HARRIS-DNA-L Archives
Archiver > HARRIS-DNA > 2005-05 > 1115275324
From: "Gregg Bonner" <>
Subject: Re: [HARRIS-DNA] Group 10 versus Group 12
Date: Thu, 5 May 2005 14:45:16 -0400
References: <19a.3335b85a.2faa75d5@aol.com> <002101c55164$b016f080$3801a8c0@bonner> <002b01c5511c$2cc6b000$445cd618@simon>
Hi Folks,
I am going to address three recent messages to the list:
1) Darlene wrote:
> Greg and James, I asked a simular question of Family tree DNA and was told
> that you have to have enough sampling to know the base DNA. What seems to
be
> two off could be where one set mutated one way and the other set the other
> way.
The probability of this happening depends on how far back you think the
common ancestor is. If you think the common ancestor is their grandfather,
then no, no way. If the common ancestor is their 12th great grandfather,
then maybe. But more importantly, another sample reveals the likelihood of
this. If you test a cousin of "A" who is less remote than "B", and he has a
mutation that is not toward this hypothetical "middle ground" (i.e., midway
between "A" and "B"), then the hypothesis just went from the frying pan to
the fire.
2) Charles Leon Harris wrote:
> With experience we'll get a better idea of how far back we have to go to
find
> the most recent ancestor with a five-step difference in results.
But we need no more experience than we already have to calculate the
probabilities of the hypothesized events. The mutation rates are known.
Everything else can be derived. There is no mystery.
3) James wrote:
> OK, I give up. You are throwing too much unsubstantiated science at us.
I am not sure if this was meant to be comedic or not. But supposing it is
not, I have to say that far from be unsubstantiated, there is nothing about
it but substance. These mutation events are discrete events that happen at a
time certain. And just like statistics of flipping a coin or rolling dice,
these follow the same pattern. The probability distribution follows the
Poisson Curve. If you have Microsoft ExCel, then you can make a simple
spreadsheet that calculates the expectations (number of mutations total)
with the Poisson command. It will calculate for you the probability of zero
mutations, one mutation, two mutations, etc.
In order to do this, you need to know:
The number of markers tested: KNOWN
The number of generations from common ancestor to DNA participants: This is
part of your hypothesis - it is assignable.
The mutation frequency of the markers surveyed: KNOWN
James continued with:
> There is no database of DNA testing for the last 1000 years, so that you
are creating
> a database and criteria from a small sampling of current DNA tests, from
which you
> and others interpret and project a database, where none really exists.
There is no database of DNA testing for the last 1000 years, this is true.
However, this does not mean that we are unable to assign a DNA type to a
common ancestor of two lines, even if he lived several hundred years ago.
Frankly, this is exactly why the technique DOES work. All you need to do is
have a perfect match from two different sons of the original patriarch. If
you can't do that, then you are going to need more samples, but it can still
be done. Remember, there are more complicated ways to explain the data
(i.e., use more mutations than are minimally necessary to satisfy the data),
but mutations are known to be very rare, so you can't hypothesize a lot of
excess mutations, or else the data (probabilities) will necessarily not fit
it due to that very basis.
James continues with:
>I must admit that after this debate, I have very little faith in DNA
testing and analysis
> because the entire structure is, in my opinion, based on a flawed premise
and I don't
> think it is of much help to anyone as long as the current criteria for
analysis is being used.
In my case, I matched exactly with my 3rd cousin, 2x removed. I was not sure
if my cousin and I were from the Virginia line or the North Carolina line.
We tested bona fide descendants of both lines. I matched the Virginia line
perfectly, and the North Carolina line was completely different. I concluded
that I wasn't from the North Carolina line. Where is the flawed premise in
this?
Jim concludes with:
>Lord or serf, they are only important if they are my true ancestors.
That's great. But that doesn't stop any other person from claiming your
ancestors even when they aren't theirs. And then when they get their DNA
done, they are only a 23/25 match with you, and it makes you wonder why.
Best,
Gregg Bonner
Ann Arbor, Michigan, USA
This thread: