GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2001-08 > 0998582436


From:
Subject: Re: [DNA] Common Ancestor
Date: Thu, 23 Aug 2001 12:00:36 EDT


In a message dated 08/22/01 2:29:17 PM Pacific Daylight Time,
writes:


> I am confused by the previous 2 posts. I thought that if I found a living
> representative in a straight line of female descendants from my Mary
> (Cockerham) Downey AND another living representative in a straight line of
> female descendants from one of her presumed sisters, Keziah (Cockerham)
> McBride, and their mitochondrial DNA (mtDNA) matched, it may be very strong
> support (but not proof) that Mary was Cockerham. These supposed Cockerham
> sisters were about 7 generations back and born ca 1760.
>

I think we need the services of a professional statistician here, but I'll
lay out the way I visualize things anyway.

I agree that a match would support your hypothesis, but the real question is
how strong is that support. I think most people would agree that a mismatch
(say more than one difference in polymorphisms) would rule out your
hypothesis.

For the sake of argument, let's assume that we'll accept relationships more
distant than sisters, say cousins of a small degree who descend from a common
maternal ancestor within a genealogical time frame. The question then boils
down to whether a match has any significance at all, when the common ancestor
COULD have lived thousands of years ago, as David Andrews has pointed out.

Mutation rates in mtDNA are still a matter of study, but let's use the figure
of 3 mutations in 705 transmission events from the Iceland study, roughly 4
per 1000. (This figure covers both hypervariable regions of mtDNA, about 1000
bases, compared to the 400 bases in Hypervariable Region 1 that are
commercially available.) This means that it's quite possible to find people
living today who share the mtDNA haplotype of the Ice Man from 5,000 years
ago. In fact, Sykes mentions one in his book and we've seen one on our own
mailing list.

By sheer coincidence, this is about the same mutation rate observed in one Y
STR marker. Since Y chromosome testing uses 10-12 or more markers, you have
much greater resolution for surname studies, where we are seeing some
single-step mutations that can be traced to one branch. This kind of
resolution probably won't be posssible with mtDNA, even if the whole 16,569
bases are sequenced.

But all is not lost -- mutations in mtDNA can occur in any combination of
positions, so there are potentially a very large number of haplotypes. It
seems that new specific haplotypes are found every time a new study is done,
although most of them can be assigned to general categories (haplogroups).

In your case, you are talking about two women who lived in a specific
location in a specific time frame. It seems to me that it's appropriate to
switch the frame of reference and ask "how likely is it that these two women
would match because two descendants of a common ancestor who lived thousands
of years ago just happened to settle in the same vicinity?" That's why I
posted the reference about 460 unique haplotypes (out of 1200 samples) found
in one small region in Germany. If you happened to have one of the rare
haplotypes, I'd say the probability of a random match is quite low, and even
if you had one of the common haplotypes, a match would still lend some
support to your hypothesis.

I should run this by a statistician before I commit it to posterity, but this
scenario seems analogous to some examples I've seen of "prior" and
"posterior" probabilities. Here the prior probability would be when you just
know one thing, the overall mutation rate, and that is David Andrews' more
pessimistic frame of reference (pardon me for putting words in your mouth,
David!). The posterior probability, "after" you know some more facts such as
the time and place and number of specific haplotypes in the vicinity, is more
optimistic about the value of mtDNA testing.

To my mind, this is an illustration of the value of a hypothesis. If I were
to set up a database where people could record their mtDNA haplotypes,
finding a match might not mean much at all. I'm actually thinking of doing
this, just out of curiosity, if it doesn't take too much effort. I would
certainly enjoy comparing notes with someone who shared my haplotype.

Ann Turner
List Administrator


References:

Iceland study:
http://www.journals.uchicago.edu/cgi-bin/resolve?AJHG991452

Number of haplotypes:
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&;
list_uids=11296889&dopt=Abstract

This thread: