Recently, I published the article, Hitting a Genealogy Home Run Using Your Double-Sided Two-Faced Chromosomes While Avoiding Imposters. The “Home Run” article explains why you want to use a chromosome browser, what you’re seeing and what it means to you.
This article, and the rest in the “Triangulation in Action” series introduces triangulation at FamilyTreeDNA, MyHeritage, 23andMe, GEDmatch and DNAPainter, explaining how to use triangulation to confirm descent from a common ancestor. You may want to read the introductory article first.
This first section, “What is Triangulation” is a generic tutorial. If you don’t need the tutorial, skip to the “Transfers” or “Triangulation at GEDmatch” sections.
What is Triangulation?
Think of triangulation as a three-legged stool – a triangle. Triangulation requires three things:
- At least three (not closely related) people must match
- On the same reasonably sized segment of DNA and
- Descend from a common ancestor
Triangulation is the foundation of confirming descent from a common ancestor, and thereby assigning a specific segment to that ancestor. Without triangulation, you might just have a match to someone else by chance. You can confirm mathematical triangulation, numbers 1 and 2, above, without knowing the identity of the common ancestor.
Reasonably sized segments are generally considered to be 7cM or above on chromosomes 1-22 and 15cM or above for the X chromosome.
Triangulation means that all three, or more, people much match on a common segment. However, what you’re likely to see is that some people don’t match on the entire segment, meaning more or less than others as demonstrated in the following examples.
You can see that I match 5 different cousins who I know descend from my father’s side on chromosome 15 above. “I” am the grey background against which everyone else is being compared.
I triangulate with these matches in different ways, forming multiple triangulation groups that I’ve discussed individually, below.
Triangulation Group 1
Group 1 – On the left group of matches, above, I triangulate with the blue, red and orange person on the amount of DNA that is common between all of them, shown in the black box. This is triangulation group 1.
Triangulation Group 2
Group 2 – However, if you look just at the blue and orange triangulated matches bracketed in green, I triangulate on slightly more. This group excludes the red person because their beginning point is not the same, or even close. This is triangulation group 2.
Triangulation Group 3 and 4
Group 3 – In the right group of matches, there are two large triangulation groups. Triangulation group 3 includes the common portions of blue, red, teal and orange matches.
Group 4 – Triangulation group 4 is the skinny group at right and includes the common portion of the blue, teal and dark blue matches.
Triangulation Groups 5 and 6
Group 5 – There are also two more triangulation groups. The larger green bracketed group includes only the blue and teal people because their end locations are to the right of the end locations of the red and orange matches. This is triangulation group 5.
Group 6 – The smaller green bracketed group includes only the blue and teal person because their start locations are before the dark blue person. This is triangulation group 6.
There’s actually one more triangulation group. Can you see it?
Triangulation Group 7
Group 7 – The tan group includes the red, teal and orange matches but only the areas where they all overlap. This excludes the top blue match because their start location is different. Triangulation group 7 only extends to the end of the red and orange matches, because those are the same locations, while the teal match extends further to the right. That extension is excluded, of course.
Matches with only slight start and end differences are probably descended from the same ancestor, but we can’t say that for sure (at this point) so we only include actual mathematically matching segments in a triangulation group.
You can see that triangulation groups often overlap because group members share more or less DNA with each other. Normally we don’t bother to number the groups – we just look at the alignment. I numbered them for illustration purposes.
Shared or In-Common-With Matching
Triangulation is not the same thing as a 3-way shared “in-common-with” match. You may share DNA with those two people, but on entirely different segments from entirely different ancestors. If those other two people match each other, it can be on a segment where you don’t match either of them, and thanks to an ancestor that they share who isn’t in your line at all. Shared matches are a great hint, especially in addition to other information, but shared matches don’t necessarily mean triangulation although it’s a great place to start looking.
I have shared matches where I match one person on my maternal side, one on my paternal side, and they match each other through a completely different ancestor on an entirely different segment. However, we don’t triangulate because we don’t all match each other on the SAME segment of DNA. Yes, it can be confusing.
