GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-12 > 1197825443


From: "Ian Logan" <>
Subject: Re: [DNA] SNPs as Clock
Date: Sun, 16 Dec 2007 17:20:21 -0000
References: <001b01c84000$9a40e120$6400a8c0@Ken1>


Ken & List

'Kivisild' is a mitochondrial expert who now seems to have moved across to the study of SNPs
in the autosomes and in particular the Y-chromosome.

So it is no surprise to me so see that the structure of 'Figure 7' shows a striking similarity to
the
mitochondrial phyllogenetic tree.

And, like with the mitochondrial tree, what is needed is a lot more data.

So what data is available at present ? Is it only the J. Watson and C. Ventner genomes ?

Is there any other published Illumina data, yet ?

And, are 'we' ready to handle the large amount of data when it does arrive ?

Ian

...........
Ken wrote ...
Figure 7 of the new Underhill-Kivisild paper could in principle be viewed as a clock. But if that
is done, it seems the SNP mutation clock rate behavior is strange. Since the discovered SNPs from
a survey of part of the y chromosome are printed one upon the other with equal line spacing, and the
numbers of SNPs which occured in the tree branch lines are therefore determining the printed length
of the branch lines shown in Figure 7, the tree ends up having some sort of time scale associated
with it.

The problem is that all the dna samples from the various haplogroups have the same date of "now".
So if the SNP mutation clock were perfect, the branch lines leading to the dna samples should all
end up at the bottom of the figure at a common vertical dimension (time = now). But since SNP
mutations are random events, there should be statistical flucuations between their actual numbers
found for each branch length and the true elapsed intervals of time for the branch lengths. So the
25 SNPs on the branch length leading to I1a should have a statistical standard deviation of plus or
minus 5, etc.

But after allowing for such natural and expected statistical flucuations, and allowing for a few
gaps in the vertical lists of SNps where tree splittings have butted into the display, it seems the
differences of "time" locations for all the present-day dna samples are too large. Was there some
sort of bias present in the produced list of SNPs, perhaps mentioned in the paper, which would
degrade the clock-like behavior of the collection of SNPs after being located into a phylogenetic
Tree?
Comments?
-------------------------------

This thread: