KIT (white spotting)

Crossing two of them

Viability: what canOccur means here

The alleles UC Davis lists as “homozygosity thought nonviable” — W22, W13, W23, W10, W5 — get canOccur false. Those are embryonic lethals, so there is no such horse — no catalogue entry, nothing in totalGenotypes(), nothing in the founder table. It is the same statement the sex locus makes about Y/Y, and it is a different statement from lethal white, which is a foal that is born.

Compound heterozygotes are allowed. W5/W13, W22/W10 and the rest are real horses. The evidence is about an allele paired with itself, and the breeding advice is to test a prospective mate for the same exact variant — so the rule in the model is one allele twice, not two strong alleles.

“No homozygote has been found” and “a homozygote has been found” are different claims, and the table says which is which. For most of the strong alleles the first is true and the model treats it as a lethal, which is what a breeder is told to do. For W15 the second is true — a W15/W15 horse is on record and it is all white — so its homozygote is allowed and lands on near-white. W4’s has no evidence either way, and picking “allowed” is the choice that does not invent a lethal the source never claims.

White finds white

The painter reads the coat handed to it and raises its own strength by 0.35 × however much of the horse is already de-pigmented. That is what makes W20 a booster rather than a decoration, and it is why a KIT pattern over a tobiano or a frame reads louder than the same pattern on a solid horse. One line in the painter, no interaction table anywhere. Full reasoning on the pipeline page.

Its gene carrot

Gene key

horsegenetics.kit

Priority

76

Alleles

W22 W4 W13 W23 W10 W5 W15 SB1 W20 W35 W32 W34 N

Combinations

every pair of its alleles, minus KitGene.nonviableHomozygotes(); one outcome each — WhitePatternGenesTest asserts the relationship rather than the counts

Outcomes

wild, minimal-white, modest-white, sabino, broad-white, extensive-white, near-white, dominant-white (masks), camarillo-white (masks)

Default allele

N

Wild frequency

the boosters first — W35 9%, W20 8.5%, W32 3.5%, W34 1% — then SB1 2.2%, and every strong W at or below 0.5%. See KitGene.frequencies(); Hardy-Weinberg over the lot, with the nonviable homozygotes dropped and the rest rescaled

Founder draws

1 × nextFloat()

Deterministic

no, except the wild type and dominant white

Epi draws

nextLong() seed, 4 × nextFloat() leg heights, nextFloat() belly, nextFloat() face

Painter

WhitePattern.sabino, at a strength per outcome

The alleles

Twelve of the thirty-plus named ones (W1–W35, with gaps — there is no W29), chosen to span the real range rather than to reproduce a testing panel. The W number is a discovery order, not a severity ranking — W20 is far milder than W5, and nothing about an allele can be inferred from its number.

Two all-white alleles, and why they are two outcomes

W22 and W4 paint the identical horse: pure white, pink skin, every other coat gene hidden and still inherited. “No pigment cells reached the coat” has one appearance, so they could have shared an outcome. They do not, and the reason is that they differ in the only thing a breeder can act on:

• W22’s homozygote is thought nonviable. Two W22 parents throw a quarter fewer live foals, and no W22/W22 horse exists.
• W4’s is not known to be. Claims about which dominant-white variants are embryonic lethal are allele-specific, and the ones made online frequently are not. Where the reference is silent this mod allows the homozygote, exactly as it does for W23.

W4 is also a named line rather than a generic white horse. It traces to one stallion, Sultan, born about 1912 and bought at the California State Fair in 1921 by Adolfo Camarillo, who bred him to Morgan mares; the descendants were parade horses for sixty years and the herd was dispersed at auction in 1987. That is why the Morgan is the one breed here carrying it, and why the gene dictionary says “Camarillo white” instead of throwing the fact away.

The boosters are most of what this locus does

W35, W32, W34 and W20 are one group and behave as one. All four are mild; all four have viable homozygotes; all four range from “no visible white beyond ordinary markings” to frank sabino-like spotting; and all four add white beside any other variant here.

W20 is the famous one — detected in twenty-five of twenty-eight surveyed breeds, and reaching an allele frequency of 0.34 in sampled Thoroughbreds, 0.33 in Gypsy Cobs and 0.21 in Paints. But in the one large commercial dataset that measured the newer alleles, W35 was commoner still at 0.168, with W32 at 0.061 behind it. That is the correction this page used to need: the mod carried W20 alone and treated the rest of the locus as a museum of rare founder alleles.

The founder frequencies here are that ordering at roughly half the magnitude, because a commercial colour-testing dataset is a population selected for having something to find. The breeds that genuinely carry them — the stock horses for W35 and W34, the Thoroughbred and the Paint for W20, the Quarter Horse, Paint, Standardbred and Welsh Pony for W32 — say so in their own tables.

The combination table

Every pair of alleles, nine outcomes, written as a cascade: the strongest allele first, then what the second copy adds. There is no dose arithmetic and deliberately so — this locus holds a real viable dose series (SB1), a booster that is subtle alone (W20), and several alleles that are “dominant with variable expression”, where one copy is already effective and the phenotype is not neatly intermediate. Only a table says all three at once.

How it paints

Every painted outcome is the same WhitePattern.sabino shape at a different strength: white grows inward from the extremities — legs, belly, face — with ragged, roaned margins, and past about strength 0.42 the body follows in torn patches, colour retreating last to the ears and crest. That is what “sabino-like” means throughout the W series, and writing a bespoke painter per allele would be writing the same painter eight times and pretending the differences were principled.

private static final double S_MINIMAL    = 0.12;
private static final double S_MODEST    = 0.24;
private static final double S_SABINO    = 0.42;
private static final double S_BROAD     = 0.58;
private static final double S_EXTENSIVE = 0.74;
private static final double S_NEAR_WHITE = 0.93;

Measured coverage through the real pipeline, averaged over three epigenomes: W20/N 5.7%, W20/W20 11%, SB1/N 15%, SB1/W20 25%, W13/N 73%, SB1/SB1 94%, W22/N 100%. WhitePatternGenesTest pins that the ladder is monotone at every step, because a ladder whose rungs a player cannot tell apart is a ladder whose descriptions are decoration.

Dominant white is a different painter, not the top of the ramp: it zeroes both pigments everywhere and masks. All-white is not a great deal of sabino, it is the absence of pigment cells altogether — and the model has one true epistatic result rather than an asymptote.

The face

The face marking comes from the shared face-marking vocabulary every white locus shares, at the outcome’s own strength and with a jag of 0.42 — so a KIT marking has the same torn margin on the face that it has on the body. Strength picks the distribution rather than the marking, which is what finally makes the weak end of this locus mean what its description has always said: W20/N is “a star and a sock”, and about a third of the time it is now literally a star — a detached patch on the forehead, with about another third of horses carrying nothing on the face at all. Up the ladder it becomes a star and stripe, then a blaze, and SB1/SB1 is a bald face three times in five.

What is at this locus, and what only looks like it is

This is the distinction the whole restructure turns on. Only alleles at exactly the same locus share a gene. SB1 and every numbered W are defined KIT variants, so they compete for the same two chromosome slots. Two patterns that sit right beside them do not:

So a tobiano can also carry W20, and a dominant white can be a roan underneath — both of which real horses are, and neither of which the old single “white” gene could say.