does recognize our genetic results. We have an OFA submission report that you can download directly from our website. So right now, you do send it and your payment to OFA separately from us, but you have a form that is prepopulated and all you have to do is send it to them—no additional work beyond that.
Judi Stella [49:16] Does Embark allow breeders to send alternate samples from litters, such as tails or dewclaws, rather than waiting until they’re old enough to get a swab?
Ryan Boyko [49:29] We don’t do that. Like I said, our goal is to provide the least expensive and most accurate testing we can. Having the single work-flow process provides that. I’m seeing a couple of comments where I think there’s a misconception. Our test is 100% accurate at being able to say, “There’s two dogs’ DNA on here.” We detect one in 10,000 dogs is a chimera. They actually have two dogs’ DNA in them. Not only do we know that, we know if a single chromosome is coming from two dogs in these cases, because they’re siblings. It’s true that you don’t want to test a dog immediately after they’ve nursed, but it’s not because they’re going to get the wrong result. It’s because they’re going to have a failure. That swab will fail. It’ll say you’ve got DNA from two dogs. If a puppy hasn’t nursed for a half hour and you can fit the swab in their mouth (which, for most breeds you can do; it’s a very small swab), it’s perfectly fine to test that dog at that point. We’ve actually done some analysis where even as young as one to two weeks old, we’re getting 97% success rates. At two to three weeks old, they’re just about at that 99% that we’re seeing for adult dogs. It’s important to try to avoid contamination by waiting about a half hour, but the result of contamination is not that you’ll get the wrong result. It’s that you’ll have to swab them again.
Judi Stella [51:13] And then somebody asks how long the cheek swabs are good for, or are they saved? Or do you have to send a new sample each time?
Ryan Boyko [51:24] We do save DNA from the dogs. We currently only use it for research, and the reason for that is that it only works to produce another sample about 80% of the time. So it’s helpful to be able to, when we’re doing research to make new discoveries, but it doesn’t really work very well commercially. But we do save all the data from the dog, and what that means is very often when there is a new result that we can release (it’s not all the time; unfortunately, the biology doesn’t work that you can make 100% promises) we can actually go back to the dog’s data and figure out what that mutation is—without having to do any new lab processing. If we can’t, if we have to do new lab processing, we do have to get a new sample from the dog. Unfortunately, we’d have to incur all the costs there. So they’d have to get a new test.
Judi Stella [52:37] Cat, Monica, do you have any other questions? Are there any on the Q&A?
Cat Matloub [52:40] Ryan, following up on your comment, can you give us an example of diseases in breeds where you’ve actually made discoveries and discovered new diseases using that data that you collect?
Ryan Boyko [53:02] There’s a couple of examples that I want to give. Keep in mind the power of the database. We’re exponentially growing, which a lot of people are now familiar with, because of coronavirus, right? The power of the database is getting much, much greater every year. Our first discovery that we made was actually the cause of blue eyes in Huskies and related breeds. Not exactly a health discovery, but very interesting. Sine then, we’ve been able to publish some work on the impacts of coefficient of inbreeding, the impacts of different ways of managing a breed and breeding populations. There’s this idea that you have one test, one health condition, and that’s the Holy Grail. What we’re seeing is the impacts of inbreeding and the differentiation of highly and less-highly inbred dogs within a single breed. It can be years of life that we’re talking about. It can have a bigger impact than many of the single health conditions. Because that impacted all breeds, we spent a lot of our early time with that. We are also working on a number of other projects right now, things like DCM and TBDD heart diseases, some seizure disorders, and several other disorders. We’ve got a couple of discoveries—you’ll see some publications shortly. I can’t, unfortunately, say right now. Stay tuned! There are some things that have been written up and are about ready to be submitted.
We’re excited because we now have so much more power. We do these annual health surveys (in addition to all the other surveys). We started last year and got around 10,000 people answering it. This year, it’s something like 40,000 people answering it. Just the data set building is getting more and more powerful. I think we’ll be able to move even faster.
Cat Matloub [55:49] Awesome. Ryan, I’ll ask you one more question before we wrap up, if you have a minute. It touches on a lot of specific questions. There are a lot of questions from breeders about why Embark doesn’t test for various specific conditions that might affect their breed. Could you speak to that, please?
Ryan Boyko [56:03] Yeah! There can be a few different answers, so I’ll try to keep it quick. One answer is that sometimes there’s not a test at all. Nobody’s made a discovery. It might be that people are misunderstanding. We might have a test in a different breed, but it’s not the same thing that causes it in that breed. Sometimes, it could be that nobody’s made a discovery yet. That’s actually the main reason why we don’t have a test for things. Often when people are asking, it’s because they think there is a test. There might be, somewhere. The most common reason is nobody knows what causes the health condition, so we can’t test for it. After that, we want to be able to stand behind our results. In some cases, companies are offering tests for things that there are no publications to support, or there are conflicting publications, where something says it does impact and something says it doesn’t. There are a number of things that we have on our testing panel that we don’t provide results for right now, because we are trying to validate whether or not it’s true before we tell people that information. In some of these cases, we actually follow up with breeders in that breed and ask with surveys, “Has your dog had this?” In many cases, we’ll actually get vet records because we want to be certain. And then if we can validate it, we’ll start offering it.
So, the most common reason is that nobody knows what causes it. The next most common reason is there’s not enough evidence for us to be convinced that it’s possible. The last reason is a few things, like the different variants of merle. Because of the kind of mutation they are, they’re not possible to test on an array with greater than 99% accuracy. The technology just doesn’t work. We’ve decided to focus, for now, on building the most powerful array that can do the most good for the most number of dogs, right now. Hopefully, we can expand our testing more in the future.
Cat Matloub [58:45] Thank you both so much. Judi, Ryan, this conversation has just been fascinating. I hope for everybody else as well. I want to reiterate something that I messaged around in the group. We are so sorry for any questions we didn’t get to, but we’ll email all of you following up and make sure that you get answers to all of your specific questions, so don’t worry about that. We want to get you the information that you need. Thank you all, again, for joining us for our weekly breeder webinar. It’s so exciting to be able to stay connected with you guys and think about how we can continue with our mission of building a better world for our dogs. Particularly now, there’s such a great demand for our dogs. We’re so grateful to you, our dog breeders, for bringing us these amazing creatures that we’ve devoted our lives to. Huge thank you to all of you, and keep an eye out for an email from us following up with answers to any of your questions.
Ryan Boyko [59:48] Thank you so much for having me on. It’s been great. Thank you!
Cat Matloub [59:52] Thank you, Ryan. We can’t wait to do more great stuff together, and we can’t wait to hear about what you guys are doing. It’s wonderful. Thanks, everybody! I hope you enjoyed it, and I hope you have a great afternoon! Let us know if you have any suggestions for future webinars. Let us know what you’d like to hear about, who you’d like to talk to. We’ll keep the conversation going! Thanks everybody!