Just remember, each of your segments, and matches, has its own individual history.
Imputation Can Affect Matching
Over the years the chips on which our DNA is processed at the vendors have changed. Each new generation of chips tests a different number of markers, and sometimes different markers – with the overlaps between the entire suite of chips being less than optimal.
I can verify that most vendors use imputation to level the playing field, and even though two vendors have never verified that fact, I’m relatively certain that they all do. That’s the only way they could match to their own prior “only somewhat compatible” chip versions.
The net-net of this is that you may see some differences in matching segments at different vendors, even when you’re comparing the same people. Imputation generally “fills in the blanks,” but doesn’t create large swatches of non-existent DNA. I wrote about the concept of imputation here.
What I’d like for you to take away from this discussion is to be focused on the big picture – if and how people triangulate which is the function important to genealogy. Not if the start and end segments are exactly the same.
GEDmatch does not utilize imputation, but in order to allow matching to chips with fewer location, they have relaxed their matching thresholds which sometimes generates larger matching segments than those shown at the original vendor.
All vendors except Ancestry offer some type of triangulation.
If you and your Ancestry matches have uploaded to GEDmatch, Family Tree DNA or MyHeritage, you can triangulate with them there. Otherwise, you can’t triangulate Ancestry results, so encourage your Ancestry matches to transfer.
I wrote more specifically about triangulation here and here.
Transfer your results in order to obtain the maximum number of matches possible. Every vendor has people in thier data base that haven’t tested elsewhere.
Have you tested family members, especially everyone in the older generations? You can transfer their kits from Ancestry or 23andMe if they’ve tested there to FamilyTreeDNA, MyHeritage and GEDmatch.
Here’s how to transfer:
- Ancestry Step by Step Guide: How to Upload-Download DNA Files
- 23andMe Step by Step Guide: How to Upload-Download DNA Files
- MyHeritage Step by Step Guide: How to Upload-Download DNA Files
- Family Tree DNA Step by Step Guide: How to Upload-Download DNA Files
The upload to GEDmatch is very easy. After setting up your account at GEDmatch, here, you’ll find the link to upload files on the right side of your page.
Triangulation at GEDmatch
Many GEDmatch functions are free, with only the more advanced falling under the Tier 1 subscription model which costs $10 per month.
Basic One to One matching (shown below) is one of the free tools.
I suggest that you enter your kit number and accept the defaults until you’ve gotten your feet wet with GEDmatch tools, changing default settings and what they mean to results.
You’ll see your matches and related information along with a link to GEDCOM files, if uploaded, as well as WikiTree links. Please provide some form of tree.
Note that matching is more “generous” since GEDmatch implemented it’s relaxed thresholds in an attempt to include files generated from vendors who use fewer DNA locations. In other words, you may find the same match at GEDmatch differs, and is larger, than the match to the same person at the original vendor. Check when possible.
While basic matching doesn’t provide you with triangulation information, triangulation information is available in multiple ways.
You can see the kit numbers of your matches, shown above in the red box, so you can look at their matches the same way you view your own. However, there are easier ways to see who you match in common with someone. (The unblurred kits are mine, so no privacy issues.)
Another free tool is called “People who Match both, or 1 of 2 kits.”
Click on that link.
You’ll enter two kit numbers. I suggest you accept the defaults for the rest of the selection criteria.
You’ll see how both of the kits you’ve entered, shown along the top, match various people on the match list that they match in common. The great news is that those “A” kits are Ancestry kits!! Unfortunately, newer kits can’t be identified by the initial letter.
To explain further, kits in column T52 (number obscured) and column M13 (number obscured) both match the top person at far left, *Roberta. Kit T52 matches *Roberta as does kit M13. Every person on this match list matches the person whose kit we are viewing plus T52 and M13, but we don’t know where they match or if they all match each other on a shared segment. To discover that information, we need to triangulate either by comparing all 3 to each other one by one and recording the results, looking for a 3-way overlap of 7 cM or greater, or using a triangulation tool.
