Sequencing proposal: what has the Great Dane gene pool kept?

October 14, 2026 · Not for distribution. Figures are expectations from recorded pedigrees.

Summary

We propose sequencing a chosen set of living Great Danes to measure how much of the breed's historical genetic diversity survives, and which dogs carry what is at risk. A pedigree analysis of 118,706 dogs shows that 58 ancestors supply half the genome of today's dogs, while 1,449 other old ancestors survive only as a trace, holding 1.3% of it. A pedigree can only predict who carries those lineages; sequencing would measure it.

We would choose the dogs for how much unheld ancestry each adds. At 100 dogs that holds 88 ancestor genome equivalents, against 74 for random dogs and 50 for the most-connected dogs. The result would give a preservation plan with specific dogs, matings and stored samples in it.

The question

Preservation breeders keep two things at once, and only one is rewarded. The breed standard describes a dog: type, structure, color. We preserve that by choosing who to breed, and the show ring checks it.

The other thing is the breed's genetic diversity, the range of variation that keeps a breed healthy and adaptable for generations. No ring checks it, and no single dog shows it.

The two are not opposed, but they can drift apart. The dogs that fit the standard best get used most, and heavy use of a few lines narrows the genome behind them. This proposal asks what the breed has kept, what it is losing, and which living dogs hold what is at risk. Sequencing is how we would find out.

What the pedigree already shows

A few dozen old lineages supply most of today's genome, and over a thousand more survive only as a trace. We measured this on a pedigree database of 118,706 Great Danes with birth years back to the 1950s. The living-dog stand-in is the 3,637 dogs born 2016 to 2020 with both parents recorded. We then asked which ancestors born before 1980 their genomes come from.

Inbreeding points the same way. Across the 7,456 dogs born 2012 to 2020, average inbreeding on the full recorded pedigree is 12.6%, and half the dogs are above 12.5%, the level of a half-sibling mating.

These are expectations from recorded pedigrees. About 8% of the genome traces to unrecorded parents, so the true picture is somewhat less certain than the table suggests.

Relatedness here counts every shared ancestor across all recorded generations, not only close family. About half of a typical recent dog's genome (51%) comes from the same 58 old ancestors. The most-connected dog in the data gets 77% of its genome from them, higher than 99.8% of recent dogs. That is why it is as related as a half-sibling to 1,501 of 3,611 recent dogs, though only 6 are its own pups, siblings, aunts or uncles: it is built mostly from the ancestors that nearly every dog shares.

The Kinship Study counts ancestors a different way (any era, dogs born 2012 to 2020) and reports 21 ancestors for half of the gene pool. This page counts ancestors born before 1980 for dogs born 2016 to 2020 and gets 58. Both are real; they answer slightly different questions, and the pattern is the same: a small set of old dogs supplies half of the breed.

Why one dog is not enough

The Great Dane reference genome came from a single female named Zoey (Halo and others, PNAS 2021). The paper does not give her pedigree or say why she was chosen, so we cannot place her among the families on our map.

A reasonable guess is that she was chosen because she is clearly a Great Dane and answered the questions the authors had. That is a good reason for a first reference. A breed is also a population, though, and a reference genome serves it best when the dog is chosen to represent that population and to help the analysis of the dogs within it. The authors were probably not thinking of the Great Dane as a population of show dogs, but that is what most living Great Danes are, and their pedigrees are what let us choose with the population in view.

What we can show is how much the choice of one dog matters. In the 3,611 dogs born 2016 to 2020, the best-connected dog is related at the half-sibling level or closer to 1,501 of them (42%). A dog with the median reach is related to only 181 (5%). A single reference dog speaks for the breed only as far as its own family does, and the pedigree is what tells us how far that is.

This is a reason to record pedigrees for reference and sequenced dogs, and to choose them with the family structure in view. It is not a claim that the choice of Zoey was poor, because we do not know where she falls.

Which dogs to sequence

The dogs that tell us the most about the breed today are not the dogs that carry the most of its past, so we would pick for diversity. Picking the most-connected dogs retains about half as much ancestral diversity per dog as picking for it.

We scored each living-proxy dog by how much of each historical ancestor's genome it is expected to carry, then built the sample greedily: each pick is the dog that adds the most ancestry not yet held. The unit is an ancestor genome equivalent, where one fully unrelated dog counts as one. The 3,036 candidates with at least 90% known ancestry could together hold about 640.

There is no small set that captures most of it. Half of the 640 takes about 525 dogs, because each lineage lives in a different dog. That shapes the plan: the first 50 to 100 dogs are the best value per dog, and every tier after that adds steadily.

