What’s in Your Toolbox? — DNA Painter’s cM Estimator

There’s no secret that DNA Painter is my favorite third-party site for DNA analysis.  Its tools for chromosome mapping, relationship prediction, and hypothesis testing are widely acclaimed and have been used by hundreds of thousands genealogists around the world.

There are also hidden gems at DNA Painter that are less widely known.  One of them is the recently-released cM Estimator.

When might you want to use this tool?  Consider the example below from 23andMe.  Three people—Mike, Lana, and Nikki—match the tester, Brandon, on chromosome 10.

 

We know that Mike, in purple, is related on Brandon’s mother’s side, but we don’t know how either Lana (orange) or Nikki (yellow) is related.

Recall that we have two copies of each autosome (autosomal chromosome), so some of our relatives will match on the maternal chromosome and some will match on the paternal one.  The testing companies, though, can’t tell which is which, so they show us where a relative matches, but not on which chromosome copy.  We can have relatives matching us in what appears to be the same place but not matching one another because some are maternal relatives and some are paternal.

The chromosome browser at 23andMe is especially nice because not only can I see which segments Brandon shares with his DNA matches, I can see which segments they share with one another.

In the next image, I’ve rearranged things so that Mike is the reference point (grey background), Brandon is in purple, and the next two rows are Lana and Nikki.

 

Neither Lana nor Nikki matches Mike in this view.  Because the 22-cM segment that Nikki shares with Brandon (yellow in first image) is completely overlapped by the larger one that Mike shares with Brandon (purple in first image), and because Nikki doesn’t match Mike (second image), we can safely conclude that Nikki is related to Brandon through Brandon’s father.

But what about Lana?  She doesn’t match Mike there either, so can we conclude that she is also paternally related to Brandon?  If so, we might expect that Lana and Nikki would match one another, because their segments overlap slightly in the first image.  But they don’t.  What the what?

 

Enter the cM Estimator

Could it be that the region of overlap between Lana and Nikki on Brandon’s chromosome is too small for 23andMe to pick up?  How big is it?  There’s no way to tell from the 23andMe chromosome browser.

That’s where the cM Estimator is so valuable.  I can plug in the start and stop points for any section of DNA on any chromosome, and it will tell me how many centimorgans it is.

To find the start and stop points of the overlap between Lana and Nikki, I go back to the comparison of Brandon to his matches at 23andMe (first image) and hover my cursor over each segment to get a pop-up with detailed information.

The overlap on chromosome 10 starts at the beginning of Nikki’s segment (position 117,780,090) and ends at the terminus of Lana’s segment (position 119,542,582).  When I plug that information into the cM Estimator, and it tells me that the overlap is only 3.6 cM.

23andMe requires that the first matching segment between two people be 7 cM or more, and this one is half that.  I wouldn’t expect 23andMe to show a match there, even if one really exists.

Comparing Mike and Lana

But what about the overlap between Mike and Lana?  I can use the same approach to figure out the centimorgans.

In this case, the overlap spans from the start of Mike’s segment (position 113,787,146) to the end of Lana’s (position 119,542,582).

The cM Estimator tells me that region is 9.4 cM, well above the matching threshold of 7 cM.  In other words, Lana genuinely doesn’t match Mike, so she must be related to Brandon on Brandon’s father’s side, like Nikki is.