Genetics has been revolutionized in the last decade by the development of amazing new technology. For decades, geneticists relied on markers called microsatellites, which are stretches of DNA with repeated element that can differ from individual to individual. In closely related animals, the repeated pattern would be similar, whereas the pattern would be more and more dissimilar in animals that are less related. 
​

For many years, SNP technology was very expensive and beyond the reach of the layperson. But the cost has come down dramatically in the last few years, and you can now send off a sample of spit to any of a number of companies that will tell you where your ancestors lived, which of your genes are links to particular traits or diseases, and even the nationalities in your background. What started as "recreational" genomics is now driving advances in diagnosis and treatment of diseases.

SNP technology offers some amazing opportunities to learn about the genes in dogs. Companies are popping up that offer to identify the breeds in a mix (some with better success than others). For breeders, high resolution SNP genotyping is providing data that can be used with powerful analysis tools in molecular biology and population genetics to improve our ability to select for the traits we want and against the ones we don't more efficiently than ever before. In fact, livestock breeders can now take a DNA sample from an animal at birth and know the traits the animal will have as an adult. Could we be using these techniques to breed better dogs?

How we can use SNP genotyping to manage genetic disease
​
We can use the power of high resolution SNP analysis to reveal the details of genetic relationships among animals. A potential sire that might look like an outcross to your bitch on paper can instead be very similar genetically because of relationships among ancestors many generations back in the pedigree. When you are  choosing a sire to improve a trait, reduce the risk of genetic disease, or increase the genetic diversity in your line, reliable information about the DNA in the dogs can be the difference between success and failure. Let's look at an example of how we can use information about genetic relationships to manage a genetic disorder in dogs.

An organization breeding Labradors for guide dogs was having a problem with elbow dysplasia (ED) (Ubbink et al 1998). They had good pedigree records and used these to map the genetic relationships among the dogs in their breeding stock using kinship to determine relatedness and a technique called cluster analysis to form clusters of related dogs.

This is a "dendrogram", a diagram similar to the branches of a tree that displays the genetic relationships among animals. On the right side of the chart is a numeric scale of relatedness. The shortest branches indicate the highest similarity between individuals; so the very short branches at the level of 0.125 (red arrow), reflect relatedness comparable to a half-sib mating. The next number, 0.0625, corresponds to the relatedness (inbreeding) produced by mating first cousins, and the numbers above that correspond to successively lower degrees of relatedness.

Using SNP genotyping to map genetic relationships

Many breeds are highly inbred and have low genetic diversity. Will this technique using DNA to establish genetic relatedness work for purebred dogs?

I have gathered some DNA data in public archives from published studies that were analyzed using high density SNP genotyping, then used cluster analysis (as was done above) to produce the map of genetic relatedness for a number of breeds. These are displayed below, and many show that there are genetically distinct subpopulations within a breed that could be exploited to look for patterns in the expression of genetic disorders. We know nothing about who these dogs are, but the charts show that there is sufficient genetic differentiation in a breed to use for this type of analysis. On each one I have indicated with a red square the first major division to make it easier for you to visualize the groups. For some of these we clearly need more data, and for all of them we need data for known dogs. But this should give you a good idea of what some of these look like.

Can we do this for your breed?

Here we describe a very powerful technique to sort out genetic relationships among individuals that can be used to improve genetic management of disorders without knowing the genes involved. There is a long list of inherited disorders that breeders are struggling to manage in the absence of genetic information for guidance. Epilepsy, cancer, heart disorders, kidney disease, neurological problems, degenerative myelopathy, and the list goes on and on. We're not making much progress.

We can do this now. We need DNA genotype data for current dogs, including those in the breeding population, to establish the genetic subgroupings within the breed. We need the incidence data for the disorders of interest.

​It might not take a lot of data to reveal the information we need. In the example of Laboradors above, data from just a few dogs would indicate the location of the cluster containing the affected dogs. You could take action using this information far sooner than if you needed 20 affected and 20 clear dogs to search for genes. 

If you're interested in using this powerful technique for your breed, please contact us. We can set up a Genetic Management Workshop for your breed where breeders and PhD population geneticists with expertise in canine genomics and breeding can discuss goals and resources, and work together to develop a breeding plan to address problems. 

We now have amazing tools to help breeders manage the growing list of genetic disorders in dogs. Let's use them!

Even 100 years ago, breeders of domestic animals realized that continuous inbreeding resulted in deterioration in the quality of their stock. To remedy this, every few generations a relatively unrelated animal would be crossed in to the population to freshen up the gene pool and reduce the level of inbreeding. In fact, the coefficient of inbreeding was devised in the 1920s as a way to estimate the level of inbreeding so breeders would know when it was time to do an outcross.

Many modern dog breeders follow a similar regime, breeding close for a few generations then out to a relatively unrelated dog. But is this as effective as we think it is?

For one thing, our dogs are much more inbred than they were a century ago. Many breeds have average levels of inbreeding above 20%, and too many are even higher. With such high levels of inbreeding and low genetic diversity, the dogs in a breed can all be more similar genetically than half- or even full siblings - or even more. How much can you expect to gain from an "outcross" that really amounts to close inbreeding?

They found that a single outcross followed by one or more generations of backcrossing reduced the beneficial effect of the outcross and sometimes even increased the rate of inbreeding. In the cross breeding, the first generation receives 50% of alleles from each of the donor and recipient parents. But in subsequent backcrosses, half of the donor genes are lost on average, so that after a few backcrosses the remaining donor contribution was inconsequential.

This result is not unexpectded, but it should give pause to those that undertake a crossing program that consists of a single donor cross followed by serial backcrosses. Kennel clubs typically prefer 3 backcrosses before admitting the descendants of a crossbreeding program to the studbook. But the simulations show that the required backcrosses essentially flush the donor genes from the recipient gene pool and there is no net benefit. Indeed, there are multiple examples of crossbreeding programs that failed or produced little benefit because the introduced genetic variation was subsequently lost.

Regular outcrossing was the most effective strategy for reducing the rate of inbreeding in the Saarloos Wolfhond, but still the outcomes of the simulations were highly variable. The scenario of regular outcrossing generally reduced the rate of inbreeding, but the inbreeding rates were highly variable, and in some simulations all of the donor genetic contribution was lost. 

What this study clearly showed is that using outcrossing to reduce the rate of inbreeding requires careful management in order to ensure success, and the benefit achieved will be small unless crossing is done on a regular basis. One key to success is to follow outcrossing with strategies to increase the effective population size.  Otherwise, inbreeding will quickly return to previous levels. 

This outcross simulation used a completely unrelated (fictitious) breed as a donor. For breeders that are crossing to what they hope are relatively unrelated dogs in their own breed, the benefit will be much less if the level of relatedness among individuals is high. 

You can now enroll your breed in the new ICB Genomic Management Workshops, in which breeders and expert scientists will work together to assess the genetic status of your breed and design effective breeding strategies to reduce inbreeding, increase genetic diversity, and improve the health of the dogs.

We are accepting breeds now, so don't miss this opportunity to be one of the first we work with in the flagship program. You can read more about the new ICB Genomic Management Workshops HERE. 

Don't miss out on our "Incentive" offer of a free course for each participant for the first three breeds enrolled!

REFERENCES

Windig JJ & HP Doekes. 2018. Limits to genetic rescue by outcross in pedigree dogs. Journal of Animal Breeding and Genetics 135:238-248.

It's the classic scene. The family sits down at the dinner table and Billy grimaces at the food on his plate. He hates peas. HATES peas. So while acting non-chalant and seeming to really dig into his meal, Billy was actually deftly using his fork to push one pea to the other side of the plate, and another up under the thigh bone from his drumstick. A few disappeared under the pile of mashed potatoes, one leaped from the plate and fell to the floor, and soon it looked like he had done a decent job of getting down most of the despised peas.

But Mom is no fool. "Billy", she says sternly, "eat your peas instead of just pushing them around on the plate!" Busted.

We risk doing a similar thing when making decisions about the ideal sire and dam for the next litter of puppies. Let's say you have a bitch you want to breed and you're considering four or five potential sires. You also have some DNA test information that allows you to compare the genetic similarity of the bitch with each of the sire candidates. Selecting the sire that is least similar to the dam should produce puppies with the lowest average inbreeding.

But does this tell you everything you need to know? Are you really improving genetic diversity? Or are you just pushing the peas around on the plate?

Let's consider 2 things. First, let's look at genetic relatedness.

