Interpreting Results from Double Match Triangulator

 
Interpreting Results from Double Match Triangulator
 
The main tool for interpreting results from double match triangulation is DMT's Map page.
 
These are currently my own suggestions for you. Once experienced genetic genealogists have had time to use the DMT tool, they hopefully will write about how they used it, and more techniques will be forthcoming.
 
But personally, I think the goal should be to use DMT to help map every segment of all of your DNA chromosomes to the ancestor it comes from. A triangulation group likely represents one ancestor, with subgroups representing ancestors of the ancestor and larger groups representing descendants of that ancestor.
 
Remember that there may be matches on both chromosomes of each pair. There will for sure if you compare siblings. But these can be for anyone, because any person could be related to both your father and mother, sometimes in unknown ways. Matches on both parental chromsomes will show up as triangulation groups that overlap, so be aware if your triangulation groups are not aligning well that this may be the cause.
 
Once you get segment match files from many of your relatives on both your parents sides (I'm currently at 118) and combine them together using the "Combine all results" option, you'll find many overlapping match groups to interpret.
 
 

Traditional Triangulation versus Double Match Triangulation

 
Traditional triangulation uses single matching. Segments where Person A matches two or more other people need to be investigated. What must be done is to ensure that the other people match each other. Too often that is not done, so a shortcut of using segments 15 cM or larger is often taken. That length has been shown to include very few false matches.
 
Double match triangulation is different. It does that match for you. The Person A to Person C matches are in your segment match file. The Person B to Person C matches are in Person B's segment match file and these are the traditional "third match". Then all you need are the Person A to Person B matches to figure out which of the double matches triangulate. Double match triangulation finds every triangulating segment between any of your matches and a Person B.
 
Not only that, due to putting two sets of matches together, the likelihood of false matches due to chance is considerably reduced. Jim Bartlett has said he believes fully triangulated segments are almost all Identical by Descent down to 7 cM and possibly even down to 5 cM. My evaluation indicates that the same is likely also true for all double matches, including missing AB matches.
 
 

You Can Only Show That Three People Match. You Cannot Show They Don't Match

 
Double matching and triangulation only works one way. If three people double match or triangulate, and it is not a random match, then you likely have a relationship with a common ancestor. If two people don't double match or triangulate, that does NOT mean they are not related. One of them may not have been passed the same segment from their ancestor as the other two. Or maybe they do match on the segment, but their total matches were not enough to be included in each other's match lists by the testing company.
 
 

Identical by Descent (IBD) versus Triangulation

 
A misconception by many people (myself included until I was corrected by Debbie Kennett) is that too often, people think triangulated segments mean that the segment is Identical by Descent (IBD), meaning it was truly passed down to the three people by a common ancestor. That is not necessarily true. A triangulated segment may have one of the 3 matches being by chance, it could have a match being maternally to one of the other people, and paternally to the other, there may be an algorithmic error by the DNA company and other reasons. So it's important to remember that:
 
- A triangulated segment is not necessarily Identical by Descent
 
However,
 
- An Identical by Descent segment must triangulate.
 
Therefore, by identifying triangulated segments, DMT is showing you all the possible segments that may be Identical by Descent for Person A and Person B. Not all are necessarily IBD, but there will be no others that are IBD.
 
 

Double Match Filtering (DMF), Phasing, and Future Possibilities for DMT

 
DMT filters out all people who are not related to Person A and Person B. If you select a parent of Person A, then DMT will display only the people who match the person and the person's parent on the same segment. The segment is likely to have been passed down from that parent to the child (although if both parents are related, the parents may have segments in common). This filtering in effect eliminates all double matches and triangulations that cannot be from that parent's side. DMF with a parent is similar to Phasing.  Phasing is a technique that compares a person's DNA to both their parents and identifies the parent. Traditional phasing does this at the level of the individual allele's (A, C, G, T). Whereas DMT filters segment matches.
 
If you compare yourself to both parents, then the triangulation groups that include matches between you and one parent and don't match with the other parent can be assigned to that first parent. Likely you can logic out the triangulation groups that have matches to some parents as there might have been a few smaller by-chance matches that snuck in.  Each of the triangulation groups can then be assigned to the Paternal/Maternal side, which would be the first level for the mapping of the chromosomes to ancestors.  See what Jim Bartlett says in section A4 of The Attributes of a TG, where he indicates that a triangulation group is the equivalent of phased data. This type of filtering would be a brand new technique that I would call:  Double Match Filtering (DMF)
 
The first step to do this uses Double Match Theorem 1 and Corollory 1.
 
I am hoping a future version of DMT will be able to separate your matches into your paternal and maternal sides for you. Again, more research is needed.
 
Remember that double match filtering can apply to more than just a person and their parent. You can select any Person A and any Person B. If they are, say, 2nd cousins, then you are filtering out anyone that is not related to both Person A and Person B. The people who are not random matches are therefore likely related through one or more common ancestors.
 
I expect that using double match filtering for a number of different relatives and combining the results ... and then with a little more innovation and thinking outside the DNA box ... it may be possible for DMT to map every segment to its common ancestors and to the descendants of the common ancestors. This would provide (1) a nearly complete segment map for Person A, and (2) a connected family tree of nameless ancestors and their DNA-descendants. This is the ultimate goal of DMT. Then traditional genealogy would be used to fill in the names and find out about the ancestors' lives.
 
