Rabbit genetics software that learns from your herd records

The engine never assumes a genotype from breed alone. It builds each locus from evidence:

• Recorded variety. A REW is c/c; a self is a/a. Those are certainties.
• Parents. Each contributes one allele, which limits what a rabbit can carry.
• Ancestors. A recessive in a grandparent counts as a 50% chance of carrying it, in a great-grandparent 25%. The engine stops at three generations.
• Offspring. A kit that shows a recessive proves both parents carry it, so genotypes sharpen as you record litters.
• Your overrides. If you have test-bred an animal, set its genotype and the engine stops second-guessing it.

One locus, made-up rabbits, the whole loop:

1. 1

   A buck is recorded as Castor

   Castor, a chestnut agouti Rex, needs at least one A. His A locus shows A with the second allele unknown: he could be A/A, A/at, or A/a.

2. 2

   A litter from him includes a self black kit

   Self is a/a, so the kit received an a from each parent.

3. 3

   His genotype resolves

   After the litter syncs, the buck's A locus reads A/a, marked exact, with his offspring as the evidence. No deliberate test breeding was needed.

4. 4

   The next pairing is predicted

   Paired with a self black doe (a/a), the A locus predicts 50% agouti and 50% self. If the pair match on their other loci, the variety list reads chestnut agouti 50%, self black 50%.

If the kit had been agouti instead, the buck would stay unresolved: an agouti kit is consistent with every possibility, so nothing is guessed. The genetics engine guide walks through the same example.

For a proposed doe and buck, OmniBurrow shows the odds at each locus and a ranked list of up to 12 likely varieties from the breed's catalog, leaving out anything under 2%. When a parent's genotype is incomplete, recessive outcomes show honest odds below 100% instead of a certainty.

Coefficient of inbreeding uses Wright's path method across up to 10 generations and reports how deep the recorded pedigree actually goes, so a low number on a thin pedigree is easy to spot. See predicting offspring and lines and COI.

The web app and the mobile app show the same results. Genotypes are worked out on the server; on the installed mobile app, pairing predictions and COI run on the phone, so they work with no signal using the genotypes already synced to it.

A prediction gives the odds for a pairing based on what your records prove. It cannot promise what any single litter will contain. It also cannot know:

• Anything you have not recorded — no variety, parents, or litters means no evidence.
• Traits outside the 13 modeled loci (A, B, C, D, E, En, Du, Si, W, P, V, Dw, and Mf): body type, fur density, ear carriage, and temperament are not modeled.
• Whether a recorded variety is correct. A wrong entry carries into every inference built on it.
• Relatedness beyond your recorded pedigree. A 0% COI on two recorded generations means no shared ancestor that OmniBurrow knows about.

It supports every ARBA-recognized breed from its reference catalog, and keeps multi-generation pedigrees alongside the genetics.