By "relatedness", we refer to the genetic similarity of a dog to the other dogs in the population. Closely related dogs will be more similar; distant relatives will be less similar. 

There will always be genetic differences between two dogs, no matter how closely related they are, because no two puppies will inherit exactly the same alleles out of all the zillions of possible combinations. So even among closely related dogs, you should always be able to identify a dog that is "least" related, and if you're shopping for sires this one might look like the best bet. 

The problem here is that you can identify a dog that is less related compared to the others, but it might still be very closely related - much more closely related than you would risk breeding to if you knew the actual genetic level of relatedness (First cousins? Half-siblings? Full siblings???).

Let's consider an example. We have a population of about 80 dogs in a particular breed, and you are sorting through them to identify which of the sires you are considering is least similar to your bitch. You make your choice and since they don't share any common ancestors for several generations, you're happy.

But the kinship coefficients of the dogs would tell a different story. The kinship coefficient for a pair of dogs is the fraction of their genes that are identical by descent. This sounds like it's related to inbreeding, and it is. In fact, the inbreeding coefficient of a dog (the fraction of the genes in an individual that are identical by descent) is equal to the kinship coefficient of its parents. Since we know that most genetic disorders in dogs are caused by recessive mutations, the kinship coefficient also tells us the risk of producing a genetic disorder caused by a recessive mutation in the puppies.

If you determine the kinship coefficients between the bitch and each of the potential sires, you choose the one with  the lowest kinship coefficient. However, the kinship coefficient reveals that this "least related" dog is actually more similar to the bitch genetically than you would expect of a full sibling. The data for the kinship coefficients with these potential sires are:

Sire number 5 has the lowest kinship coefficient so is least related. But that kinship coefficient indicates that he and the bitch are more similar genetically than you would expect for full siblings, which would have a kinship coefficient of 0.25.

Wow. You wouldn't do a full sibling mating, and this mating would be a similarly close inbreeding. From the data for the other sires, it looks like there is very high relatedness among many of the dogs in the population, whose kinship coefficients go as high as 0.693!

When you compared the dogs relative to each other, one dog looked like a good choice. But after looking at the actual kinship coefficients of the potential parents, you decide that the marriage is much too close for comfort.

Above, I noted that there were two things we needed to consider when making mate selections based on information about relatedness from DNA data. The first was the absolute, and not just the relative, genetic relatedness.

The second issue comes from the first. Genetically similar dogs will have similar mutations. The least related pair based on kinship coefficients will nevertheless share many mutations if they are as similar as full siblings. They might be from different lines and not share any kin for many generations, but genetic similarity brings with it high risk of producing a genetic disorder. Mating of full siblings, with a kinship coefficient of 0.25, has a 25% risk of pairing up two recessive mutations.

Let's look at another example using this same breed. Below is a "dendrogram", a family tree that is based on kinship coefficients determined using a high-density (> 200,000 markers) SNP panel that has considerable power for resolving relatedness among closely related individuals.

On the panel on the left, I have indicated the results of genetic testing for the mutation (VWF) responsible for von Willebrand's disease in this particular breed. Individuals that tested clear have a little "c" at the left end of the branch. Individuals that are carriers of the mutation have two asterisks (**), and dogs that are homozygous (and therefore affected by the disorder) are indicated with four asterisks (****).

You should be able to scan from the top of the chart to the bottom and easily pick out the clear, carrier, and affected dogs. What you will probably notice is that the VWF mutation is widespread in this breed. If there wasn't a test for this disorder, you might think you're avoiding it by choosing dogs that are not closely related to each other. But, as we saw above, all of these dogs are very similar genetically, regardless of how unrelated they might appear on paper.

By Carol Beuchat PhD

ICB is proud to announce a new program for breeders, the ICB Genetic Management Workshops. Now for the first time, breeders can get a scientific assessment of the genetic status of their breed as well as professional guidance in setting up a breeding program aimed at improving the health of their dogs. 

The Workshops will partner breeders with veterinary and population geneticists to do a detailed assessment of the genetic health their breed using both DNA and pedigree information. Then the scientists will work with the breeders to design a breeding program that will address the genetic issues and provide a foundation for sustainably breeding healthy dogs into the future.

The mission of ICB since it was founded in 2012 is to provide purebred dog breeders with the tools and information they need to make breeding decisions that will improve the health of dogs. Breeders will have unparalleled access to guidance from world experts in canine genetics, as well as the tools and expertise of conservation geneticists.

