GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2012-03 > 1332450684


From: Jim Bartlett <>
Subject: Re: [DNA] Phasing 4th Cousins
Date: Thu, 22 Mar 2012 16:11:24 -0500 (CDT)


On 03/22/12, John Chandler<> wrote:

Jim wrote:
> The deal is that every match has at least one large segment - the
> companies believe most are IBD - and the number of cousins increases
> faster than the probability of one matching decreases.
You have pasted together two statements that don't mesh. Yes, the
matches that are reported to you are graced by large half-identical
segments. However, the probability of sharing large IBD segments
decreases faster than you mught expect, because of the combination of
dilution and fragmentation, and the number of cousins increases slower
than you might expect, because of pedigree collapse.

My response: Sorry for the two statements. It appears you agree with the
first one: each atMatch has at least one IBD segment that we share.
Restating the second statement: yes the probability of sharing large IBD
segments decreases quite quickly - I agree. It looks like you agree that
the number of cousins does increase with each generation going back in time.
The question remains - is this combination of decrease and increase a net
decrease or a net increase. From my experience, it appears that although the
sharing decreases with every generation, in fact we see more atMatches as we
go back in time. We have more 3rd cousin matches than 2nd; we have more 4th
cousin matches than 3rd; we have more 5th cousin matches than 4th; we have
many Distant cousins. In my case this increase continues out to about 7th
or 8th cousins. I attribute this to having a robust documented Ancestry,
based largely on Colonial Virginia. Many folks have trouble finding matches
because they don't know their ancestors back 8-10 generations. Now you
might argue that at 8-10 generations there shouldn't be any IBD segments
left, and you may be correct. However, three of us have now determined that
each has at least a 13cM segment on Chr19 from 38-48m with each other (A to
B, B to C, and C to A), and we descend from 3 different children of John
HIGGINBOTHAM b c1695 and wife - they are my 7G-grandfparents, making the 3
of us 8th cousins. Maybe it's a fluke, and we 3 descend from some other
closer, hidden ancestor we haven't found yet. I now emailing my 23&me
atMatches with that same segment - maybe one of them will also match John H;
maybe we'll match elsewhere; etc. - we'll see.

Simultaneously, there is the growing problem of composite segments
that are identical by shared ethnicity. This is where the issue of
redundancy enters. If you want a reliable "road sign", you need to
verify the same segment in *three* distantly related descendants of
the same pair of ancestors.

My response: I don't completely understand the phrase "composite segments
that are identical by shared ethnicity". The conversation is inching over my
head... If you mean I can't rely on just me and another person matching to
declare a quilt piece, I agree. I treat those results as tentative (and
sometimes my atMatch and I agree on two different Common Ancestor couples);
and continue looking for others who share that same segment to see if they
have the same ancestors. See HIGGINBOTHAM above. In fact I'm convinced that
it's even possible to have one segment eventually be found to go back to two
of my ancestors - but one would have to be on paternal side and one on
maternal.
> I'm not interested in completing the quilts, I'm
> interested in identifying road signs that point to specific
> ancestors.
Suppose someone comes along who may descend from a specific pair
of ancestors, and you want to know whether it's worth the expense
of autosomal testing for this person to see if that link can be
substantiated. The likelihood of having a useful "road sign" is
proportional to the degree of completion of the relevant quilts.

My response: What you say is true, and I agree. And in this case the
statistics for a successful atMatch are tiny when a new person is involved.
I would not approach it that way. I've already got 1,200 atMatches, and
only worked out Common Ancestors with 62 of them. I'm looking at the
inventory I have - sorting the spreadsheet by Chr and start location for all
large segments. I have already identified over 50 groups of at least two
others who share the same large segment with me. I can weed that group out
pretty quickly if they don't share that segment with each other, or b) they
can't determine a match on a Common Ancestor. This is what I mean by a
patchwork quilt - it's not specific ancestors that I am targeting, it's
accepting whatever matches I can determine to be quilt pieces. As more
atMatches come in, I can look at them and see if they share some of the
quilt piece road signs I already have; or whether their large (IBD) segment
forms another 3-way with two of us already in my spread sheet.

Another part of this that excites me, is the opportunity to help the
adoptees. If one of my adoptee atMatches is in a 3-way segment match, or
even better a 4-way, with me, the Common Ancestors the rest of us find would
be a very important clue. If I were an adoptee, I'd be sorting through my
own spreadsheet and asking the 3-ways if they matched and/or had found a
Common Ancestor. By the same process, I may be able to find some maiden
names and other ancestors just beyond my brick walls. Their names and BMD
info won't be handed to me on a platter, but the road signs will be
invaluable help.

I hope you can see this from the perspective of using this tool to find
cousins, confirm (or deny) ancestry, and find new ancestry.

This thread: