Unable to triangulate DNA even though they all share DNA?

Question

I'm having a problem confirming DNA between three individuals and I'm unsure why.  The individual I'm trying to confirm, Felix Thompson, has three individuals that have taken DNA tests.  These three individuals have common ancestors of John Willis and Martha Smith (relations from the common ancestors to the DNA test takers are Baby Boy Phil's 5x great grandparents, Jennifer's 4x great grandparents, and
Donna's 3x great grandparents).

Comparing each of the DNA to each other, one on one, seems to work okay, but I get a null result when I try to triangulate with all three.

I'm new to the triangulation method so I'm not sure exactly what this means.  Can anyone help me with this?

Comments

There are four people you mention. Did you do six possible one-to-one comparisons between them? And when you do those comparisons, does the chromosome browser show that they are matching in each comparison in the same section of the same chromosome?

Answer 1

If I'm reading your question correctly, I think the chance of all three of your test takers sharing a common segment of DNA large enough to show as a match at most companies is very small. Based on your question Jennifer would be Donna's 4th cousin once removed (4C1R) and Phil would be Donna's 4C2R. Jennifer and Phil would be 5C1R. According to the research paper "Cryptic Distant Relatives Are Common in Both Isolated and Cosmopolitan Genetic Samples" the chances of these people having detectable matches would be roughly as follows. Jennifer/Donna 15%, Jennifer/Phil and Donna/Phil 4%. This is a rough estimate as the paper does not give probabilities for removed cousins, so I assumed the relationships to be 5th (4C1R) and 6th (4C2R and 5C1R) cousins. Remember that this means there's only a 4% chance that either Jennifer or Donna will have a match with Phil. The chance of those matches overlapping is exceedingly small. This is a very simple answer to an extremely complex question, in addition to being in an area of what is far from settled science, but hopefully it sheds some light on the issue. Perhaps some of our fellow Wikitreers with far more expertise (and much better funded keyboard replacement budgets) can fill in some of the details.

Answer 2

Felix,

I will assume that the data you are working with at either Gedmatch or MyHeritage has been in their system long enough for everything to be calculated.  (Gedmatch takes a day or two not sure how long MyHeritage takes.)

You do not mention the size of the overlaps.  When you compare two of the tests.  Is the overlap small?  When you look at the visual display - do the tests overlap?

As you have not mentioned other family relationships - are you certain that there are no other connections between the test takers?  

Were all of the tests done by the same testing service?  I have found cases where tests done at Ancestry show a significant match on one segment.  (There is only one segment match)   When the tests were moved to MyHeritage - there was no match shown.  

I am sure others will think of other issues - these are just some I have found.

If you have the information on Gedmatch - if you list the ID's perhaps others can look at it.

Comments

All of these are on GEDmatch and they should have been in there long enough.

AQ6848363 <-> QD3146867 is showing an overlap of 14.08cM on one segment, chromosome 3.

QD3146867 <-> T366951 is showing an overlap of 50.22cM on two segments, chromosome 3 and 4.

AQ6848363 <-> T366951 is showing an overlap of 14.04cM on one segment, chromosome 2.

And... I guess that answers my question.  The overlap isn't the same between all three.  I feel like I need a drawing board to figure out what's going on here.

---

Eric,

Try the DNA Confirmation Citation Maker, it's great for composing and/or validating DNA confirmation source statementsl.

---

Eric,

Since you are using GEDMatch, try 

Multi Kit Analysis under Tier 1 tools, then select Triangulation from the next page. This will quickly tell you if there's a triangulation match.

---

Ken, it's not necessary to use a paid Tier 1 tool in this case. It's pretty straight forward, the 3 people aren't matching even on the same chromosome, let alone the same overlapping segment. Eric has found that out himself already.

---

Yes, I understand that.  I saw his earlier post. This was for reference in case he  (or anyone) runs across this in the future.

---

The principle of DNA/segment triangulation is so simple that it doesn't require the Tier 1. If it can be done for free, why should anyone pay for it? You just have to compare 3 kits in this, so you note down/copy all matching segments.

Paste them into a word editor or Excel file and that's all that is needed.

“If you give a man a fish, he will be hungry tomorrow. If you teach a man to fish, he will be richer forever.”

Answer 3

I just commented on a the same problem at Roberta Estes blog when she wrote that DNA triangulation (aka segment triangulation) is possible at FTDNA. Hint: it's not

The problem is that in order to confirm segment triangulation, all 3 (or more) people need to match at an overlapping locus on the same chromosome. I have a visual representation (screenshot) of what I mean by overlapping locus here.

Now as Paul Chisarik has rightfully pointed out, the distant relationship in this specific case gives very little chance that indeed such segments qualifying for DNA triangulation have survived the random recombination events (due to the many generations and thus, many recombination events).

At the same time, it's a perfect example why you can't do segment triangulation at FTDNA as it doesn't show you where match B matches match C. Yes, they do match but on chromosome 2 and not on the chromosome 3 where the other two match (assuming that match A is QD3146867)

Here's Eric's segment matching data again for easier reference (given by himself in one reply):

All of these are on GEDmatch and they should have been in there long enough.

AQ6848363 <-> QD3146867 is showing an overlap of 14.08cM on one segment, chromosome 3.

QD3146867 <-> T366951 is showing an overlap of 50.22cM on two segments, chromosome 3 and 4.

AQ6848363 <-> T366951 is showing an overlap of 14.04cM on one segment, chromosome 2.