GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-01 > 1170120173


From: (John Chandler)
Subject: Re: [DNA] Worth getting yet more extras?
Date: Mon, 29 Jan 2007 20:22:53 -0500 (EST)
References: <293619.90409.qm@web31504.mail.mud.yahoo.com>
In-Reply-To: <293619.90409.qm@web31504.mail.mud.yahoo.com> (message fromJonathan Day on Mon, 29 Jan 2007 12:38:09 -0800 (PST))


Jonathan wrote:
> Numbers are fun, but it pays to be cautious -
> probabilities aren't additive and mutations
> accumulate.
>
> This means that for every mutation, there is an 11/12
> probability that it will result in the person
> diverging from my initial 12 markers, no matter what
> their markers are,

No, that's not what it means. If you want a statement that is true,
no matter what their markers are, you're out of luck.

> will take them above the 12 and outside of the tree
> under consideration.

Sorry, but there is no "tree" here. There is a spectrum of alleles
for DYS392, as for any other STR marker, and it is possible to jump
from any state in that spectrum to any other state in a single
generation. In other words, you cannot know that the "12" state you
have now was preceded by an "11". It could have come from a "10"
instead, or a "13". Of course, the odds do strongly favor an "11" as
precursor, but you seem to be looking for statements that are
absolutely true "no matter what", and you simply can't get such.

> (This is per iteration, and since you can't just
> multiply up the probabilities, you'll need to do some
> definite integrals here over the 100 generations that
> my branch of DYS392=12 seems to have existed.

Since you have tested only yourself, you have no evidence whatsoever
about how long ago your line came to have "12". You don't even know
whether the "12" is a fresh mutation in you. As I pointed out in my
previous message, there are something like 20,000 hgI men in the world
today who have a fresh mutation to DYS392=12, and you could be one of
them. If you recruit and test a 2nd cousin, you can discover whether
the mutation happened before or after your ggf, but it's going to be
exceedingly difficult to push that sort of test more than ten
generations back, since you seem to be starting from scratch.

> Remember also, I'm only considering I1a haplotypes -
> which only account for 35% of Is in the areas they max
> out in - and am discounting all I1as who carry a
> mutation I can positively identify as predating the
> DYS392=12 branch I am on

Sorry, but there is no "tree" here for other STR markers, either,
except for the *extremely* slow mutators or within a very young
subpopulation. For that reason, there are *very* few I1a's who carry
alleles that can be "positively" identified and dated as specific
mutations, even in the loosest sense.

> But for each of those people carrying that mutation,
> there was a p(0.00005) that they would suffer another
> mutation that would result in a value of 13, and some
> other probability that the mutation would have
> reverted to an 11.

No. The mutation rate for DYS392 is 0.0005 in total, and there is no
evidence that the proportion of up mutations is as low as 1/10. For
the sake of simplicity, in the absence of evidence to the contrary, it
would pay to assume that the proportion is 1/2.

> (This assumes that mutation is just as stable as a
> value of 11, which might not be the case. If it can be
> shown that there are sizable gaps in the genetic
> record that should not exist, then that would be
> grounds for believing that the mutation is NOT
> stable.)

That is called "grasping at straws."

John Chandler

This thread: