Sunday, July 10, 2016

FTDNA's New Phasing Tool--Ashkenazi Edition

FamilyTreeDNA has a new tool that attributes genetic matches to your maternal or paternal sides.  Roberta Estes has a comprehensive overview here where she also explains how to link your parents' results to yours to generate this sort of breakdown.  Roberta's example shows her matches nicely broken out between her father's side (700 matches) and her mother's side (487 matches), with only 3 matches inherited from both parents.  She also only has a total of 1604 matches--while I have 7053.  Because of endogamy, Ashkenazic Jews' DNA isn't as straightforward to deal with as the general population.
My FamilyFinder Results

I've tested myself and both of my parents.  I have my parents' direct ancestors documented back at least 200 years on all branches (and further back on many branches).  Documentation shows that my father's family and my mother's family lived relatively distantly from one another during that entire time, and I have no indication of them being related on paper.

But even with that, I have 576 individuals who match me on both my paternal and maternal sides with long enough segments to be included in the "both" tab.  (Note that FTDNA seems to be erring conservatively, so some people who FTDNA predicts are my distant cousins may not appear on the "maternal," "paternal," or "both" tabs.

In fact, most of my closest matches on the "All" tab (other than very close relatives whose relationships I already know and who tested because I asked them to test) appear on the "Both" tab.  While I don't have any really huge segments in common with them, I do have lots of segments in common with them--some from each of my parents.  This is why sometimes looking at longest segment can be more beneficial in endogamous populations than looking at total shared DNA.

For those who haven't been able to test parents and who are from endogamous populations, realize that this is why your potential "2nd-3rd cousin" may actually be much further but actually related through both of your parents, and maybe even all four of your grandparents.  Take a look at my results.

The second-closest match which FTDNA says is connected on both my parents' sides is a woman I'll call "S."  FamilyTreeDNA predicts S to be my 2nd-3rd cousins, with our 137 shared cM and a longest segment of 10.  2nd-3rd cousins sounds promising, right?  But not so fast....
My match with S.
I looked for S on each of my parents' lists of matches.
My father's match to S
My father's connection to S is predicted to be much more distant--5th to remote cousin.  They share 104cM with a largest segment of 9cM

So S must be very closely related to my mother, right?
My mother's match to S

Nope--S is predicted to be my mother's 5th-remote cousin as well.  They share 110cM with a largest segment of 10cM.

So how can this be?  Looking more closely at shared segments explains this phenomenon:
My DNA overlap with S (showing shared segments 3cM & longer)
I have a lot of shared segments with S, driving that relatively close prediction.  But why did my parents have a much further predicted relationship?
My Mother's DNA overlap with S (showing shared segments 3cM & longer)
My mother obviously has significantly less DNA in common with S than I do.  And how about my father?
My Father's DNA overlap with S (showing shared segments 3cM & longer)
My father also doesn't share much DNA with S.  But it seems like I inherited almost all of the DNA that S had in common with both of my parents, making the predicted relationship much closer than obviously could be the case.

Now notice that I said that S was my second-closest match predicted to be both maternal and paternal.  So why didn't I lead off with the closest one?  Well, my closest match predicted to be both maternal and paternal is E.
My Match with E
E is also predicted to be my 2nd-3rd cousin, based on our 141 shared cM, with a longest shared segment of 15cM.  And how does E match my parents?

She doesn't.  E shows up as a match to neither of my parents, not even distantly.  How could this be?
My DNA overlap with E (showing shared segments 3cM & longer)
While I have a bunch of shared segments with E, the only reasonably large one is a 15.5cM segment on chromosome 4.  None of my other known relatives who have tested match me on this segment.  My assumption is that part of this shared segment came to me from my mother and part from my father, meaning each of them have a relatively small shared segment on chromosome 4, and FTDNA decided the DNA that each parent shared with E just wasn't enough (particularly with its endogamy adjustment) to qualify E as matching any of them.

So this is another proposed 2nd-3rd cousin who's only related much (much) more distantly.

So be careful with assumptions, particularly when dealing with endogamous populations--sometimes a closely predicted match just isn't so close after all.  Had I not tested both of my parents, I may have continued to chase down these matches, trying to find the connections.

And don't get too excited just yet about matches that are only associated with one of your parents--they may be getting DNA through both of their parents.

Isn't endogamy fun?

Note:  I'm on Twitter.  Feel free to follow me (@larasgenealogy).

Want to get future blog posts emailed to you automatically?
Enter your email address:


  1. Sigh---this all makes a lot of sense to me. I haven't yet tested my father (he's a bit difficult about such things), only my mother. But now your explanation has made me think about whether it's worth the aggravation and cost of getting my father tested just so I can also parse out these supposed 2d to 3d cousins whom, I know, are nothing close to that. Thanks, Lara!

  2. The key in accurately determining if it is a close or distant "predicted" relationship, like a 2nd - 3rd cousin match would come from the segment size. In general, endogamous populations will have multiple segments. It does not help that FTDNA will show very tiny segments. Many of my matches can have easily more than 50 segments. My first couple of pages of matches can have 100+ segments, and the total shared are usually above 100cM, yet largest segment (longest block at FTDNA) are as small as 8cM. That is a big sign that they are distant, so while my brothers and I may rarely show slightly more shared total with a few matches than our mother, it should be obvious by the longest block size that these matches are much more distant.

  3. Your results look much like my husband's: 1734 Paternal matches; 1869 Maternal matches and 411 Both.

    Do you think there a minimum longest block that we should be looking at?

    Unfortunately I cannot go back as many generations on all of my husband's lines as you can.

    1. I heard CeCe Moore mention 20cM as the minimum she wants to see when working an Ashkenazic Jewish case.

    2. That makes sense based on what I'm seeing in my husband's results. Thanks!

  4. Thinking about getting started in this direction and your detailed explanations are very helpful! Thanks!

  5. Lara - As much as 20cM before considering a match? About how close a relative would that be? Where would someone who has several blocks in the 10-15 cM level be? I have one close friend who I'm "related" to. She has two blocks in the 8-9 cM and two more in the 13-15 cM range. We're getting a kick saying we're related, but I'd like to have an idea how far back. Thanks.

    1. I actually have a partially-written post looking at largest segment size shared with known relatives. It may help. I'm in Ukraine now but will post it when I'm back.