DNA mystery: any tips how to proceed?

[Blythe]

I had my DNA tested at Ancestry.com. I don't know who my father is, so I was delighted to discover two hitherto unknown "first cousins" in my matches who are about the same age as I am. It turns out they are brother and sister and they are interested to help me find out how we are related. So far so good, but I have come across a puzzle that doesn't make sense to me, and I would be grateful for any tips on how to solve it.


It turns out that one of their first cousins on their mother's side of the family (offspring of their mother's sister) is a 1st or 2nd cousin match to me as well. Two other cousins from their mother's side of the family (offspring of their mother's brothers) don't match me at all.


What I don't understand is, how can I be related to some 1st cousins from their mother's family but not to others? Is it possible that the positive matches are wrong?

Known relatives on my mother's side of the family show up as matches to me with the correct relationship, so at least I know that my DNA sample wasn't switched with someone else's.


I uploaded my raw data to MyHeritage (that site correctly identified my relationship to my daughter who had been tested there), but found no other matches closer than 2nd cousin. I have submitted another sample to 23andme to see what I find out there but they haven't received it yet.


Is there anything else I should be doing to solve the mystery? I would be grateful for any ideas.

[John McCoy]

First, are you basing your estimate of relationships solely on Ancestry's estimate? It would be a good idea at least to keep track of the actual shared cM values and to be aware of the other possible relationships that might be consistent with the shared cM. Even better if you can manage to get the relevant kits uploaded to GEDmatch so you can identify the shared segments and be sure that the kits match in all combinations, consistent with the apparent relationships.

However, depending on exactly what the relationships are, it is not impossible to end up with kits in the 3rd cousin range or more remote that don't match each other at all, but which do match other actual 3rd cousins. The expected level of matching for 2nd cousins is so high, I would not expect ever to find a pair of them that don't match.

It is also entirely possible that there are other complications, such as uncertain parentage of one or more of the "cousins" or, for that matter, of their parents. If that is the case, the degree of matching (and possibly also of X chromosome matches) among those cousins should be checked carefully.

Regarding the matches that you have, if the shared cM values are within the range as suggested (even a first cousin once removed would be in the range of 215-650 cM, according to the popular and very useful "DNA Detectices Autosomal Statistics Chart", easily found on the internet through a Google search), there is no way these matches can be "false". The problem is to combine the genetic evidence with the paper trail evidence and then to sketch out ALL of the possible family trees that could yield the observed result. Keep digging!

[Blythe]

First, are you basing your estimate of relationships solely on Ancestry's estimate? It would be a good idea at least to keep track of the actual shared cM values and to be aware of the other possible relationships that might be consistent with the shared cM. Even better if you can manage to get the relevant kits uploaded to GEDmatch so you can identify the shared segments and be sure that the kits match in all combinations, consistent with the apparent relationships.

However, depending on exactly what the relationships are, it is not impossible to end up with kits in the 3rd cousin range or more remote that don't match each other at all, but which do match other actual 3rd cousins. The expected level of matching for 2nd cousins is so high, I would not expect ever to find a pair of them that don't match.

It is also entirely possible that there are other complications, such as uncertain parentage of one or more of the "cousins" or, for that matter, of their parents. If that is the case, the degree of matching (and possibly also of X chromosome matches) among those cousins should be checked carefully.

Regarding the matches that you have, if the shared cM values are within the range as suggested (even a first cousin once removed would be in the range of 215-650 cM, according to the popular and very useful "DNA Detectices Autosomal Statistics Chart", easily found on the internet through a Google search), there is no way these matches can be "false". The problem is to combine the genetic evidence with the paper trail evidence and then to sketch out ALL of the possible family trees that could yield the observed result. Keep digging!

Thank you very much for your input!

Yes, I based my estimates of relationships solely on what Ancestry.com told me the relationships could be. They said that instead of 1st cousin, my newfound relative could be any other relationship with the same "distance" (my words as I understood it), for example, I could be my newfound relatives' great-great-aunt, but those relationships are even less likely than cousin because of our ages.

I didn't know about GEDmatch and will look into it. With my current level of know-how I wouldn't be able to do any kind of segment comparison, but I suppose I could acquire the knowledge by doing research. I may not match anyone there though.

My match with my three newfound "cousins" (or whatever they are) range from between 343 centimorgans shared across 18 segments to 901 centimorgans shared across 43 segments. Ancestry seems to think that those are ranges that are very likely to indicate a relationship as close as a 1st or 2nd cousin relationship. And since you say the matches with those values can't be "false", I will persevere in my "digging". Maybe my test results from 23andme will give me more information. I already tried to sketch out possible family trees that would yield the observed result, but I came up blank.