About the Workshops
The ICB Genetic Management Workshops will be ICB's flagship program and the first of its kind. In these Workshops, breeders and scientists will sit down together, evaluate the genetic status of the breed, diagnose the problems, and come up with breeding strategies and recommendations that will improve health. The goal will be to provide a plan that is feasible for breeders to follow and which will have short- and long-term outcomes for health improvement. Genetic assessment of a breed and development of breeding strategies will use the the new ICB Breeder Tool. It will provide breeders with a detailed assessment of the genetic health and issues of their breed and the information they need for making the best possible breeding decisions going forward.

Genetic Assessment from DNA
The genetic assessment of the breed will use DNA genotyping and mutation testing information. The current commercial provider of suitable data is Embark Vet, which uses a state-of-the-art, high-density Illumina SNP analysis with more than 200,000 markers distributed over the entire genome, including all 38 autosomes, the sex chromosomes, and mitochondrial DNA. This is the same platform used by canine genetics research laboratories worldwide and provides the most comprehensive and detailed analysis available. Because this platform is so widely used, marker data produced by research laboratories can be added to our databases to increase our sample size and the breadth of the gene pool represented. 

Genetic and Demographic Assessment From Pedigree Data
Assessment of the genetic status of the breed should ideally include a pedigree database, which provides information about the genetic history of the breed, founder dogs, bottlenecks, changes in population size.  Information from the DNA analyses can be used to verify pedigree relationships. In the absence of a useable pedigree database (e.g., because of errors, missing data), the DNA can be used to establish the genetic relationships of the current dogs, which will provide the information needed by breeders for genetic management going forward.

A good pedigree database, or genealogy reconstructed from DNA data, can be used to locate the genes causing a genetic issue without needing to identify the specific genes responsible. This is especially valuable for complex and polygenic disorders for which it would be difficult and expensive to identify the causative alleles (e.g., cancer, epilepsy). A pedigree database also provides the foundation for estimated breeding values (EBVs), which substantially improve the efficiency of selection.

Our Goals
The Breed Workshops are not courses, although the participants will surely learn a lot. Our overriding goal is to solve problems and provide breeders with the guidance necessary to work towards healthier dogs when planning their very next litter. 

Commitment
The Breeder Workshops involve a partnership with a shared commitment from both the ICB team that is working with the breeders, as well as from the breeders, who need to be willing to commit to implementing the plan produced by the workshop. If breeders cannot make a commitment to implement the strategies they have developed with the assistance of the scientific team, we will not accept them for a workshop. We want to see RESULTS that make a significant difference in the health and welfare of dogs.

For this reason, we have some basic requirements before we will agree to run a workshop for a breed. Any plan will require  the breeding of dogs, so we must have sufficient breeders involved to support this. We will also need to know that the puppies produced can be placed in homes that will properly care for them and retain them until they can be bred if they are suitable. This will probably require many non-breeder participants. People that have left breeds because of  their  health issues might be ideal for raising puppies produced in the program.

The minimum involvement for a workshop is 50 participants with a fee of $200 ea. These can be breeders as well as non-breeders who wish to support the effort. For large breeds, it will take a substantial commitment by breeders to make a meaningful improvement in the genetics of the breed. A few enthusiastic breeders cannot make a significant difference. This will need to be a substantial, collaborative effort with a long-term commitment. With proper planning and scientific support, the breeding program has the greatest chance of success.

Arranging a Breed Workshop
Any group of breeders can request a workshop for their breed. It would be ideal to have club support, but if that is not forthcoming we will work with a group of independent breeders that can meet the requirements for participation and commitment.

We will need the DNA genotyping results from Embark (or other sources using the Illumina canine SNP chip) for a sufficient number of dogs to be able to do meaningful analyses. For some breeds, there are data already available for anonymous dogs from research studies to provide baseline information about the breed. Participants will add to this the genotyping and mutation/trait testing information for current dogs as necessary to fill any gaps and to provide the information needed about the current breeding population of dogs. There is a table below with a current inventory of data, to which data will be added on an ongoing basis from research studies as well as dogs in the breeding population.
​

How Long Will It Last?
If we have the necessary information organized properly ahead of time, we anticipate that analysis and design of a breeding program should take 8 to 10 weeks. But we'll keep working until we're done. The final product is a breeding strategy to be implemented by workshop participants.
​

What You Will Need
This will be work, for both the participants and the scientists involved. It will be a commitment of time as well as learning so that participants understand the rationale for the breeding plan and its goals. You will need enthusiasm and dedication to the project for the long term if you are truly going to make a difference.

