HARRIS-DNA-L Archives

Archiver > HARRIS-DNA > 2005-05 > 1115173301


From:
Subject: Re: [HARRIS-DNA] Group 10 versus Group 12
Date: Tue, 3 May 2005 22:21:41 EDT



Hi Gregg,

I'm sure we are all amateurs in regard to DNA analysis and I would have no
basis to disagree with anything in your current analysis, but in some of your
earlier correspondence you said that even in perfect 25/25 matches, there has
been at least one or two mutations backwards and forwards, and also that 2
stage mutations can be a single event, which has been proved recently in the
DNA tests of a father and son.

So, what are we to think, as amateurs, with so many variables and abstract
possibilities and probabilities, that even a perfect match can be terribly
flawed to the extent that unknown mutations have occurred, of which, we do not
know how many, or whether they were one or two stage mutations, as a single
event or as double events. I have to ask you, as you seem to be an expert in DNA
analysis: With all of the unknowns, variables, possibilities, probabilities
and assigned values to the mutations and results, can anyone say with any
certainty, that any result can be compared sensibly with any other result, if
seemingly perfect matches include an unknown number of mutations, impossible to
determine, and if all this is true, then is genealogical DNA study really
helpful to researchers?

We have been encouraged to spend large sums of money to find DNA matches,
and yet the comparitive analysis seems to be much in doubt and the results are
questionable to the point that one wonders if any supposed match can be
trusted to reveal a true family relationship, that is, according to the experts,
who seem to generally disagree about what really constitutes a match beyond our
own generation.

Sincerely,
James E. Hargraves


In a message dated 5/3/2005 6:17:42 PM Pacific Standard Time,
writes:

Hi Folks,

I think it is worth pointing out technical features of DNA mutations of this
type. Imagine some arbitrary marker having allele value = 10. Then, in a
later generation, it mutates to 9. Then, somewhat later, it mutates again.
What is the probability that it will mutate to 8, as opposed to mutating
back to 10? The point of this is that even though a value of 8 versus 10 is
only a genetic distance of 2, it is actually less probable. Generally
speaking, when people assign "mismatch penalty" to these things, they get
the square of the difference in penalty. This is to say, a 2-increment
mismatch gets a "4" penalty; the same as 4 separate and independent
mutations.

The other thing is that just because a marker is fast-mutating, that does
not mean it gets lesser penalty for mismatch. The increased mutational rate
is exactly balanced by the increased mutation observations.

So, just to address the issue, let's take 8895, 28636, and 8871. Versus
group 10 modal through 12 markers, 8871 has:
1-increment mutation at marker 3 (14 versus 15; penalty 1)
2-increment mutation at marker 6 (16 versus 14; penalty 4)
2-increment mutation at marker 12 (28 versus 30; penalty 4)

So, in sum, 8871 is not even remotely close to Group 10. Kit 25394 fares not
much better, but I would at least give it "possible".

If anything, I would split group 12, not combine other groups or members
thereof.


Best,

Gregg Bonner
Ann Arbor, Michigan, USA






This thread: