Shade — how far black spreads over a bay

What it does

Crossing two of them

Its gene carrot

See it doing something — the agouti preview offers this locus as a modifier, because that is the only place it has anything to show.

Gene key

horsegenetics.shade

Alleles

ShD Sh ShL

Outcomes

five, and all of them wild type — it paints nothing alone

Read by

agouti, through Gene.coatDependsOn()

Default allele

Sh

Priority

95 — the “paints nothing” band

Founder draws

1 × nextFloat()

Model

common.genetics.BayShade, common.genetics.genes.ShadeGene

The three haplotypes

It is additive, not dominant: ShL/ShD and Sh/Sh land on the same place because their dosages cancel, which is what an incomplete-additive haplotype means and what the two-outcome “dominant/recessive” vocabulary could not have said. Each combination gets its own sentence in the gene dictionary even though none of them paints, because “carried, and pushing dark” and “carried, and pushing light” are different things to read.

The shade score

BayShade sums four things. Three of them are genotype and fix the label; the fourth is a per-horse roll and moves the look inside it.

The four bay labels are bands of the genetic part, so a breeding prediction never depends on a roll. The roll is deliberately narrower than a band: two horses of the same genotype differ visibly, but neither of them contradicts the name the gene dictionary gave it by more than an edge.

Over the wild population that comes out at roughly 15% blood bay, 61% bay, 19% liver bay and 5% seal brown — the seal being scarce because it needs ShD/ShD and a dosage term on top of it. The exact numbers are arithmetic on ShadeGene.founderTable(); the shape is the point.

Silent, and carried by everyone

Every combination of this gene is a Expression.wildType(), so affectsCoat() is false: it never enters the coat pipeline, it collapses to one pen in the gallery whatever its dosage, and it is invisible in the genome display as an expressed thing. Agouti names it in coatDependsOn(), which is what folds its alleles into a bay’s texture key — and, because that fold is conditional on agouti actually painting, leaves them out of a chestnut’s, so two chestnuts of different shade genotype still share one baked texture.

That is the same mechanism PATN1 and PATN2 use, with one deliberate difference: the appaloosa modifiers only reach a founder that already rolled LP, and shade reaches every founder. The regulatory region is present in every horse. A chestnut carries and transmits it with nothing to show for it, which is exactly how a chestnut line hands a seal brown to a grandfoal.

Interactions

• It does nothing without a bay. No black to move on a chestnut; nothing restricted to move on a black. Both still carry it.
• The dilutions run after the shade is resolved, so a buckskin has a bay shade underneath it — a cream over a liver bay and a cream over a blood bay are different buckskins. The same goes for dun, silver and champagne.
• Grey hides it progressively, the way it hides everything; the score stays in the genome, so a grey horse is still on record as having been born a seal brown.
• Breeds all name it. Shade is a modifier a coat gene reads, so BreedFounder would force it wild on any breed that did not — leaving every breed horse shade-neutral while only feral mixed ones varied. Breed.Builder.shadeAny() is called by all three agouti helpers so that cannot happen quietly.

What it is modelling

A genome-wide association study of 126 genetically confirmed bay horses found a ~0.5 Mb region on equine chromosome 22, immediately upstream of ASIP and taking in the neighbouring RALY gene, very strongly associated with the continuum from light bay (black confined to mane, tail, lower legs and ear margins) through to near-black bay (black over most of the body with only small red areas left). It is by a wide margin the strongest known signal for bay shade — much stronger than the ordinary A/a test.

The causal mutation has not been identified. The lead marker sits in a high-linkage block and nobody has yet disentangled whether the functional variant is in ASIP, in RALY, or in a regulatory element controlling one or both. The leading biological guess is a regulatory variant that alters when, where or how strongly ASIP is expressed: broader ASIP activity inhibits MC1R over more of the horse and biases it redder; narrower activity lets eumelanin cover more of it.

So there is no allele to name, and this gene does not pretend there is. What it models is the thing the association actually describes: a haplotype dosage, as three alleles worth -1.5, 0 and +1.5, summed.

Why there is no At

A tan-point at allele is genuinely mapped in several other mammals, and the horse version was borrowed from them by analogy. But the commercial equine agouti test distinguishes A from the recessive 11 bp deletion a and nothing else, and seal-brown-looking horses test as ordinary A/A or A/a — genetically bay at that locus. The chromosome-22 region above is a far stronger modern lead.

So a mod that advertises genetic accuracy (see philosophy) does not get to ship At as a real allele. There were three honest options — give each shade a made-up allele (arcade), ship At and label it a simplification (fine for a teaching game), or model a shade locus plus the dosage terms. This is the third.