Embark’s Test for Genetic Coefficient of Inbreeding (COI): For millennia, dog breeders have intentionally bred relatives as a way to fix traits in a lineage. A century ago, Sewall Wright devised the coefficient of inbreeding (COI) as a way to measure inbreeding, a statistic still popular today. Like humans, dogs tend to be 99.8-99.9% genetically similar to other members of their species. Even other species can exhibit similarities—dogs and humans are 64% similar at the base pair level. But genetic variation is the spice of life, and the 0.1-0.2% of the genome that differs encodes a myriad of variation. Some of these variations we have intentionally perpetuated, like body shape, coat color, and behavior. Unfortunately, other less desirable variants confer potentially harmful effects on health, longevity, and reproductive success.
Harmful mutations come in three main varieties; recessive, dominant, and additive. The harmful dominant and additive mutations are quickly weeded out in large, outbred populations. This occurs because the individual carrying these mutations has a reduced fitness. Recessive mutations, on the other hand, are different. A harmful recessive mutation might “break” a gene. This has little or no consequence if an individual has a working copy of the gene from his or her other parent. However, this can have disastrous consequences when an individual inherits two broken copies. Outbred individuals almost never inherit two broken copies. Therefore, natural selection or breeders cannot effectively select against them unless there is a genetic test for the mutation. For example, if a mutation is at 1% frequency in an outbred population, any given dog has a 0.01% chance of inheriting two copies of the mutation—clearly a very small chance. As such, every dog population—or in the context of purebred dogs, every dog breed—contains an abundance of rare recessive mutations that were either present in a founder individual or arose spontaneously in the dog population sometime afterwards. These rare mutations are hardly ever problematic for outbred individuals because they almost always inherit at least one working copy; however, they can cause real problems for inbred individuals—animals that arise from the mating of closely related parents.
Let’s consider what happens with dogs in a mother-son mating. A mother passes along 50% of her genome to each pup, so each rare (<1% frequency) recessive mutation carried by the mother has a 50% chance of being transmitted to a son. Offspring from a mother-son mating would, therefore, have a 25% chance of inheriting two bad copies of the mutations that have been passed down to the son. This is a greater than 100-fold risk compared to an outbred dog! Inbreeding in dogs has real consequences. Research in the Embark Co-Founder Adam Boyko Lab has shown that a 10% increase in inbreeding can lead to a 6% reduction in adult size (poor growth) and a six- to ten-month reduction in lifespan. Reduced litter size and fertility are also likely. These risks occur from both classical inbreeding and from drift in small populations where every individual is a not-so-distant relative. Assessing these risks depends on accurately quantifying the likelihood that mutations will be identical-by-descent, or inherited from the same ancestor.
Calculating coefficient of inbreeding (COI)
There are three ways to quantify the coefficient of inbreeding (COI): (1) Using a pedigree, (2) Trying a small set of polymorphic markers, or (3) Testing a genome-wide marker panel. How do you easily find out?
Pedigree-Based coefficients of inbreeding are based on the relatedness of individuals in a pedigree. 25% is the value from a mother-son or full-sibling mating; 12.5% being the value from a grandparent-grandchild or half-sibling mating; and 6.25% being the value from a first cousin mating. These values accumulate. Logically, all individuals have COIs between 0% (completely outbred) and 100% (completely inbred). So, three generations of full-sibling matings would lead to a COI of 50%. Ideally, the pedigree is complete all the way back to the founding of the breed. However, in reality, most pedigrees only go back, maybe, 5 to 10 generations. Most COI calculators assume that the original ancestors in the pedigree are unrelated. Therefore, a COI calculated from a 5-generation pedigree could be much lower than that calculated from a 10-generation pedigree. This is likely much lower than the true COI if the complete pedigree back to the breed founders was known. For this reason, there’s no one answer for what a “good” COI is; it all depends on how complete the pedigree is. Furthermore, because of the principle of segregation, two individuals with identical expected COIs from a pedigree may have very different levels of inbreeding. This depends on which individuals inherit which chromosomal segments.
Marker-Based Inbreeding uses dozens or hundreds of widely spaced markers to estimate inbreeding. Each marker can be heterozygous or homozygous (identical by state). The overall locus heterozygosity (HL) of the panel is generally correlated with inbreeding. However, the absolute values of HL depend on the markers that are chosen. Because a rare marker being homozygous is stronger evidence of inbreeding (identity by descent) than a common marker being homozygous, different weightings may be used to calculate statistics like internal relatedness (IR). This varies from -1 to 1. Nevertheless, most of the genome is not linked to any marker. Therefore, estimators do not detect most inbreeding tracts. As a result, it makes marker-based estimators poorly suited for differentiating between individuals with similar COIs (less than 5-10% different).
The genome-wide genetic coefficient of inbreeding is the gold standard for measuring inbreeding. It requires at least tens of thousands of markers spread across the genome. The Embark Dog DNA kit includes Genomic-wide COI results. With this resolution, the actual inbreeding tracts can be directly observed as tracks of homozygous markers. Above a certain size, these runs almost always represent identity-by-descent, and thus we can easily calculate the coefficient of inbreeding (the proportion of the genome that is identical by descent). At Embark, we use about 1 million basepairs, known as 1 centimorgan, as the minimum size of each track. This is because we are interested in inbreeding all the way back to a breed’s founding; remember, for most domestic dog breeds, this is usually 50-100 generations ago.
Calculating the genetic coefficient of inbreeding directly using genome-wide data has several advantages. It doesn’t require a pedigree. Also, it doesn’t depend on marker frequencies or require complicated statistics to correct for rare/common markers. And finally, it is directly comparable across studies because it doesn’t depend on the specifics of the markers used or the populations being studied.
Inbreeding tracts are apparent using genome-wide data. Pedigree-based and marker-based estimators often miss these tracts. Comparing an individual to the genetic coefficient of inbreeding distribution for the breed lets you know whether a dog is more or less inbred than expected for its breed. You can visualize the inbreeding tracts to see where in the genome they are found. Accurate determination of inbreeding tracts is crucial for identifying recessive disease mutations through homozygosity mapping. It is also crucial for more precisely understanding the risks of inbreeding within and across breeds. Although some level of inbreeding cannot be avoided for most purebred dog breeds, and inbreeding risk shouldn’t be the only consideration when selecting mates, reducing the inbreeding load in a population is a valuable goal.