Archiver > GENEALOGY-DNA > 2009-12 > 1260930211

From: "Alister John Marsh" <>
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent- back mutations
Date: Wed, 16 Dec 2009 15:23:31 +1300
References: <><><3C7226E21D8B42C3B704E165C522F268@john><004a01ca7d97$6c230ad0$6400a8c0@Ken1>
In-Reply-To: <004a01ca7d97$6c230ad0$6400a8c0@Ken1>


You say that variance takes account of back mutations and multi step
mutations. I believe I have a "rough" understanding of this. You are right,
if correctly allowed for, back mutations as I experienced should not on
average distort age estimates by variance means, and would be reflected in
confidence intervals.

Perhaps in shorter time spans, like 750 years, when back mutations would not
be expected to be common, an unexpected cluster of back mutations could
muddy the water.

The back mutations in my example would I think confuse Tip, in the time
spans under 24 generations.


-----Original Message-----
[mailto:] On Behalf Of Ken Nordtvedt
Sent: Wednesday, December 16, 2009 4:01 AM
Subject: Re: [DNA] R-U152 and R-L21 on the European Continent- back

----- Original Message -----
From: "Alister John Marsh" <>
I now know that in addition to the obvious one step
> difference, my own CDYb has mutated to =39, then mutated again back to
> =38.
> This means that although only 1 step mutation is visible at first glance,
> it
> is apparent that there have been at least 3 mutations difference between
> us
> on that marker in the past 750 years, and for all we know there could be 5
> or 7 mutation steps difference between us on that marker. If 5 or 7
> mutation steps are being concealed by the obvious 1 step difference, the
> TMRCA may be much further back than we or Tip think.

If you are using GD for your trmrca estimates then corrections need to be
made for the probabilities of back mutations and multiple steps of mutation
in the same direction. But variance takes these things into account
automatically. That has been emphasized by several people on this forum
over the years dozens of times.

You are apparently talking about Tip. How they adjust their estimator for
back and multiple steps of mutation would be revealed from the interworkings

of the Tip algorithm. I believe Tip is based on GD of some sort.

The scenarios concerning back mutations you mention are "could have" types
of things; they happen with some probability. In some sense they are the
same "could have" types of realities associated with total number of markers

which mutated? If we "expect" on average n such mutations, it could have
been n+1 or n+2... or n-1 or n-2 mutations, each actual outcome happening
with some probability. The main consequence of all those "could haves" is
the statistical confidence interval about the most likely or expected
outcome --- i.e. the reality that the outcomes form a distribution.

To unsubscribe from the list, please send an email to
with the word 'unsubscribe' without the
quotes in the subject and the body of the message

This thread: