Modern DNA-analytical methods such as breed analysis by Generatio can measure to what extent a dog attributed to a particular breed actually carries the genomic components typical of that breed.
In breeding planning, the analysis makes it possible to identify dogs from other subpopulations that meet the breed standard in appearance and additionally carry genomic components that can enrich your own population.
There is no generally accepted definition of „breed"; the term is nonetheless widely used to classify dogs and other populations.
For the current systematics of dogs the most important body is the FCI; the Vertebrate Breed Ontology portal of the EMBL-EBI is also very helpful.
According to the FAO, a breed is a sub-specific group of animals with definable and identifiable external characteristics that permit visual appraisal (> phenotype) and separation from similarly defined groups. Which characteristics these are in each case, and how they are assessed, is not regulated and remains fluid.
A dog belongs to a breed if it is entered in that breed’s studbook and its parents were. For admission, the parents undergo a phenotype assessment. The breeding organisation and its studbook define membership, not the genome. DNA-based parentage verification supports the correctness of the studbook entry.
Dog breeding outside any association is not regulated. The individual breeder alone decides which breed name to assign. Depending on their focus, some follow the recognised breed clubs of the breed in question; others have specialised in breeding hybrids, or set out to realise their own conception of a particular breed.
Whether Great Dane, German Shepherd or Chihuahua: only a few specific gene variants underlie the manifold dog breeds – for most traits studied, no more than three loci explain the bulk of the variation. The specific genomic background arises above all from the restriction to selected breeding animals in closed studbooks. When an already established breed is bred on with animals kept apart by geography, regulations or working purpose, their differences give rise to subpopulations of their own. Breed analysis can uncover those differences and make them usable.
Where a breed is established in a country from a handful of imported dogs, that population carries only part of the breed’s diversity from the outset. Founder effects account for much of the differentiation measurable today between populations on different continents.
Italian Greyhound and Shetland Sheepdog show clear geographical subdivision: US populations form clusters distinct from their European counterparts. Within a single country, differential breeding strategies split show and working lineages in the same way.
Studbook entries rest on what was reported. Where parentage is not DNA-verified, a misassignment stays undetected – and with it the genomic share it introduced. It does not disappear: it is passed on and segregates unrecognised through the population, invisible in both the phenotype and the paperwork.
Breed analysis compares a dog’s genome against the reference background of its breed. It therefore answers more than „belongs or does not belong": it shows to what extent genomic components other than the breed-typical background are present.
By phenotype the two dogs are equivalent: each of them looks like a true Dobermann. Animal B, however, carries the greater genomic diversity and would therefore be an enrichment for your own population – provided the inbreeding analysis and the mating assessment do not argue against it.
The analysis quantifies to what extent the genomic components characteristic of the breed are present in the dog. A yes/no question becomes a graded statement – comparable between animals and traceable across generations.
It equally shows whether, and to what extent, genomic components are present that do not belong to the breed’s reference background. After a few generations such components are no longer visible in the phenotype – in the genome they remain detectable. No intent need lie behind them: they may have entered the population through a parentage record that was never verified.
The result also shows how close the dog is to the reference background, or how clearly it departs from it. Deviation is not a flaw: it may point to a distinct line or to a subpopulation from another country – in other words, to exactly those dogs that are of interest for refreshing your own population.
The starting point: a male from another population looks promising on individual assessment – both in his breed background and in his own inbreeding level.
How much of that actually reaches the puppies, however, is decided not by his own inbreeding value but by his relatedness to the intended dam. Every candidate assessment therefore has to include the question of what the mating does to the inbreeding level of the offspring.
The genomic inbreeding coefficient GIC (FROH) states how much of the genome is homozygous in the candidate himself. That is a statement about him – on its own it says little about the planned mating.
The expected inbreeding level of the puppies follows from what both parents share at the same positions. A candidate with a higher own value can therefore be the better choice if he has little in common with the dam.
The aim is not the lowest value at any cost. The breed-typical homozygous regions are to be preserved – what is lowered is homozygosity outside them. It is the genome-wide evaluation that makes this distinction possible.
Breed analyses are often advertised with the number of breeds they cover. Some providers promote more than 400 breeds – suggesting that almost every dog breed recognised worldwide can be told apart. The published data do not support that.
The most comprehensive publicly available study covers 1,346 dogs from 161 breeds. 146 of them could be clearly delimited as a breed of their own; 15 could not. One large current reference panel distinguishes 114 breeds; its authors name missing reference data explicitly as a limit on their assignments.
For the remaining breeds of the roughly 400 recognised worldwide, no comparable published basis exists.
Many breeds sit genomically very close to one another: they share a common breeding history and differ only across small parts of the genome. The smaller that distance, the more reference animals it takes to differentiate two breeds reliably.
A list covering many breeds is therefore no mark of quality in itself. What counts is how many animals stand behind each individual breed – and whether a separation that is reported holds up at all.
We assess every dog examined at three levels of resolution, with respect to its composition and its genomic similarity:
These breeds have a genomic composition that yields a clear assignment against the reference data. Our result names the individual breed or breeds and their share of the total genome of the dog examined.
Combined are breeds that sit so close genomically that they cannot be separated individually with confidence. The group itself is clearly delimited – against the distinct breeds and against the other breed groups.
Examples: Spitz breeds with their size varieties – bully type with Pit Bulls and Staffordshires – Molossers – wolves – French scenthounds – Coonhounds
These groups comprise, for example, the local village dogs of various regions. They are mixed-breed dogs that have arisen from a wide range of local original and imported breeds and carry a genomic background of their own. At present we maintain 9 such pools: with a wide genomic spectrum, but a clear delimitation to the outside. Examples: Central Asian livestock guardian dogs – European livestock guardian dogs – Turkish livestock guardian dogs – African village dogs – East Asian village dogs – Latin American/Pacific village dogs – Middle Eastern/North African village dogs – Southeast Asian village dogs – South Asian village dogs
Note: our data material from South America, Asia and Africa is still being built up. As the number of these so far rather exotic references grows, individual breeds may be taken out of a pool and new pools may be added.