We will need the genetic data files from analysis using the Illumina SNP chip, either from Embark or research projects using the Illumina chip. DNA analyses from the Davis genetic diversity program use microsatellites (STRs); the data are not compatible with data from other sources and lack the resolution of the SNP data. Mutation testing from MyDogDNA/Mars can be used, but their DNA genotyping data is not compatible because they use a proprietary chip. So, stick with Embark if you do any commercial DNA analyses.

We will be accepting breeds for workshops on a first-come, first-served basis. We need a commitment with payment for at least the minimum 50 participants at $200 ea to get a reserved spot on the calendar. If small breeds are unable to meet the requirement of 50 participants, please contact us to discuss whether we can design a program for the breed with fewer breeders.
​

Data Already Available
DNA data already available for some breeds that can be used in analyses. We have data for most AKC breeds and at least 50 dogs of these breeds (# in parentheses):

• American Cocker Spaniel (70)
• Border Collie (50)
• Boxer (160)
• Cairn Terrier (60)
• Cavalier King Charles Spaniel (50)
• English Setter (80)
• English Springer Spaniel (50)
• German Shepherd (130)
• Golden Retriever (250)
• Irish Wolfhound (210)
• Labrador Retriever (630)
• Maltese (80)
• Miniature Schnauzer (60)
• Rottweiler (80)
• Vizsla (90)
• Yorkshire Terrier (180)​

Enroll for a Genetic Management Workshop
We are scheduling workshops on a first-come, first-served basis. You will need a minimum commitment of 50 participants. 

Please contact us to discuss setting up a Genetic Management Workshop for your breed.

​
Contact:

Carol Beuchat PhD
Scientific Director
Institute of Canine Biology
carol@instituteofcaninebiology.org

To learn more about the genetics of dogs, check out
ICB's online courses

***************************************

Visit our Facebook Groups

ICB Institute of Canine Biology
...the latest canine news and research

ICB Breeding for the Future
...the science of animal breeding
​

Sewell Wright, who first formulated the method for estimating inbreeding from pedigrees (hence, "Wright's Coefficient of Inbreeding"), also described another way of assessing inbreeding using something called a "fixation index", or Fis.

In a large, genetically diverse population in which mating is random, you would expect inbreeding to occur only rarely because the probability of two related individuals mating is low. If we were to find that there was a higher level of inbreeding in this population than what would be expected just by chance, that would be evidence of non-random mating of related individuals; i.e., preferential inbreeding.

If we know the allele frequencies of a particular marker in the population, we can estimate the genotypes to expect under random mating. These would be predicted by the "Hardy-Weinberg Equation" (p^2 + 2pq + q^2), where p and q are the frequencies of the two alleles for the marker of interest. We can use the Hardy-Weinberg equation to predict genotype frequencies under random mating in a population. If we measure the actual genotype frequencies and find them to be different than expected based on H-W, then we know that mating has not been random.

We can use this tool to detect non-random mating using Wright's fixation index, Fis. Animals in a population with a level of inbreeding higher than expected based on H-W will have a positive Fis statistic. Animals that are outbred relative to what is expected under H-W will have a negative F statistic. An Fis of zero indicates a level of inbreeding equal to the expected value based on allele frequencies in the populaion.

In a nutshell, a positive Fis indicates inbreeding; a negative Fis indicates avoidance of inbreeding.

We can use our thousands of SNP markers to estimate Wright's Fis statistic for a population of individuals in a breed. The graphs below show the distribution of inbreeding coefficients (Fis) in Boxers and Irish Wolfhounds. For both breeds, there is "excess homozygosity" - a higher incidence of homozygosity on average than you would expect to see in a randomly breeding population. In both breeds, breeders are preferentially mating dogs that are more related than average in the population to produce higher levels of inbreeding.

 
These data show that better choices when selecting breeding pairs could slow the rate of increase in inbreeding in both breeds and also lower the risk of genetic disorders caused by recessive mutations. 

