Dog10K.

To know what is unusual in one dog, you first need to know what is normal across many.

Dog10K is a large international consortium formed to sequence thousands of canine genomes, dogs across hundreds of breeds along with village dogs and wild canids, and to assemble from them a comprehensive reference catalogue of canine genetic variation. It is the closest thing the dog has to a map of its own diversity: which positions in the genome vary, how, and how often.

A reference panel like this does something practical as well as descriptive. Most dogs are genotyped on arrays that read only a fraction of the genome directly. A deep, diverse reference lets the rest be imputed, inferred with confidence from the patterns the reference has already seen. That is how a modest genotype becomes a genome-wide one.

Dog10K is the reference the population layer stands on.

The genotypes behind the Sniff Atlas, the CanVAS release, are harmonized and imputed against a Dog10K-scale reference. Dog10K is what lets a sparse array genotype be filled out to a genome-wide picture, and what defines the baseline of normal canine variation that the atlas is read against.

Imputation is powerful and it has limits, and we respect them. Its accuracy depends on how well the reference represents the dog in front of it, and that honesty about confidence is carried through into what the atlas is willing to claim.