MITF (splash white)

Crossing two of them

Deafness

The splash pattern comes from melanocytes failing to reach the skin, and the same cells line the inner ear — so a horse white enough at this locus is very often deaf. A homozygote at this locus (no N copy) gets the splash-deafness condition.

It costs the horse nothing. The mod has no hearing for a horse to lose, and inventing a penalty to stand in for one would be worse than saying so: it is INFORMATIONAL — named in the info panel and the paper dump, and that is all. See the horse’s body.

The condition is shared with PAX3: one disorder with two causes. A horse homozygous at both loci is told it is deaf once, not twice — TraitBuilder de-duplicates on the condition id for exactly this case.

Its gene carrot

Gene key

horsegenetics.mitf

Priority

78

Alleles

SW3 SW1 SW5 N

Combinations

10, of which 9 a horse can carry, landing on 5 outcomes

Outcomes

wild, splash-minimal, splash, splash-bold, splash-extensive

Default allele

N

Wild frequency

founders carrying one copy: MitfGene.WILD_SW1_PERCENT, WILD_SW5_PERCENT, WILD_SW3_PERCENT. Heterozygotes only — a doubled splash is bred, not caught

Founder draws

1 × nextFloat()

Deterministic

no

Epi draws

nextLong() waterline seed, nextFloat() waterline height, then the shared face-marking draw (1 long + 8 floats)

Painter

WhitePattern.splash, at a strength per outcome — and over a range for SW1/N

The alleles

The combination table

SW1 is the one allele here with a documented viable dose effect, so its homozygote is a step up rather than more of the same; SW3 is the strong one, so any second variant beside it tips the horse into the extensive outcome.

SW1 is the common one, and it is usually invisible

SW1 is a 10-base-pair insertion in MITF’s melanocyte-specific promoter — a regulatory change, an alteration to the switch rather than to the protein. Three things follow, and together they make it unlike every other allele on this page.

1. It is old. Several hundred years, predating most modern breed formations, which is why it turns up across stock horses, ponies, warmbloods and gaited breeds alike rather than tracing to one stallion the way SW5–SW8 do.
2. It is widespread. In a 2022 survey it was the splash allele found in by far the most horses and the most breeds; the others are rare and family-limited.
3. It is the most variably expressed thing in the white loci. One copy can be a snip and one white foot, or ordinary-looking face white beside a single blue eye, or a full blue-eyed splash — on the same genotype, in the same family.

The paper that identified it says the third point outright: in a minimally expressed horse, splashed white cannot be told from common white markings by eye. So SW1 is the allele that hides in plain sight, and the one worth testing for when a horse’s markings look unremarkable.

Which is why SW1/N is painted over a range

Every other outcome in the white loci passes WhitePattern.splash a single strength. SW1/N passes a pair, and the horse’s own epigenetic roll picks a point between them — the same roll that then decides how high the waterline sits, so a horse that drew a strong splash draws it high too. It costs no extra draws; only what they are used for changes.

Modelling the widespread splash allele at a fixed strength would make it the one thing it is documented not to be: predictable. The outcome is named splash-minimal after the usual case, and its description says so.

How it paints

The horse dipped in white from below. One level waterline in absolute body-space Y rises with strength, taking the legs and belly whole, with a blaze widening to a bald face above it. The line is hard — wobbled at two octaves so it reads torn rather than ruled, but never faded:

• a sharply bounded margin is the diagnostic difference from KIT’s ragged sabino edge, and the thing a horse is actually looked at to tell them apart;
• the red/black gradient has no grey between a coloured texel and a spent one, so a half-scaled black texel samples the warm diagonal and reads gold. A one-pixel fade paints a tan fringe along the entire waterline. (It did, in the first bake.)

private static final double S_MINIMAL_LOW  = 0.10;   // SW1/N is a RANGE...
private static final double S_MINIMAL_HIGH = 0.44;   // ...not a value

private static final double S_SPLASH       = 0.38;
private static final double S_BOLD         = 0.62;
private static final double S_EXTENSIVE    = 0.86;

Measured white coverage over the standard seed set (the same averageWhite helper WhitePatternGenesTest uses): SW1/N 12%, SW5/N 17%, SW1/SW1 38%, SW3/N 39%, SW3/SW1 69%. One copy each of SW1 and SW2 comes out at 38% — well past either alone, which is the claim the split exists to make and which that test pins as a threshold rather than a number.

The face

Splash’s face marking comes from the shared face-marking vocabulary, with a jag of 0.11 — the same crispness that makes the waterline diagnostic, carried onto the face. But splash runs its face much hotter than its body: the marking strength is the outcome’s strength plus SPLASH_FACE_BOOST (0.34), because a splashed white horse is defined as much by a bold blaze as by the white climbing its legs. So splash reads as a blaze, and splash-bold / a two-locus horse as a bald face. (Without the boost, single-copy splash landed on stars and snips and horses came out bare-faced — the regression this fixes.) splash-minimal is the exception and deliberately so: it runs the ladder from a star at the bottom of its range to a blaze at the top, which is what “you cannot tell it from ordinary markings” has to look like. PAX3 draws from the same vocabulary with the same boost, so a horse splash at both loci gets one bold marking per locus rather than two competing conventions.

Why splash is two genes

A gene has two slots. If SW1 and SW2 were alleles of one gene, they would compete for those slots — and a horse could not carry both. In life a horse carries N/SW1 at MITF and N/SW2 at PAX3, and comes out with markedly more face, leg and belly white than either alone. One gene cannot express that genotype at all.

Two MITF variants can still meet — SW1/SW3 is a real compound heterozygote — and that is what this locus’s own table is for.