The Coefficient of Inbreeding (COI)
Most breeders are familiar with Wright's Coefficient of Inbreeding (COI), which can be calculated from pedigree data. The COI provides an estimate of the predicted level of inbreeding based on the simple assumptions that alleles are inherited independently and with equal probability. From this, you can predict the average COI of the litter of puppies produced by a pair of dogs. Because COI is based on the probability of inheriting alleles across all generations of a pedigree, it cannot predict the COI of a dog exactly; only the average and the expected variation. However, it usually correlates well enough with "realized" inbreeding (based on DNA homozygosity) that it remains the single most useful statistic used by animal and plant breeders today to assess the level of inbreeding in an individual or the inbreeding to be expected from a particular breeding.

Runs of Homozygosity (ROH)
The most common method of determining inbreeding from DNA data uses a technique that quantifies the amount of inbreeding represented in "runs of homozygosity" (ROH), which are regions of the chromosomes where there are many consecutive homozygoous loci (Ceballos et al. 2018; Curik et al 2014).

The homozygosity produced by inbreeding is not randomly scattered all over the chromosomes. Selection, both natural and artificial, will produce "hot spots" of homozygosity in the regions of the chromosome where there are genes under selection (Sams & Boyko 2018). If the genes remain under continuous selection over generations, as would be the case for the genes for type in a purebred dog population, the blocks of homozygosity will tend to get longer and longer each generation (Kim et al. 2015). 

​Let's look at some examples.

There are panels below for two breeds. The horizontal strip at the top represents the chromosomes lined up end to end to form a continuous strip the width of this page, with each of the autosomal chromosomes numbered consequitively from 1 to 38.

I've blown up a part of the strip to show you the first 7 of 38 chromosomes.
​​
Below this is a chart depicting the blocks of homozygosity for about 80 Coton de Tulear, with each dog represented by a horizontal strip with dog IDs indicated on the left. Below this is a similar chart for a similar population of Stabyhouns. SNPs that are heterozygous are pink, while homozygous SNPs are blue.

​You can see scattered across the chromosomes for the individual dogs that there are blocks of blue of various sizes. These are the runs of homozygosity. 

Shared Homozygosity
Something you will notice is that there are regions that are homozygous in multiple dogs, which is evident because they appear as vertical blue line across the rows where these regions all line up. There is an obvious one in chromosome 23 in the Stabyhoun, which shows up as a blue stripe about 25% from the right edge of the chart. There are other areas of shared homozygosity here and there in both breeds.
​

Besides usefulness as a measure of inbreeding, runs of homozygosity are also relevant because they are enriched for deleterious mutations. That is, mutations get trapped in blocks of inbreeding and cause the genetic load of detrimental mutations to increase over the generations. If breeders can see where blocks of homozygosity are in the genome, they can select sires that do not share the same ROH, which will prevent the blocks from being passed on to offspring. The two individuals might have the same amount of inbreeding but distributed differently across the genome, and being able to visualize the actual inbreeding on the chromosomes can allow breeders to break up blocks using the heterozygosity at the same place in other parent.

Although there seems to be general agreement that ROH currently the method of choice for estimating the level of inbreeding from DNA, there is still discussion about the size of the blocks of homozygosity to include in the calculation (Sams & Boyko 2018). Certainly, the longest blocks represent the most recent inbreeding and should be considered, but should older inbreeding also be included? This can amount to a substantial difference in inbreeding estimates, and as yet there is no consensus on what provides the best estimate. In the meantime, comparisons should be made among individuals or between breeds using the same minimum block size, which should provide comparable estimates of the most recent inbreeding.

You can learn more about runs of homozygosity and how to use ROH to estimate inbreeding in the two-part tutorial below.

ROH Tutorial: part 1
​ROH Tutorial: part 2

REFERENCES

Curik I, M Ferencakovic, & J Solkner. 2014. Inbreeding and runs of homozygosity: a possible solution. Livestock Science 166: 26-34.

Ceballos FC, PK Joshi, DW Clark, M Ramsay, & JF Wilson. 2018. Runs of homozygosity: windows into population history and trait architecture. Nature Reviews Genetics 19: 220-234. 

Kim E-S, TS Sonstegard, CP Van Tassell, G Wiggans, & MF Rothschild. 2015. The relationship between runs of homozygosity and inbreeding in Jersey Cattle under selection. PLoS ONE 10:e0129967. https://doi.org/10.1371/journal.pone.0129967.

