There’s been some discussions recently as to why iwdb shows different (lower) values of inbreeding than some other systems. We would like to take the opportunity to explain how we calculate Coefficients of Inbreeding (COI), how that differs from Wright’s formula for doing the same thing, and why we’ve chosen the method we have.
When describing any inbreeding, it’s necessary to calculate the distance between all ancestors being present on both sides of the pedigree. This is done through a normal path-counting method, counting the distance in generations from the appearance on the sire-side and the appearance on the dam-side.
This is what IWDB does. When Sewell Wright constructed his formula in 1922, he also added that the COI of the common ancestor itself should be added to the calculation. This is the way it’s done in most pedigree-programs, but not in IWDB. Thus IWDB will report a somewhat smaller
There are technical reasons for our choice to omit additions of common ancestor COI. IWDB calculates COI on anything from 3 through 10 generations. In order to include ancestors COI, we would have to store the COI of all dogs that were bred from in the database in 3-10 generations. In order to provide accurate and fast calculations for trial matings, we would have to include all dogs in the database in this dataset as well as COI of all trial matings. The COI would have to be recalculated for all descendants any time we add an ancestor to the dataset. All of this is very time consuming. Typically, in the neigbourhood of 30 minutes each time it’s done.
What difference does it make? It varies a lot and all depends on the ancestors, their inbreeding and their usage. IWDB will normally report a bit lower COI (Anywhere between 1-3 percentage points) than other programs.
We are comparing apples to apples, though, so when comparing to breed averages, the calculations done are the same.