MRCA - probabilities

[jhgren]

Hello all,

Last year I had my dad submit a Family Finder test puls a 12 marker Y-DNA test.

I may have found the connection to one of his matches through FF. If the paper trail is correct, then my dad and him are 7th cousins once removed. Now FTDNA suggests that they are 4th cousins with a range of 3rd to 5th cousins with these values:
Longest block = 15,75 cM
Shared cM = 41,04 cM

Also, a distant relative of of my dad's has also tested her autosomal DNA and we have compared through Gedmatch, and the result is this:
Longest block = 31,7 cM
Shared cM > 3 cM = 40,2 cM
Estimation of MRCA: 4,2 generations
Through traditional genealogy we know that my dad and this relative are "double" 6th cousins once removed (she has two lines to their MRCA because of a marriage between two 1st cousins).

So, now I am wondering... The estimations don't seem to fit with what our research show. I realize that estimations are based on statistics and that there can be large individual differences. But what are the chances of results as the ones above? Anything else that I should take into account?

Thanks for any input,

Jacob

[thetick]

Yes the ranges are vast estimates. Also any type of inbreeding which was very common in up until recently may make some very distant relatives appear to be much closer due to multiple line contributing to your DNA.

Also the majority of your ancestors that contributed to your DNA quickly vanish see http://www.genetic-inference.co.uk/b...-share-our-dna

[Frederator]

Yes the ranges are vast estimates. Also any type of inbreeding which was very common in up until recently may make some very distant relatives appear to be much closer due to multiple line contributing to your DNA.

Also the majority of your ancestors that contributed to your DNA quickly vanish see http://www.genetic-inference.co.uk/b...-share-our-dna

Concurr. That sounds like a reasonable interpretation.

Though I have to admit that I'm a little surprised about how often 1st cousin marriages result in extraordinarily large single blocks. I know there's a large element of random chance here, but we're talking about something like a total of 7K Cm total, aren't we?

The odds of recombination resulting in a single block ~30 Cm should be like 1 in 108,889--if you'd expect a block of about 15Cm and adjust for the possibility that the database's matching protocol can't distinguish between materna/paternal DNA (i.e, 7,000 / 15=467; 467^2/2=108,889; the odds that two ancestral 15 Cm segments realign contiguously and consistently for 2 separate individuals).

Maybe that's not the right way to look at it. Even though I've heard of this happening plenty and don't doubt for a second that the double-6th cousin attribution is correct, I just am shocked by how often this results in unusually large single blocks.

[Colin]

How would second and third cousin marriages affect the dna?

[jhgren]

Thanks for your very reasonable input!

Does anyone know of confidence intervals for the predicted match ranges?

Along with what was written above I have been wondering how it may affect the predictions that my dad's ancestors were on his mother's side from a specific island in Denmark and on his father's side from a not so large area of Sweden...?
I know of only one case of intermarriage on his mother's side (but so far back that noone would have known). But very far back there could have been a lot more. What I mean is that the "ancient gene pool" for his ancestors must have been less diverse than if his ancestors had been from many different countries/backgrounds. Would this lead to more matching segments based on chance or residue? To what degree? Would estimates be systematically too recent? Is there any research on this?

Can someone add?

Thanks again,

Jacob

[thetick]

@Colin

The more distant cousin marriages the higher chance more DNA will be shared between two people. This adds even more error to our already very large cousin prediction range.

@jhgren
I have never read anything on confidence intervals since I think any calculation would be so vast they would be meaningless for beyound 3rd cousins. This is the nature of random DNA segments being passed every generation. There is nothing to be done to improve this resolution. Note relatives 3rd cousin and closer can be very reliably predicted. I think Ann Turner was working on a paper concerning cousin matching prediction. Hopefully it will soon be published.

This is really a tangent, but one analysis we can do is run a BGA on a shared segment. This works very well if the shared segment is from a genetically distant group from the other parents DNA. This is because the BGA of a segment contains two strains one from each parent. If you can phase your DNA to your maternal and paternal lines then the BGA can be even more effective.

The above has proved very effective in determining my Jewish matches since all my other known ancestry is Eastern and Western European. This would not work well at all if I had Italian ancestry also since the Italian and Jewish are genetically similar.

[jhgren]

Thank you for your reply! I have been travelling, therefore the late reply.

It sounds like a very interesting article by Ann Turner. I look forward to it.

I'm still rather new to this, so what does BGA mean? It doesn't sound like a realistic option though. My father's father was from Southern Sweden and my father's mother was from Southern Denmark, so my father is entirely Southern Scandinavian - I wouldn't expect much genetic diffference between the two sides of his family. (My father was the one tested.)

[cjm95]

BGA is Biogeographical Analysis which is basically the Population Finder results you get with the Family Finder test.

http://www.familytreedna.com/faq/search.aspx?k=bga&c=&o=0&ps=25

[jhgren]

Ahh, I see. Thanks!

Well, for my dad the result is:
Europe Orcadian 100,00%