The most-connected dogs (the ones that cover the most living relatives) are still the right choice for a different question, namely what the breed is like today. A sample could take both: a small set chosen for coverage of the living population, and the rest chosen for diversity.

Rare lineage sits in a minority of dogs. Among dogs with well-known pedigrees, the top 10% carry about 2.2 times the average amount, and the top 1% about 6 times.

What sequencing would answer that a pedigree cannot

A pedigree predicts how much of each ancestor a dog should carry. Sequencing measures how much it does carry. The gap between the two is the finding.

• Realized versus expected retention. For the lineages the pedigree says are nearly lost, are they really still present, or did chance and heavy use of a few sires remove them sooner than the pedigree implies?
• Haplotype diversity. How many distinct stretches of genome survive across the breed, and how many sit in only a handful of dogs.
• Breeding history. Pedigrees and photographs record what breeders chose and when. Sequence adds what the genome kept: how much diversity each era left behind, and where the trends in the records (popular sires, color, size, type) show up in the DNA.
• Why it narrowed. Chance loss looks different from loss through repeated use of a few sires, and both look different from loss through selection for type, coat color or size.
• Health variants. Which carried-along variants travel with the rare lineages, so that preserving a lineage is a choice made with its health cost in view.
• A way to keep it. Which dogs, matings and stored samples hold the most of what is at risk, so a preservation plan has specific dogs in it.

Pilot genome

One dog from our own program anchors the design: an assembled, public Great Dane genome linked to a deep pedigree. The rest of the sample is chosen by the pedigree scoring above, not by who we happen to own. One dog of ours is included because we can control the sample, the consent and the release, not because the scoring picked it, and the proposal says so.

The dog is Mia. By the pedigree she is at the 90th percentile of recent dogs for how much of her genome comes from the 58 key ancestors (71%, against 51% for the typical dog). She represents the core of the population. On rare-lineage score she sits at the 25th percentile, below the median of the cohort. She is not a rare-lineage dog, and we do not claim she is.

• Mia: long-read sequencing (about 30x HiFi, or ultra-long nanopore) and a scaffolding library, assembled de novo and compared with the Zoey reference.
• Her parents, Rolex and Piper (optional): short reads only. With them the assembly separates into a paternal and a maternal haplotype (trio binning), so one assembly also shows two ancestral contributions.
• Annotation: transfer the reference annotation first, then add RNA from a few tissues if funding allows.
• Release: reads, assembly and annotation public, with the pedigree of the sequenced dog. Living dogs outside our program stay anonymous.

These are pedigree expectations, not measurements. A second pilot dog chosen for rare lineage would add what Mia does not.

Plan, samples, and budget

This section needs real quotes. Every price is blank on purpose.

The pedigree work already gives the dog list, ranked and screened for pedigree completeness, so the lab only needs to receive samples.

For the core lab: sample sheet

One row per dog for the pilot. Status cells are blank on purpose until each is confirmed.

What we would bring to the first conversation:

• What we want: one de novo assembly and, if the parents are included, two haplotypes from trio binning.
• Platform and depth, and whether a Hi-C or similar scaffolding library is worth adding.
• Sample handling the lab requires: collection tube, volume, shipping, and how fresh the blood must be.
• Data release: reads, assembly and annotation public, with the sequenced dog's pedigree.
• A quote per dog and for the whole run, and who is listed as an author.

Limits and risks

• Pedigree expectations, not measurements. Everything above comes from recorded pedigrees. Realized values are what the sequencing would supply.
• Incomplete pedigrees. About 8% of the cohort's genome traces to unrecorded parents. A dog with a shallow pedigree can look falsely unique, so the picks screen for at least 90% known ancestry.
• A proxy for living dogs. Birth years 2016 to 2020 stand in for dogs alive and breeding now. The data holds no alive-or-dead status, and the most recent years are thin because registrations lag.
• Rare is not the same as good. A lineage can be rare because breeders moved away from it for a reason, including health. Sequencing is what separates useful diversity from carried-along problems.
• What survives is all we can read. Sequencing living dogs can only show variation that persisted. Variation already lost needs old samples, such as stored semen or tissue, to recover.

Open decisions

• Who is this proposal for: a funder, a lab, a breed club, or all three?
• How many dogs is realistic, and on what budget?
• Can the pedigree database be cited by name, or should the proposal describe it only as a large pedigree database?
• Do we include dogs born before 2012 where samples or stored semen exist?
• Are our own dogs part of the sample, and how are they shown?
• Do we treat health variants as in scope, or leave them for a second phase?

Companion: The Living Great Dane Family Map, an interactive view of the same pedigree.