[Emona]

This table (based on real relationships and real numbers) may also be helpful:
https://dnapainter.com/tools/sharedcm

and another version, with probabilities:
https://dnapainter.com/tools/sharedcmv4

Just in case: In some endogamous populations, people seem to be more closely related than they really are. The same genes are circulating in the community and they all share them; examples: a small remote island, or Ashkenazi.

Do I need to mention that I am curious how this story is going to unfold? This said, always remember that the well-being of your family members is more important than curiosity of anyone here on the forum.

Good luck.

[ltd-jean-pull]

Is there anything else I should be doing to solve the mystery? I would be grateful for any ideas.

Have you also uploaded your DNA here to FamilyTreeDNA?

[Blythe]

Have you also uploaded your DNA here to FamilyTreeDNA?

I would have, but I couldn't find an option for uploading the data. I will look again, since your question seems to indicate that it is, indeed, possible.

[Blythe]

This table (based on real relationships and real numbers) may also be helpful:
https://dnapainter.com/tools/sharedcm

and another version, with probabilities:
https://dnapainter.com/tools/sharedcmv4

Just in case: In some endogamous populations, people seem to be more closely related than they really are. The same genes are circulating in the community and they all share them; examples: a small remote island, or Ashkenazi.

Do I need to mention that I am curious how this story is going to unfold? This said, always remember that the well-being of your family members is more important than curiosity of anyone here on the forum.

Good luck.

Thank you so much for pointing out those tools! How fascinating! You gave me a great starting point to work on!

By the way, there don't seem to be any endogamous populations in my background or in the family tree of my "new" relatives, so I think there is no danger of false positives from that direction.

[Emona]

Here are the instructions how to upload to ftdna:

https://www.familytreedna.com/learn/...nsfer-program/

Just a thought.
what if:

-you are related to the 2 siblings through their father?
(you do not match their two maternal cousins)

-your relation to their remaining maternal cousin is through the person's father?
(and the match is here, creating this unexpected situation)

Keep trying and the pieces of your puzzle will eventually fall into their place.

[Blythe]

Here are the instructions how to upload to ftdna:

https://www.familytreedna.com/learn/...nsfer-program/

Just a thought.
what if:

-you are related to the 2 siblings through their father?
(you do not match their two maternal cousins)

-your relation to their remaining maternal cousin is through the person's father?
(and the match is here, creating this unexpected situation)

Keep trying and the pieces of your puzzle will eventually fall into their place.

Thank you for sending me the link! It appears that I or a member of my family has to be tested at FTDNA first before I am able to upload data from another provider's test. Maybe I will do that.

The idea that I may be related to the two siblings through their father is one that I will have to pursue. It was the less convenient road to investigate at first because their father had no known siblings. It is a real possibility, however, and the links you sent me detailing possible family relationships will be very helpful. I can't thank you enough!

[betadams]

Blythe

Is it Ancestry DNA you want to transfer to FTDNA ?

If so, you will need to use this link

http://www.mapmy23.com/tools/ancestry_ftdna_fix.php

[Jim Barrett]

Is it Ancestry DNA you want to transfer to FTDNA ?

Only if it is new Ancestry, within the last one or two years.

[Germanica]

My match with my three newfound "cousins" (or whatever they are) range from between 343 centimorgans shared across 18 segments to 901 centimorgans shared across 43 segments. Ancestry seems to think that those are ranges that are very likely to indicate a relationship as close as a 1st or 2nd cousin relationship. And since you say the matches with those values can't be "false", I will persevere in my "digging". Maybe my test results from 23andme will give me more information. I already tried to sketch out possible family trees that would yield the observed result, but I came up blank.

Are those the numbers for the brother and sister? 343 cM could easily be 2nd cousins. Even so, 2nd cousins should match each other (about 10% of 3rd cousins won't share enough DNA to match and 50% of 4th cousins won't), so you should still match all of them and not some of them. Plus, 901 cM is too high to be 2nd cousins.

[loobster]

It appears that I or a member of my family has to be tested at FTDNA first before I am able to upload data from another provider's test.

No, complete misunderstanding.
I would love everyone to test at FTDNA, but to transfer/upload data/results from another provider to FTDNA absolutely does not require anyone to test at FTDNA first (or at all).

[Dabney Carr]

No, complete misunderstanding.
I would love everyone to test at FTDNA, but to transfer/upload data/results from another provider to FTDNA absolutely does not require anyone to test at FTDNA first (or at all).

You just need to create an account and log in first before you try to transfer the DNA file.

[loobster]

If you are transferring DNA results for yourself, and you do not already have an account / have not already tested at FTDNA - OR you are transferring for somebody else who does not have an account/has not already tested at FTDNA -
Follow the directions you get if you click on "New Customers" in the link given before -- https://www.familytreedna.com/learn/...nsfer-program/ -- do not go trying to create an account and log in some other way - just follow the directions.