SMOC2
short-face candidate
SMOC2 chr1:56.4Mb (Marchant 2017). Top brachycephaly hit in our village-baseline selection scan at |dP|=0.85. The H1 hardening test (logistic regression of brachy phenotype on the rep SNP) showed R^2=0.05 alone, vs R^2=0.96 with a genome-wide PRS - the rep SNP is necessary-but-not-sufficient.
short-face allele common at this rep SNP - BUT brachy is polygenic and SMOC2 alone does not cleanly tag the phenotype (some non-brachy breeds also read high here, e.g. Afghan Hound, Cardigan Welsh Corgi)
long-face allele common - this direction reliably distinguishes non-brachy from brachy/intermediate (Bull Terrier, Leonberger, Newfoundland, Saint Bernard read at the bottom)
Where the brachycephaly variant sits across every breed with data, by allele frequencyAllele frequencyWhat it isHow common a gene variant is across dogs, from 0% (none carry it) to 100% (all do).For your dogA high number means it's common in the breed. It does not tell you whether your own dog carries it.PreciselyThe proportion of sampled chromosomes in a population that carry the variant allele.Sniff Atlas (CanVAS) · measured · Ask about this → and grouped by AKC breed groupAKC breed groupWhat it isThe American Kennel Club's grouping of breeds by the job they were bred for: Sporting, Hound, Working, Herding, Terrier, Toy, Non-Sporting.PreciselyAKC parent-group classification. A registry convention reflecting historical use, not a genetic grouping.American Kennel Club · registry convention. Each tick is one breed; a variant fixed in one group and absent in another shows up as a gap. These come from one marker per trait, so read them as a tag, not a verdictA trait tag, not a verdictWhat it isThese frequencies come from one marker per trait. It predicts the trait for most breeds, but there are real exceptions.For your dogRead it as a tendency for the breed, never as a guarantee about your dog. One marker is not the whole story.PreciselySingle-SNP trait proxies are population tags, not deterministic genotype-to-phenotype calls. Penetrance, modifier loci, and breed-specific haplotypes cause exceptions (for example, the MSRB3 ear-type tag misreads Toy Poodles and Belgian Sheepdogs).Sniff Atlas methodology · known limitation, stated · Ask about this →: accurate for most breeds, with real exceptions.
- 100% n=102
- 100% n=96
- 100% n=38
- 100% n=33
- 100% n=31
- 100% n=26
- 100% n=24
- 100% n=21
- 99% n=67
- 99% n=59
- 2% n=25
- 2% n=44
- 12% n=182
- 13% n=37
- 17% n=23
- 20% n=40
- 22% n=192
- 26% n=31
- 28% n=23
- 29% n=381
105 breeds with fewer than 20 genotyped dogs are not ranked here. At that sample size a single dog swings the frequency, so the figure is not yet stable enough to compare.
Frequency is measured at the typed-backbone 1:56398416 on chr1, 56.4 Mb (inside the gene body). Alleles A/G. Coordinates from ensembl symbol UU Cfam GSD 1.0. Per-breed frequencies are computed across all CanVAS dogs labelled with that breed (missing genotypes excluded).
SMOC2 as it is catalogued across the genomics world. Each link is the canonical record, so this gene composes with everything those resources know.
In humans, this gene's counterpart is SMOC2. That ortholog is what connects SMOC2 to a century of human medical genetics. The dog and human proteins are 90% identical.
In people, SMOC2 appears tolerant of loss-of-function variation (gnomAD v4.1 constraint, LOEUF 0.82). Constraint measures intolerance to loss-of-function only and does not indicate importance; some tolerant genes cause disease through other mechanisms.
In people, variants in the SMOC2 gene have conflicting classifications in ClinVar, and none is expert-reviewed. The evidence is unsettled, not that variants here are benign.
The per-breed allele frequencies on this page are derived from the open Sniff Atlas v1.0.1 (Gehring 2026, doi:10.5281/zenodo.20566358, CC-BY 4.0). The underlying genotype substrate is CanVAS (Brundage 2026, doi:10.64898/2026.04.13.718238), and disease associations are grounded in OMIA. Full citation formats including BibTeX, RIS, and CITATION.cff at sniff.world/cite.