 
1

Full Triangulation

1. Full Triangulation
 
Look first for the matches that are marked in green in column N (STATUS) as Full Triangulation. The matches and people shown here form triangulation groups and may have a common ancestor who passed this segment down to the people who do not randomly match.  Look in Column G (CM-AC) to see the size in cM of the match. Anything more than 7 cM is likely IBD (Identical By Descent) indicating that the segment comes from a common ancestor. Double matching eliminates many by chance matches so even some of the smaller than 7 cM matches may be IBD, especially because we know that all the people in these files are matches of Person A. Look at the smaller segments and see how many boundaries match with others nearby. Those that don't match any others may be by chance, but those aligning with others could very well be IBD.
 
 
2

AB Match

2. AB Match
 
Next find the AB segment that produces the triangulation. Note the base addresses of the Start (172592941) and End (176835759) of the double match part of the segment (shown in green) as well as the length of the AB match (6.91 cM).
 
3

Start and End Base Addresses

3. Start and End Base Addresses
 
The Start and End Base Addresses are crossovers that happened since the common ancestor that reduced the size of the common AB segment. The Start and End are each "owned" either by Person A or Person B. Determining which crossover is owned by whom is important, because others who share that crossover and don't match by chance are going to be descended on the same side as the owner of the crossover.
 
Note that either Person A or Person B may match alone to Person C prior to the Start or following the End. These matches prior and following may be relevant, but they are not double matched and therefore are more likely to include some random matches by chance at either end. There may still be some useful information that can be determined from them, especially if several Start addresses or several End addresses match.
 
4

Unique Small Segments

4. Unique Small Segments
 
For double match segments under 7 cM, it is very possible they are random by chance matches if the Start and End addresses of the double match shown in green are unique. So avoid assuming anything about those matches before further investigation.
 
What you want to find are multiple matches at the same crossover points. There's safety in numbers and many matches with the same crossovers will lessen the possibility of chance happenings. This area has not been studied, so there are no rules of thumb as to how many matches you need to indicate some sort of common ancestor or descendant of said ancestor had a crossover point, and how small the segments can be for a given number of matches.
 
Be careful not to use the Start and End of the AC and BC matches shown in red or blue as the delimiters for segments. Since Person A and Person B "own" these crossovers, and it is their match files we are using, they will cause all segments to truncate at their boundaries, whether they are from a common ancestor or whether they are by chance. Also, those portions are only single matches and by chance matches can be as long as 15 cM over those parts of the segment.
 
5

Deep Ancestors

5. Deep Ancestors
If the triangulation group is from a common ancestor, then there may be smaller identifiable groups within it that are either from a later crossover down one descendant line (in which case all the people having that crossover are in that line) or are from an ancestor of the common ancestor. In the latter case. everyone with the segment to the left or the segment to the right can be placed into a deeper ancestor triangulation group. Some research will be needed on how exactly to determine and do this and it may be possible one day for DMT to identify these for you.
 
6

Missing AB Matches

6. Missing AB Matches
 
Despite not being fully triangulated, missing AB matches still have a double match that Person A and Person C share with Person B and Person C. These may again be chance matches. Or these might be matches that Person A matched to one chromosome of a chromosome pair of the common ancestor with Person C, and Person B matched to the other chromosome of the chromosome pair of the common ancestor with Person C. However, due to the double matching reducing the likelihood of a chance match, any missing AB Match more than 7 cM should be looked at carefully. They will not be segments passed down from a common ancestor, but are instead segments on different chromosomes of the chromosome pair. That has a special interpretation, that I started looking at in a blog post: Triangulation and Missing a-b Segments and Revisiting Missing A-B Matches.
 
Note that missing AB matches are different than single matches that don't triangulate. In the latter case, we are matching Person A to Persons C1 and C2. But we find C1 and C2 don't match on the segment and may or may not related. Whereas a missing AB match is where Person A matches Person C and Person B matches Person C. Person A does not match Person B on the segment, but we usually use a Person B who is related and matches on other segments.
 
Missing AB matches are a new concept that first came out through double matching. More study needs to be done to determine the best way to use this information.
 
This information about double matches that don't triangulate is very likely useful and relevant, so it is included in the DMT map file for your analysis. I believe there is valuable hidden information in this that should not be thrown away.
 
7

Lots and Lots and Lots of Stuff Here

7. Lots and Lots and Lots of Stuff Here
 
There is a lot of information given in the Map file. How to use it is something I hope will become clear as more expert genetic genealogists discover this DMT tool and start working with it.
 
I believe the boundary points and the triangulation groups could ultimately enable mapping of all the segments to common ancestors. We should be able to build a tree of nameless DNA ancestors, known because of the segments they pass down and the descendants that pass them down to. Then the genealogist could do the investigation to determine who each of those nameless DNA ancestors are.
 
If you develop techniques to use the DMT Map file information, please let me know at:  www.doublematchtriangulator.com/feedback and if it can be automated, I'll attempt to add it into DMT so everyone can benefit.