Triangulation is one of the Tier 1 paid features.
Click on “Triangulation,” enter your kit number and accept the search criteria by clicking on “Submit” at the bottom of the page.
On this page, you’re viewing sets of people who triangulate with you. In this example, we’re looking at chromosome 7 where I triangulate with 6 people. You can view how the segments stack up at the far right, in green, as well as the start and end positions, or addresses to the left of the green bars.
In this display, you’re looking for everyone to match everyone else in the same location.
For example, I triangulate with LRRY in combination with Rebecca, Buster and Charlene, respectively. Each of those people also triangulate with the others, and me, as well.
We form a triangulation group of 5 individuals who match on a reasonably sized segment, between 19 and 34 cM, of the same chromosome.
The person in this group whose genealogy I know the furthest back in time is Buster, and we both descend from Lazarus Estes and Elizabeth Vannoy. Therefore, I know that the triangulated segment that I share with all of these people descended to me through those ancestors. These people may not (all) descend directly from Lazarus and Elizabeth, but if not, they descend from the ancestors of either or both Lazarus Estes and Elizabeth Vannoy.
Other Resources to Help Identify Common Ancestors
GEDmatch offers a variety of tools, including:
- Segment Search which allows you to search for a specific segment to match
- Phasing generates kits reflecting the portion of DNA inherited from each parent if at least one parent has tested
- Lazarus tool reconstructs ancestors from descendants, although this does not always work accurately depending on the descendants and variables involved
- Evil twin which allows you to utilize the half of your parents’ DNA that you did not inherit, if the parent has tested
- GEDCOM file search and compare (please upload a GEDCOM file)
- Clusters that show groups of people who match you and each other
I wrote recently about how to work with triangulation at FamilyTreeDNA, MyHeritage and 23andMe. Join me soon to see how to work with triangulation at DNAPainter.
I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase the price you pay but helps me to keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.
Thank you so much.
DNA Purchases and Free Transfers
- FamilyTreeDNA – Y, mitochondrial and autosomal DNA testing
- MyHeritage DNA – ancestry autosomal DNA only, not health
- MyHeritage DNA plus Health
- MyHeritage FREE DNA file upload – transfer your results from other vendors free
- AncestryDNA – autosomal DNA only
- 23andMe Ancestry – autosomal DNA only, no Health
- 23andMe Ancestry Plus Health
Genealogy Products and Services
- MyHeritage FREE Tree Builder – genealogy software for your computer
- MyHeritage Subscription with Free Trial
- Legacy Family Tree Webinars – genealogy and DNA classes, subscription based, some free
- Legacy Family Tree Software – genealogy software for your computer
- Legacy Tree Genealogists – professional genealogy research
Fun DNA Stuff
- Celebrate DNA – customized DNA themed t-shirts, bags and other items
This is such a timely article for me. I am working on a new German Connections project on Gedmatch. I was about to write something up for participants to try to explain some of the results they are seeing. Thank you so much for your clear explanations.
MORE GOOD STUFF! Thank you for being such a good ‘splainer! It is great being able to point folks to your work.
Now that I use both network node and matrix graphing side by side with a chromosome browser where I use triangulation as the gold standard, I am getting a much stronger understanding of ICW patterns – especially observing connections among matches sharing multiple segments vs single segments. Now that DNAPainter takes cluster input from the very same DNAGedcom and Gedmatch files I use for clustering, things are even more interesting.
Having double cousins in my grandparents’ generation (thankfully not in my direct line) who cross my father’s maternal and paternal line, it is now sobering to see how many matches I originally would have called wrong with only Ancestry ICW shared match lists, based on subsequently finding them in the other platforms with chromosome browsers. and putting them to the triangulation test.
Have you written a post on Genome Mate Pro? I find GMP very helpful and it eliminates a lot of the manual work required by use of spreadsheets.
