Is there a glitch in GEDmatch's "People who match both kits" tool? [closed]

Question

I don't know if this is a glitch or a change in the way DNA is compared by GEDmatch.com now or not, but I have found the strangest circumstance when comparing two kits in the "People Who Match Both Kits" utility. One would expect that if two GEDmatch kits have a number of other kits that match both in DNA, there should be some genealogical connection between the owners of Kit A and Kit B, right? I discovered that if I just took a random GEDmatch kit number and ran that utility, every kit had a number of kits that matched the other kit.  I was trying to find matches in a particular family to my DNA. I discovered that even if there was no DNA matching between me and another kit, if I ran the above utility tool, there were always kits that matched both of us. I started comparing random kits and discovered that there were always matches between the two kits! I find this very disconcerting. Does this mean that everyone really is related to everyone else? I somehow thought there would have to be a DNA connection between the two kits before they would have a number of kits that matched both of them. Can someone explain before my head explodes?

Comments

Re: re-categorization

Jillaine, don't you think this would be of interest to the tree at large? It's not really a tech question. It's more of a DNA question.

---

Fixed-it Edie. You are very right this has nothing to do with WikiTree Tech but is a discussion that may be relevant to WikiTreers in general.

Mags

---

Thanks, Mags!

Answer 1

It's not a glitch.  That tool tells you if someone is matching both A and B; it doesn't mean they all share a common ancestor.  The person that matches both kits could share one ancestor with Kit A and a completely different ancestor with Kit B.

Comments

I know, but isn't that crazy that you can pick someone - anyone - at random and you have shared matches?! I have not yet found anyone I don't have shared matches with.

---

I compared myself with you, Edie, using the Gedmatch default threshold of 10 cM. Gedmatch tells me:

Kit 1: M162952 (*Enviroellen)
Kit 2: A346257 (*EdieK)

Found 3864 matches for kit M162952.
Found 6404 matches for kit A346257.

Out of those thousands of matches, Gedmatch found only 49 kits (in several cases, there are multiple kits for one person, but I did not adjust the kit count to account for those duplicates) that match both of us at the 10 cM threshold or higher. That is a very small number, and the closest calculated relationship either one of us has to any of these matches is 4.1 generations. I assume that a good-sized fraction of these matches are spurious (basically, coincidences), so maybe there are about 25 people in the Gedmatch database who share a common ancestor with both you and me (but may not share the same ancestor with both of us) within 8 generations or so. That did not seem at all far-fetched to me, particularly considering that we both live in the United States and our ancestries include some of the same national origins.

After I did that, I discovered that Relationship Finder shows that you and I map out as 9th cousins once removed, but to my way of thinking, that is a relationship distance that would not be discernible in DNA and is basically the same as saying we have ancestry in the same populations.

Have you compared with anyone who is, for example, a Chinese person living in China?

---

No, I am planning on trying different nationalities. The key is finding some who have DNA results in GEDmatch.

---

But, see, isn't that crazy to think how interrelated we all are! It sure blows my bubble to think I will be able to break through my brick wall with that tool though.

---

Yes, Edie, it is definitely disappointing!  I have almost given up on exploring those people who match both of two other people.  They rarely get me anywhere!

---

A few years back a geneticist who specializes in European DNA wrote a paper that said if you have European ancestry the odds are that if you go back 10 generations you will find you are related.   That is based on the odds and population studies.  I can sort of believe hat because I belong to an organization of people who are descendants of glass and crystal makers.   We took all of our gedcoms and combined them into one big one.  About 90% of us share a common couple, Michel Andres and Apollonia Krieger his wife. who married in the 1600s in Wangenbourg.  They had 12 children.  We have over 450 people in the organization's email list....   We call them our mythic couple.  

I am one of the Adoption Angels on WT and I use the People Who Match 1 or 2 Kits a lot.  I love it!  It has helped me solve a number of adoptee puzzles.  There are two parts to the report.  I ignore the bottom part.   I only use the top part that shows matches to both kits.  Then I run the One to One comparison for any that look promising.  But I also key those results to the One to Many Report and can see points of unions where someone in the one to many report shows up on more than one of the 1 or 2 kits report.

---

Whoa, that is so interesting about the organization of descendants of glass and crystal makers! I believe that about being related to everyone ten generations or so back now that I've discovered this "glitch." I had been using the top part to try to find distant relatives that might help me fill in the gaps in my family tree, but I became discouraged when I saw that I had shared matches with almost everyone with whom I was comparing DNA. I guess I won't give up yet!

---

Here is the thing, DNA recombines at each generation so in the strict sense you may not match person B but your sibling might by strict definition.   So if you match your sibling as a full sibling and they match person B then you can infer that by paper trail you relate even if the DNA does not support that.  Why because your full sibling does match.  

My fourth cousin and I match something like 7 different people at over 20 cms but we do not match each other.  Why?  Because at 4th cousins the DNA starts to dilute.  This is why it is very important to not just rely on DNA but to work DNA with a paper trail that has supporting documentation.  

