Double Match Triangulator, Version 5.0 - Wed, 18 May 2022

I’ve released version 5.0 of  Double Match Triangulator (DMT), a tool to help you analyze your autosomal segment matches and help you determine how you might be related to your DNA matches.

Six Months of Analysis

Over the winter and spring, I was working hard on on a project with Terry, my wife’s 3rd cousin, to see if we could confirm via DNA that their great-great-grandparents and 4 others were in fact siblings. Terry had got 44 people who were descendants of the 6 potential siblings to DNA test at Family Tree DNA. He was able to supply me with Segment Match Files for 26 people.

My goal was to see if I could use DMT to provide us with useful segment information that could confirm/deny the sibling relationships. I was hoping that maybe I could use DMT to help build a chromosome map of each tester’s ancestors up to the potential sibling, and then see if that tells us anything.

I was not able to get quite that far, and I’ll hopefully blog more about this once we get some reportable results from this study.

But what did happen is that I made a number of useful changes to Double Match Triangulator that helped me, and now I’m making these changes available in my release of Version 5.0 of DMT.

Triangulations on One Parent

Up until now, DMT was based on the idea that a segment match with a relative generally matches only on one parent.

For example, a 2nd cousin with whom your common ancestor is your father’s mother’s parents should have all the matches on your father’s chromosome.

A 1st cousin with whom your common ancestor is your maternal grandparents should have all their matches on your mother’s chromosome.

Your nephew or niece or child will have each segment matching either on your father’s side or on your mother’s side.

The major exception to the rule are siblings. They will have on average 1/3 of their matches on your father’s side, 1/3 on your mother’s side, and 1/3 on both.

There are other exceptions (false matches, being related more than one way, endogamy, etc.) but these are less common and the thinking was that segment matches would be more often correct than incorrect. And using the best consensus should provide the correct answer more often.

What DMT does is determine all the triangulations that Persons A and B have with other people. For those triangulations where the MRCA (Most Recent Common Ancestor) of Person A with Person C was known, the ancestral path was “painted” to the triangulated segment.  Once that was done, the consensus ancestral path was determined for triangulations at each Mbp (Mega base pair).

Of note here is that there was only one ancestral path for all triangulations at any Mbp. All other triangulations and non-triangulations for people without MRCAs were then assigned the same ancestral path. Every person could only match either their father or their mother on one segment, but not both.

Siblings were assigned the consensus ancestral path over their matching segment. It would not include both parents when it double matched, but it wouldn’t be wrong either.

This has always seemed to work fairly well.

Triangulations on Two Parents

For our study, we had a number of siblings involved, including Terry and his brother and sister. I did want to be able to make more use of the sibling data with DMT. So I experimented and came out with a new idea.

Instead of one consensus ancestral path from the triangulations over a segment, how about determining two:  One for any triangulations on the father’s side, and one for any triangulations on the mother’s side.

If all triangulations are on just one parent’s side, then we’d have exactly what we had before. But if both sides had some, then we’d have new information to use.

So now we have four possibilities:

1. Triangulations only on father’s side.
   Other triangulations can be mapped to the father and the ancestral path.
   Other non-triangulation can be mapped to the mother.
2. Triangulations only on mother’s side.
   Other triangulations can be mapped to the mother and the ancestral path.
   Other non-triangulation can be mapped to the father.
3. Triangulations on both father’s side and mother’s side.
   Other triangulations and non-triangulations have unknown side.
4. No triangulations.

The nice thing here is that cases 1 and 2 happen a lot more often than case 3. So you can quite often map all the other segment matches whose parent cannot be determined on their own to the correct parent depending on whether that segment triangulates or not.

And in case 4, you will occasionally have a single match on its own with someone whose MRCA tells you what parent it’s on. One single match, or even two on opposite parents is fine. But there shouldn’t be more than one on the same parent because if there were, then they should be triangulating. That’s how the logic of all this works out.

The long and the short of this all is that I made this major change to DMT. I have done extensive testing on it. Overall it makes better use of sibling information and seems to do a slightly better job of assigning ancestral paths than it did before.

Other Changes

There are a lot of other changes I snuck into DMT while doing this. Check the DMT history page for a full list of all changes.

If you have an older version of DMT on your computer, you can run it and click the “Check online for new version” link and follow the instructions for a quick and easy upgrade. If you don’t have DMT, you can download and try it at:  www.doublematchtriangulator.com

—

By the way, Jonny Perl, developer of DNA Painter will be the keynote speaker on Friday Aug 19 for the SCGS Jamboree 2022. Jonny will be talking about “The Need for Third Party DNA Tools” and will include DMT in his talk.