Marsden CD, D )rtega-Del Vecchyo, DP O'Brien, and others. 2016. Bottlenecks and selective sweeps during domestication have increased deleterious genetic variation in dogs. PNAS 113: 152-157.

Sams AJ & AR Boyko. MS. Fine-scale resolution and analysis of runs of homozygosity in domestic dogs. bioRxiv preprint (7 May 2018). doi: http://dx.doi.org/10.1101/315770.

By Carol Beuchat PhD

One of the most basic things to know both about a breed and an individual dog is genetic diversity. In a breed, genetic diversity is the variation in DNA that results in the differences in traits from individual to individual. Genetic diversity also allows populations to adapt or evolve in the face of changes in the environment or disease. 

In an individual animal, genetic diversity is realized in the fraction of loci that are heterozygous. Inbreeding increases the fraction of loci that are homozygous, and this in turn raises the risk of the expression of a recessive genetic disorder. Homozygosity also results in an overall deterioration in health referred to as "inbreeding depression". So, at both the individual and population (i.e., breed) levels, it is important for breeders to be monitoring genetic diversity.

There are several different ways to assess genetic diversity using SNP data. Below I will describe how each is estimated, what the data mean, and how breeders can use the information.
​

A very common method of assessing genetic diversity from DNA data is by computing heterozygosity.

A locus with two copies of the same nucleotide (e.g., AA, aa) is homozygous. A locus with two different nucleotides (Aa) is heterozygous. This is part of the data spreadsheet from SNP genotype which lists the genotypes found at 4 SNP markers (e.g., BICF2G...., etc) on the chromosomes of 12 dogs (one in each row). 

For the first marker (column 618), you can see that some dogs are homozygous for that SNP (either AA or GG), while some are heterozygous (AG). For the next two SNPs, all but one of the dogs are homozygous for the same nucleotides, and all of the dogs are homozygous at the last marker.

​We can determine the degree of heterozygosity of each of these markers by tallying up the different combinations of alleles for each marker in each of the dogs in our sample. From these information, we can then compute heterozygosity, which is simply the fraction of all of the marker that were heterozygous, averaged over all of the dogs. The higher the heterozygosity, the greater the genetic diversity. This is usually referred to as "observed heterozygosity" (Ho).

The graphs below show the frequency distribution of observed heterozygosity (Ho) determined for dogs in four different breeds: the Labrador and Golden Retrievers, the Coton de Tulear, and the Stabyhoun. You can see immediately that although there are dogs in both of the retrievers with low heterozygosity (0.2 or lower), there is some variation in the population such that the distribution extends up to about 0.35.  The average and minimum Ho are similar to these in the Coton de Tulear, but the highest heterozygosity is about 0.3. The population of Stabyhoun has the lowest heterozygosity with most individuals less than about 0.25 and little variation around the mean of 0.2.

The data for heterozygosity for these four breeds reveal that the Labrador and Golden have the highest average heterozygosity, the Coton de Tulear, somewhat less, and the Stabyhoun the least. Individual dogs with the highest values for Ho will produce puppies with lower levels of inbreeding on average than dogs with lower Ho.
​

The world of the dog breeder is undergoing a sea change. We have had DNA tests for individual mutations for   a few years, but now (finally!) breeders can get a genotype that documents the DNA of all 38 autosomal chromosomes, the two sex chromosomes, and even the mitochondrial DNA. This is a massive amount of data - the file for a single dog has one row and more than 200,000 columns. 

These DNA tests are using what is now the gold standard for plant and animal genotyping. This technology uses markers called single nucleotide polymorphisms (SNPs), which are just what the name says - a locus on the chromosome at which the nucleotide present can be an A, C, T, or G. These nucleotides are the "alphabet" of DNA, a code that contains the instructions for making a protein that has some function in the body. The SNPs used as markers tend to be variable, or polymorphic, and this variation allows us to use them to estimate genetic diversity, determine the relatedness of animals, and even visualize the distribution of homozygosity (inbreeding) on the chromosomes. The massive amount of information contained in a SNP genotype, together with the reduction in cost over the last few years, have made this technology the method of choice for most genetic studies in both plants and animals. 