I have another 3rd cousin where we do match with DNA and I match her sister and I match her son.  But I do not match another member of the family who they all match as full relationship matches.  

So I have to be related to the other person because those three who I do match have full matches to that person.  It cannot be a non paternal event because they would not match at full relationships.  

DNA can help prove relationship but using a lack of a match cannot disprove a relationship... it just means the DNA can't prove it.   Now, that last statement is only correct for cousins beyond 2nd cousins.  

I hope this is making sense...

---

A chromosome map would make that perfectly clear.  My siblings and I share many segments, but not all, because for both our paternal and maternal chromosomes, the DNA of our grandparents has recombined.  So, for example, from my two paternal grandparents George Kelts and Margarete Van Wye, I may have Van Wye DNA where my siblings have Kelts DNA.

Edited to correct typo.

---

That makes sense. I think it encourages me a bit not to give up on the tool. I just need to look a little more closely at how some of those shared matches match other relatives. Unfortunately my only full-blood sister died before her DNA was tested...well, it was unfortunate that she died at all. But, my mother is still alive at 102 and her DNA has been tested, so that is helpful in my search. It is actually the search for her great-grandfather's family that I am just continuing for her. She started it in the late 1960s. It's much easier now.

---

Lucky you!  Your mother's DNA is more valuable than your sister's, for your maternal line.  For your paternal line, if you have paternal half-siblings, they will be useful.  So will paternal cousins.

---

I descend from the Stangers who started the glassworks in Glasboro, New Jersey. Can I belong to the group, or do you have to be descendants of some specific glassmaking family? :-)

---

Barry then we are related! I have the full genealogy going back centuries.  All the books written in the USA are dead wrong.  Solomon came from Wingen Sur Moder and we have the birth and marriage records to prove it.  Shoot me a private message and I will put you in direct contact with Luc Stenger (Stanger is the Americanization of
Stenger) who is the president of the World Heritage site for Vallerysthall.   

And yes GenVerrE is open to anyone with glass ancestors or who is just interested in glass and crystal making.  I can explain on this on a more direct email.  I will send you my contact info.  In the Private Message give me your direct email   I have files you are going to want to have.

Answer 2

Edie,

I do not think there is an error.  Just for fun I checked our DNA connection, we do not share any, at least at the default levels for Gedmatch.  (Wikitree shows us as 11th cousins once removed - so would not expect to see a significant DNA connection.)

The very first kit that shows up in the list at Gedmatch shares 76 cm with me, I know who she is but her profile is hidden.  She also shares 10 cM with you.  However, and this is the important part there is no overlap between you and me and the third person.

Remember when you run the report you are asking for all of those "People who match both kits, or 1 of 2 kits"  so if any exist they will show, however there is no reason to believe that kits 1 and 2 have a DNA connection.

So I have a common ancestor with the third person and you have a common ancestor with the third person as shown by DNA, but you and I do not have a DNA connection.

Comments

I see that we're related back to Thomas Moore. I didn't even know I was related to a Thomas Moore. I guess that would be a fun thing to do: Check my genealogical connections on WikiTree with all of these GEDmatch shared matches. That would help my brain not explode out of my head!

---

Edie,

I have worked through all of my Gedmatch matches that are on Wikitree - in many cases I had to add to the files for the person I matched.  While the effort was interesting, not very valuable.  I often found common ancestors, but because I have DNA tests on a fair number of close relatives I could tell that the found ancestor was not the one that provided the DNA match.   In some cases I could tell that the match had to be on my maternal side but the common ancestor found was on my paternal side and finding a valid 3 way au match is very difficult for these cousins, all in the 6th or more cousin range.

Hopefully as more individuals with DNA tests are on Wikitree and on Gedmatch I will be able to fill in some gaps.

Answer 3

This is the opposite of a glitch! It is a distinguishing feature of GEDmatch!

Testing companies only let you run searches and analysis with people they have identified as your genetic matches. But sometimes, you know of, say, a fourth cousin who also tested and you just don't have any DNA in common. This is frustrating if you are just using the website of the testing company where you both tested -- your cousin doesn't show up in your match list, and there is nothing else to be done.

But if you are both on GEDmatch, you can still see your shared matches, just as if you would if your cousin showed up in your match list at the testing company and you clicked "shared matches". And the people who match both of your kits on GEDmatch can be almost as useful as a the more typical list of shared matches -- the only deficiency is that you will not be able to triangulate any of those matches.

Answer 4

GedMatch can't match my kit with my kit. I ran this test after some strange strange things happing in matching my chromosome six. Same kit. But on chromosome 6, it came back that I was a low confidence of match with myself in several segments. I've asked them three times  how this can happen and they've refused to answer my question. Literally refused. I don't trust their "determinations" of relatedness anymore. I'm a reporter. If anyone else has noticed glitches in GedMatch matching, drop me a line.