How Predictions Are Calculated

The calculator takes the genotype you enter for each parent, one locus at a time, and works out the probability of each genotype in the offspring. For a single locus this is the Punnett square you may remember from school: an Ee × Ee cross gives one quarter EE, one half Ee, one quarter ee.

It does this for every locus at once, and combines them. Because the coat-colour loci are inherited independently, the chance of a particular combination is the product of the per-locus chances. The calculator then translates each resulting genotype into the colour it produces and groups the outcomes, so the percentages you see are the share of puppies expected to show each colour, on average, over a large number of litters.

“On average, over a large number of litters” is the important qualifier. A percentage is not a promise about one litter of six. A cross that reads 25% brown will sometimes throw no brown puppies and sometimes throw four; the number is the long-run expectation, not a quota.

The maths rests on a few standard assumptions, and it is worth naming them:

• The genotypes you enter are correct. The calculator is only as good as its inputs. A guessed genotype produces a confidently wrong answer, which is worse than no answer. Where a locus matters, confirm it by DNA test.
• Loci assort independently.The colour loci sit on different chromosomes or far enough apart to be treated as independent, so one locus’s outcome does not drag another’s along.
• Each parent passes one allele per locus at random. Standard Mendelian inheritance, one copy from each parent, no bias toward either.
• Untested loci are unknown, not absent. If you have not entered a locus, the calculator does not assume the dog is clear there. It tells you it cannot speak to what it has not been given, rather than quietly reporting a zero.

This is the heart of it. Some traits are settled by genotype and the calculator can speak to them with confidence. Others are polygenic or have no valid test, and for those the calculator deliberately steps back. Mixing the two up is the single most common way colour prediction goes wrong.

Three things in particular the calculator will not tell you, by design:

• The exact shade of a red-based dog. Whether an ee puppy is deep red, apricot or cream is polygenic and not captured by any single-locus test. The calculator reports the dog as red-spectrum and leaves the shade to your eye. It is recorded, never guessed. See why red poodles fade.
• How far a greying dog will clear, or whether it will. Progressive greying has no identified gene and no valid test. The calculator carries the G locus as something you observe from the adult coat and record — never as a genotype it predicts. See the greying gene.
• How much white a spotted dog will carry. Whether a parti is 50% or 80% white, and the untestable Irish-spotting and Whitehead modifiers, are polygenic or unidentified. The calculator predicts whether a dog is parti, not how it is distributed.

A puppy’s genotype is fixed at conception; how it reads on the coat is not always a clean translation. The calculator reports the genotype and its most likely expression, but several real effects sit between the two:

• Masking. One locus can hide another. A KB dog is solid whatever its A locus says, and an ee dog shows none of its eumelanin pattern — including merle. The allele is still there and still inherited; it just does not appear on that dog.
• Age. A greying dog is born one colour and clears into another over one to three years. The birth coat and the adult coat are different photographs of the same genotype.
• Modifiers. Polygenic modifiers move the result within a range — the depth of a red, the spread of white on a parti — without changing the genotype the calculator reported.

The calculator is under active review against the framework in Poodle Color Genetics Volume I. The genetics it uses have been corrected substantially and continue to be checked; where a locus is still being validated, it is noted in the tool itself. If you find a result that does not match what the science says, that is worth reporting — email research@poodlegenetics.com.

← Back to the calculator