GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2010-08 > 1281428259
From: Sasson Margaliot <>
Subject: Re: [DNA] SRY10831 at the root ?
Date: Tue, 10 Aug 2010 11:17:39 +0300
> The small change is producing a completely new tree
> because the root gives rise to 3 major lines
> about equally distant from each other.
The three lines are SUPPOSED to be about equally distant from each.
This is the 'empirical' prediction that your hypothesis makes.
Is it confirmed by 'experiment' ? No.
Did you see the results Tim Janzen posted a year ago?
Here I repeat only results of comparing A with all other Haplogroups using
A/B node: 35089
A/C node: 63495
A/E node: 57130
A/F node: 49312
A/G node: 57469
A/H node: 62347
A/I node: 54443
A/J node: 81744
A/L node: 58009
A/N node: 82927
A/O node: 81833
A/Q node: 61898
A/R node: 63410
A/T node: 71635
You see that A is much closer to B than it is to the rest, contrary to what
is predicted by your hypothesis
But lets compare compare the average distance of A to groups C,E,F,G,H
( which is 58K )
to the average distance of A to groups
I,J,L,N,O,Q,R,T ( which is 69K )
Not only both numbers are much grater than the distance A to B,
but, in agreement with "IJK first" hypothesis,
the distance of A to groups within IJK is systematically greater
than the distance from A to the groups outside IJK.
This is exactly what is predicted by the model in which it is IJK
that branched out first.
Notice that the absolute numbers should not be interpreted as "years",
but as some abstract measure of relative genetic distance.
> I agree that we need more haplotypes from cases near the root ;
> it should be a goal for the future.
Whereever the root is, we cannot have "haplotypes from cases near the root",
because all the present time haplotypes are, by definition, equidistant from
But we can find the Root - by looking for an Edge in the Graph where
splitting the Graph into two Trees leads to the most compact couple of
The combination of the Graph established by SNPs with quantitative
information coming from STR ( I mean by running Ken's Generation5 program
for various collections of 67-marker samples using largest number of
Ideally we should not use "one letter haplogroups, like I and J", but
"letter-digit haplogroups, like "I1, I2, J1, J2, J* etc ). At this level of
resolution we may get to some statistically significant results.
The preliminary results we have today hint towards a possibility of arriving
to interesting surprises.
On Tue, Aug 10, 2010 at 2:24 AM, <> wrote:
> The small change is producing a completely new tree because the root gives
> rise to 3 major lines about equally distant from each other. This is somehow
> equivalent to the tree between Eucaryotes, Procaryotes and the
> I agree that we need more haplotypes from cases near the root ; it should
> be a goal for the future.
> About group B , I wasn't aware of cases in Pakistan.