Archiver > GENEALOGY-DNA > 2007-05 > 1180449520

From: "Ken Nordtvedt" <>
Subject: Re: [DNA] Do we Understand y-Haplogroup F?
Date: Tue, 29 May 2007 08:38:40 -0600
References: <BAY111-DAV3656D5EDFF08CDE7E0445B12F0@phx.gbl>

----- Original Message -----
From: "Lawrence Mayka" <>
To: <>
Sent: Tuesday, May 29, 2007 4:52 AM
Subject: Re: [DNA] Do we Understand y-Haplogroup F?

> Asian clades are _extremely_ relevant to your tree.
> Remember your own fundamental assumption that each line on your graph
> contains a negligible number of back-mutations on your slow markers. This
> is only true if each line represents a small interval of time.

The expected number of back mutations can be calculated. Keeping that
number under control is one of the reasons to use only the slowest markers.

And a stretching or enlargment factor can be ascertained for the the long
branches which adjusts them for the expected number of back mutations. It's
not going to double their length.

I think I mentioned that 21,000 years or 700 generations is a typical length
between an E clade and the other clades of Europe. My average marker hardly
has time to have one expected mutation in that time. Only half the cases of
a marker having two one-step mutations will be back mutations. When the
smoke clears I could see a 20 or 25 percent correction to a 700 generation
path length occuring --- but it looks like that length needs to be doubled
or tripled to allow F* to be over 50,000 years old as others suggest it is
in age.
> My suggestion is that your current tree is still far too elided,
> particularly at the macrohaplogroup levels, and is therefore missing many
> back-mutations. Asian clades will eventually bring these out.
> Even with respect to Europe: Keep in mind that right now, you have
> deliberately focused on populous, easily distinguishable clades, and have
> avoided consideration of the outliers in every haplogroup who apparently
> belong to clades now almost extinct (or common only in parts of the world
> not yet well tested). Once again, these additional clades, when
> identified,
> will show how many back-mutations are hidden in your current tree.

Again, I think you envision much too much adjustment needed for back
mutations. I will have to dig out the calculations I did concerning them of
a few months ago and get the numbers --- I forgot the results because it did
not turn out to be very large. If the fundamental problem right now is
that clades of E have too small a GD from clades such as G,H,I,J,K,...R and
necessarily include clade F* (beginning of F), adding missing clades can not
solve that problem. Missing clades added later might cause other problems,
like not being able to find a branch off point from their cousins already in
the tree, but they can't make this distance-to -E -clades problem better.
The fact that new clades found later must find their branches fitting into
the existing tree is important in my view. It means the tree model is
refutable, because a new clade added to the GD matrix could conceivably have
a set of GDs to the other N clades which would be impossible to accomodate.
That has not happened yet.


This thread: