Archiver > GENEALOGY-DNA > 2008-06 > 1212941159

From: "Victor Villarreal" <>
Subject: Re: [DNA] mushing
Date: Sun, 8 Jun 2008 11:05:59 -0500

> Back mutations are COMPLETELY incorporated into the variance measure of
> growth of variation in a population which has descended from a common
> ancestor. Those who use steps of genetic distance (GD) as measure of how
> far back their common ancestor probabilistically existed must eventually
> make corrections to the simple models for back mutations, especially if they
> push GD applications to cases of large sizes.
> My rough estimate of the fraction of grandfather/father/grandson trios who
> will exactly match at the 2 generation level after having mutation at one
> generation level on N marker haplotype because of back mutation is N m^2 / 2
> Put N = 40 and m = 1/316, and that fraction is 2 / 10,000 Such a study will
> require lots of field data. Father/son rate studies can be analyzed to see
> if "up" rates differ from "down" rates. They probably do in general, but if
> self variance is used, difference of up rates from down rates does not
> perturb self variance use as age estimate.
> Ken

It is precisely that "incorporation" of back-mutations into the
variance measure where I think the age estimates could slip. I
understand that there is no other way to do it and that's 0k if, and
this is an important if, it wasn't done in a wholesale fashion for
every used marker. We can't just assume that the up rates are the
same as the down rates, nor that the differential between these two
applies equally at every loci.


This thread: