GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-06 > 1182459578


From: (John Chandler)
Subject: Re: [DNA] TMRCA Corrections
Date: Thu, 21 Jun 2007 16:59:38 -0400 (EDT)
References: <004101c7b33f$d2296bf0$6400a8c0@Ken1><REME20070620221136@alum.mit.edu><012b01c7b3ba$b580baf0$6400a8c0@Ken1><001101c7b41e$fd023f70$27129a8e@PAKINCAID><00f601c7b421$a045fc10$6400a8c0@Ken1><001901c7b42e$82732cf0$11139a8e@PAKINCAID>
In-Reply-To: <001901c7b42e$82732cf0$11139a8e@PAKINCAID> (7kincaids@primus.ca)


Peter wrote:
> I am just looking at possible application of this
> in relation to surname projects.

The problem is to avoid circular reasoning. If you estimate the TMRCA
based on the assumption that the two men are both descendants of a
common ancestor 600 years ago, you will necessarily get an estimate
that is no more than 600 years ago. If you take away that assumption,
then the clade of context is not the surname group, but the haplogroup.

> You noted that some outliers
> could have their TMRCA inflated. I'm was trying
> to get a sense of what a point of no return might
> be in terms of who should be in a cluster.

That's a matter of outliers with respect to the clade of context.
Unless the family cluster is very close to the modal haplotype of
the clade, then the outliers with respect to the cluster are not
necessarily outliers with respect to the clade.

Also, note that Ken has not yet done the calculations for any
comparison except identical test subjects. The corrections for
unequal testees require the weighted summing of increasingly
many different terms spanning a variety of distances from the
ancestral haplotype.

Another complication: Ken's model doesn't yet include mutations
of more than one step, but these can make an important difference
in the calculations.

John Chandler


This thread: