23andMe - How Relative Finder Works (sorta)

By Belinda Dettmann

The mathematics of the Relative Finder facility is based on the proportion of total genome that relatives share.    For instance, parent and child share 50% of their genome on average, siblings share 50 % on average, niece/nephew and aunt/uncle share 25%, 1st cousins share 12.5%, 2nd cousins share 3.125%, 3rd cousins share 0.78%, 4th cousins share 0.195%, 5th cousins share 0.05%, on average.    However individuals don't always get the exact average, so there's a lot of variability built into the system.

The system also takes into account the number of shared discrete chromosomal segments and the minimum size of a contiguous segment that two relatives share, and for the Relative Finder system thie size is set to a minimum of 5cM, where a centiMorgan (cM) is a measure of likelihood of a chromosomal block splitting at a given point.

The system works on a time scale of generations, not years.    If you wish to calibrate it in years, the approximate measure for the last 1000 years of European ancestry works out to about 30 years per generation, on average, but this varies widely between individuals and family lines so I think it's a mistake to think in terms of average generation times in this context.    You should be looking for a specific ancestor, not for some theoretical average person.    Few individuals are average.

Relative Finder does not use any information from Y DNA or mtDNA.     I believe it is calibrated on data from autosomal chromosomes only.    It has nothing to do with mutations, but relies on the pattern of recombination of maternal amd paternal non-sex chromosomes.

Belinda