This post has been updated.
The What Are the Odds? tool (WATO) at DNAPainter.com has revolutionized how we identify unknown parents, grandparents, and even great grandparents of someone who has taken a genealogy DNA test. It lets you analyze multiple DNA matches at the same time to see where the tester fits best into the family tree of the matches. (You can watch a video introduction here.)
Best of all, you can use autosomal DNA matches from any of the DNA matching services. Even so, there are some quirks you should be aware of, because each company has its own unique matching algorithms.
The Quirks
AncestryDNA has the largest database by far, with more than 20 million people tested. Chances are, most of the matches you use in a WATO analysis will come from there. They use an algorithm called Timber that down-weights some segments that they think are less informative. As a result, the same match at Ancestry might share a bit less than at the other companies. AncestryDNA also reports the “unweighted” amount of shared DNA, before Timber is applied. For WATO, you can use the “weighted” centimorgan amount that’s reported on the main match page rather than the “unweighted” amount. After all, they invented Timber to correct for unreliable segments, and the underlying data for WATO came from AncestryDNA in the first place.
23andMe has the next largest database, with more than 12 million kits sold. You will often find useful matches there who haven’t tested elsewhere. They have two quirks that might impact a WATO analysis. First, they include segments on the X chromosome in their DNA totals. If you are able to see the chromosome browser for a match, subtract those X segments out before you use the shared DNA amount in WATO. If you can’t see the chromosome browser and can only see the percent DNA shared, multiply that percent by 74.5 and use that number in WATO.
No adjustments are needed for matches at either MyHeritage or Living DNA. Just use those shared DNA amounts as reported.
(UPDATE: As of August 2021, adjustments at FTDNA are no longer necessary. This paragraph has been left intact for posterity, but you can ignore it and use FTDNA totals as is.) FamilyTreeDNA includes tiny segments in their total that are usually false positives. It’s essential that you subtract out all of the segments less than 7 cM before you use an FTDNA match in WATO. If you don’t, the tool could give you misleading results.
Finally, GEDmatch has two different matching algorithms: a quick one for the one-to-many match list and a more precise one for one-to-one comparisons with a given match. Use the one-to-one values in WATO.
The Fix
A quick and easy way to correct your match totals is with the Individual Match Filter tool, also at DNAPainter.com. Simply copy the match data into the tool and it will do the rest. It even has instructions for how to obtain the data. Remember, you only really need to correct the totals for matches from 23andMe (in cases of X segments) and, especially, at FamilyTreeDNA.
Summary
The table below summarizes the main matching sites, their quirks, and any adjustments needed to use WATO.
Updates to This Post
29 March 2022 — Noted that adjustments are no longer necessary for FTDNA matches.
Good morning! I have a question. I put in the Target and the question….added some information, but how do I add to what I’ve written?
Here’s the scenario: Person, the target was adopted, I was adopted but know my family history. The Target is a1-2 cousin with 485 cm to me. How do I add his family history as much as we know, to my maternal history where are matches come from? I’ve tried starting with his biological grandparents but need to add my history, to get to identify his father. Father comes from my biological maternal side as the Target matches all my 5 1/2 siblings and my cousins from my matches.
Thanks, Rita W Neher
A 485-cM match is probably a half first cousin or a first cousin once removed. I’d probably start the WATO tree with your maternal great grandparents. You’ll have to use shared matches to figure out which set of GGPs to use. If you need more help, I recommend that you join the dedicated Facebook group here: https://www.facebook.com/groups/WhatAretheOdds
I am working with WATO and I have three siblings whose mother’s parental origins are unknown. The three of them share many maternal matches with people from both sides of their mother’s family. These matches range in cM values from 30 up to 542 cM and levels vary between the three siblings.
Should I average these values or enter each sibling as a separate match in the WATO tree?
You can use a technique called “twinning” (or tripletting, in this case) to include all of the match data in the same tree. For each match, create two more twins. Then for each “twin”, enter the shared DNA amount for one of the three siblings. For example, if John, Mary, and Michael are the sibling, and they all match Fanny, add two siblings for Fanny such that she is now triplets. For Fanny 1, use the cM amount for John; for Fanny 2, use the cM amount for Mary; and for Fanny 3, use the cM amount for Michael. Do the same for each of their matches in the tree.
Do you put the three siblings as the target person?
No, there’s still only one target person. You twin (or triplet) the matches.
Do WATO and other tools account for sex-related inheritance peculiarities? I just now realized that I don’t remember seeing any mention of that in discussion of that for WATO, etc.
No, WATO does not account for Y-DNA, mtDNA, or X-DNA inheritance. However, if you know that an hypothesis cannot be true because of evidence from these inheritance patterns, you can manually delete the hypothesis.
I’d suggest that authors of affected tools include a note about that, and point to an article explaining the subject; some of it sounded a bit convoluted last time I read about it.
Some things such as inheritance of mtDNA from a male parent isn’t impossible, but is extremely unlikely; there’s at least one documented naturally-occurring case, and my question was triggered after reading about mitochondrial disease being fixed using mitochondria from the male parent. I suspect the same might be true for other inheritance patterns.
Hello, and thank you for writing this post. I do have a question, though. One of the people in the WATO tree I created tested at MyHeritage (60 cM) and at Ancestry (44 cM, weighted, 60 cM, unweighted). Which figure should I use? 44 or 60? Thank you and best regards, Chris
I use the weighted figures from Ancestry. You’ll find that the difference between 44 and 60 doesn’t affect the probabilities much.