SNP genotyping of dogs is now commercially available at a reasonable price ($100-200), and breeders are jumping at the chance to take advantage of this new tool. But in many cases the breeder with raw data file in hand doesn't know what to do with it. The thousands of columns of nucleotide data reveal no information that a breeder can translate. So how can we make it useful?

With the right software and some expertise in population genetics, we can extract a tremendous amount of information from the file for an individual dogs, and even more if we have data for a modest sample of dogs in the breed. The technology is too new for books or online resources designed for use by the layperson or dog breeder. But I am summarizing in a series of posts the basic information breeders need to understand what their data mean and how they can use it.

First, I will review the basic tools that can be used by breeders to assess the genetic status of their breed. These evaluate genetic diversity, genetic relationships, and population structure. Then I will show how to use the data for individual dogs when evaluating the suitability of potential mating pairs. Finally, I will provide examples of some resources that breeders can use to develop breeding strategies for sound genetic management and to reduce the incidence of genetic disorders in a breed, including the use of cross-breeding programs when necessary to restore genetic health.
​

​A DNA Primer for Dog Breeders

Quick Start Guide

1) Genetic Diversity

• Heterozygosity (Ho)
• Inbreeding (Runs of Homozygosity)
• Inbreeding (Fis)

2) Relationships

• Kinship Coefficient
• Mean Kinship (genetic value)
• Kinship Matrix
• Dendrograms

3) Population Structure

• Principal Components Analysis

​

The most basic tool of the dog breeder is the pedigree. In essence, it is the documentation of the genetic and demographic  history of an animal, and it can hold a treasure trove of useful information.

Often, however, the quality of the pedigree information available for a dog is not what you would like it to be. Most often, dog pedigrees are incomplete, either not including generations back to the founding of the breed, or missing animals in the subsequent generations. Of course, there are also errors, both deliberate and inadvertent, that can change the historical picture. Finally, a pedigree database might exist for a breed but is not accessible, perhaps for technical reasons such as a computer platform that does not communicate well with the rest of the world (e.g., self-designed pedigree software), an out-of-date computer (yes, there are still data being kept on old Commodore computers...), or not computerized at all (think folders of paper). Perhaps the most distressing reason for lack of accessibility is the unwillingness of people who have custody of the information to allow public access. (This means you too, AKC.) It's hard for me to imagine how anyone could argue that this is in the best interests of the breed; it can only be selfish or ego-driven in someone who fails to see that data that can't be used might just as well not exist. These databases die with their creators, with the records being tossed out by relatives or even strangers who see no value in them. Such as waste.

For breeds without an adequate pedigree database for whatever reason, the data produced by SNP testing can be used to reconstruct the genetic relationships among all of the individuals you can get samples for. Software compares the similarity of the genotypes based on SNPs among the tested individuals. From this information, a tree can be constructed that displays the organization and degree of relatedness among the members of the population. This is based on the same information as used to create a kinship matrix.

In the chart below, I have used the SNP genotyping information for a group of dogs to reconstruct their pedigree relationships based on genetic similarity. The tree structure outside the square is the "family tree" illustrated as a dendrogram. (Learn how to read a dendrogram HERE.) The square in the middle is a "heat map" of the kinship values that depicts at just a glance the degree of relatedness of groups of animals in the breed as reflected in the dendrogram. (Learn how to read heat maps HERE.) The same animals are on both axes in this heat map, so it is symmetrical across the diagonal. (You can match up the clusters across the top with those along the left axis.)

In the absence of pedigree information, I have been able to reconstruct the genetic relationships among this group of dogs using a very large number of SNP markers that are distributed across all of the chromosomes. Where pedigree information is not available or is incomplete, this tree of relationships can be used to establish the genetics of an existing population of animals. If good pedigree records are kept going forward, they can be an excellent tool for assessing relationships among individuals and making predictions of litter inbreeding coefficients without the cost and logistic hassles of DNA analysis. The pedigree relationships can also inform the strategic choice of animals to be genotyped in the situations where that information is needed (e.g., trait or mutation testing).

​For breeders trying to breed more strategically to reduce the incidence of genetic disorders in their dogs, but might not have the best pedigree information, DNA analysis using high density SNP markers can provide research quality genetic information for a modest cost. We can convert the data for thousands of SNPs into a chart that displays the true genetic relationships among a group of dogs.