cMs = centiMorgans; HIR = half-identical regions; IBD = identical by descent (HIR + FIR); 1C1R = 1st cousin, once removed. Relationships not listed in the table above do not have a 0% probability. They have a < 0.1% or ~0% probability under the following assumptions: no endogamy, double relationships, or pedigree collapse.
Your submissions to this new DNA match survey will greatly help improve and build more tools like this. A double cousin predictor is also available. You can find mobile apps. for relationship predictions in the Apple Store and on Google Play.
The "cMs" input box can be used for Ancestry, FTDNA, and MyHeritage. The methods used at 23andMe are different in that total IBD sharing is used rather than HIR. GEDmatch predictions can be obtained by using the 23andMe cMs box and checking for "HIR." Currently, the number of segments only affects predictions when 3rd cousins and closer are possible.
Population weights are applied to most relationships in this tool. They increase the probabilities for more distant cousins because a person likely has about 5x as many cousins from one generation of ancestors compared to those from the next generation of more recent ancestors.
Please subtract any X-DNA from cM values before using cM input boxes and leave in any X-DNA for percentage boxes. 23andMe includes X-DNA in reported percentages. As of April of 2022, this tool includes probabilities generated from data that include X-DNA, a first for genetic genealogy tools. For probabilities at any other site, it's important to subtract X-DNA off of the total. The X-DNA amount can be significant: several relationship types share up to 182 cMs of X-DNA and full-sisters share up to 364 cMs. Recent discoveries have shown that including X-DNA in the total is better than ignoring the X-DNA amount in relationship predictions.
*Parent/child and full-sibling relationships are easy to distinguish from each other or any other relationships. Parent/child relationships consist of a half-identical match across the whole length of the genome. Full-siblings share 12.5% fully-identical regions (FIR), on average, while parent/child and half-sibling relationships typically share 0 FIRs. Genotyping sites will take this into account in their relationship predictions. You can generally trust the label "Sibling" (meaning full-sibling) at the original testing site.
Probabilities are included for relationships as far back as 8C1R. A great advantage of this tool, other than the accuracy of the data, is that it treats close relatives as not being in the same group because the probability curves are significantly different. For distant relatives, there's much less certainty about the genealogical relationship for your DNA matches. Matches as low as 8 cM can be checked here, however the relationship may be farther back than 8C1R. While the relative probabilities are accurate for the relationship types shown, one also has to consider that the true relationship is a type not listed, such as 9th to 15th cousins. Any of the probabilities shown above are only relative to the other relationships listed, therefore they’re only meaningful in comparison to the other relationships. Not only are very low cM values difficult to assign to a recent ancestor, but segments of 20 cM or 30 cM may be on pile-up regions and therefore come from very distant ancestors.
The data used for these predictions came from Caballero et al. (2019). In this case, the refined genetic map of Bhérer et al. (2017) was used as well as the crossover interference parameters of Campbell et al. (2015).
The functionality of cM-only predictions here is similar to that of a tool at DNA Painter, which uses data that came from simulations conducted by AncestryDNA. The simulations resulted in probabilities that were shown in a graph in their 2016 matching white paper. These probabilities were then copied using an online plot digitizer. The probabilities here using only cMs have been shown with empirical data to give a higher probability to the correct relationship, on average, than the Ancestry simulated probabilities. Predictions using the # of segments perform far better than when using only cMs. But you will see that the real advantage, by far, comes from seeing the individual probabilities for relationship types and ancestral gender paths (i.e. maternal half-sibling rather than simply showing the probability for the group that includes grandparent/grandchild, half-sibling, and avuncular relationships).
The average probability given to the correct relationship for close family relationships is astoundingly about 50%. The expected value for a predictor for which splitting up the groups added no value would be 17% when only the close group is possible, 6% when only the 1st cousin group is possible, and lower for in-between values when both groups are possible. By using SegcM you're treating yourself to relationship predictions that are an improvement by orders of magnitude. The Orogen relationship predictor has been providing amazing predictions since February of 2022. You may have seen links to DNA-Sci predictors in Jonny Perl's newsletters. So hopefully you've been taking advantage of that to get the most accurate predictions and speed up your genetic genealogy work.
If you're one of the many people who have benefited from DNA-Science predictions for the past two years, don't keep it a secret. There are many other people who haven't been so fortunate. Help answer people's questions by showing them the most accurate probabilities. They deserve to see the correct relationship at the top of the list.
DNA-Sci is also the original home of DNA coverage calculations, since June of 2019.