No I have not. I found the setup onerous and never managed to get through that.
Thanks for explaining the triangulation option at Gedmatch. I did a segment triangulation for my kit and noted the triangulated matches for one kit. I then did a segment triangulation for that kit and noted the triangulated matches for my kit. Some of the matches are the same but many are not. Could you explain why this is?
From what you’ve described, if I’m understanding, it sounds like they should be. I’d suggest contacting Gedmatch. I have noticed sometimes their matching is “off” since they incorporated the Genesis kits.
I think I figured it out. I asked for triangulated matches for my top 500 matches. The results show that I triangulate with kits A and B. When I asked for all the triangulated matches for kit A, the results do not show that kit A triangulates with me and kit B because kit B is not in kit A’s top 500 matches.
Ahh, that makes sense.
Pingback: Triangulation Resources In One Place | DNAeXplained – Genetic Genealogy
Any thoughts on how small of segments should be to be considered a legit match? I’m in a project where we are looking to connect members from a particular surname back to the 1600s. Is a triangulated segment of 5 cM too small to pursue? Thank you!
5 cM is more likely to be false than true. However, if it triangulates with everyone else in that line who triangulates on a larger segment that includes that smaller segment, then it would be accurate.
Thank you for the reply! No larger segments since we are going back 10-12 generations. Out of 50 people in the project we just have a handful of triangulated segments of 5-7 cM where we have 3-4 people in the triangulation group. So likely false segments or true but not provable.
I’ve been following DNAeXplained since we met at Cape Hatteras in about 2011. And I must say, it seems to me you have gradually gotten more conservative about this segment length issue. Personally I thought you were right when you fought for the ability to see very short segments, in known distant (paper trail) cousins. Anyway I’m glad to see somebody else raising the concern, rather currently. We have a GedMatch project now for Germanna (VA) colony descendants; and like the 17th century MRCA issue Patrick had last month, our matches can easily be from ten generations back. That does not offer a realistic hope of long matching segments, but if we lower the bar (e.g. to the minimum of 3 cM), we do share matches of the same short segments, at the same loci, on the same chromosomes. You (Roberta) use to argue that that was meaningful. What has changed?
In short, the smaller the segment, the more evidence you need. Most people jump to unwarranted conclusions. The other thing that has changed is that we have so many more matches now that we don’t really need to chase those small ones, for the most part. Smaller matches are more likely to match simply by chance. If you are unfortunate enough to carry the most common alleles in those locations, you will match a lot of people by chance. The other thing that has changed is imputation. The smaller the segment, the more likely it is to be “filled in” by imputation and you have no idea if you’re looking at imputed values or real values. Lastly, the only way to deal with smaller segments is to make sure everyone matches each other, they are all “not” the most common values in build 38, they do have common ancestor, their trees are complete, they phase with parents and triangulate. Imputation was a real game-changer, and the vendors did not have a choice in this matter. Illumina forced it. Believe it or not, there are some people who use a higher threshold than 7, unfortunately. Ancestry deleted everything below 8.
Very helpful, thanks. I tend to trust you, or I wouldn’t have asked. I’d just be happier trusting the test results, also. Gedcoms are a good check, too, if we have those (in GedMatch, or elsewhere) for most project members. I don’t think they are inherently more trustworthy than short segments, but they surely improve our odds.
I take your point about the ever-larger database alleviating the need, these days, for looking at the tinier hints — more likely to come from a source other than the one we know about. With GedMatch, and a rationally defined sample group, we may have the luxury of working backward to the MRCA, stepwise if we so choose. Like the countdown for a rocket launch, but starting at 8 — and not going down very far!
Hey Roberta, I manage 25 DNA kits at FTDNA. I am thinking about getting an account with GEDmatch. I know to get the Tier 1 feature is $10.00 a month, would I be able to have all of my 25 kits under one account or is each kit a separate account?
I believe it’s for all the accounts you manage under your ID. That’s how mine works.