GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2005-04 > 1114553913


From: "gareth.henson" <>
Subject: Re: [DNA] Sometimes seeing pink
Date: Tue, 26 Apr 2005 23:18:33 +0100
References: <001e01c54a74$1f6e20e0$c6559045@Ken1> <01bc01c54a90$6bb70630$b9d82f50@Masterbedroom> <REME20050426151238@alum.mit.edu> <01f601c54a97$dfbeaea0$b9d82f50@Masterbedroom> <REME20050426161905@alum.mit.edu>


John

OK, but how do you get to your 500,000 men alive 3000 years ago? What was
the overall human population at that time? Divide by 2 to get males, by
another 2 to get adult males, possibly by another 2 to allow for brothers
with identical Y chromosomes? Then what proportion "daughtered out" over the
3000 intervening years?
If the 500,000 is about right (and I agree you need upstream markers for an
unambiguous tree) how skewed could the distribution be by 2005? Presumably
x% of SNPs will cover 95% of the modern day population, where x is less than
95 (possibly a lot less?)

Gareth



----- Original Message -----
From: "John Chandler" <>
To: <>
Sent: Tuesday, April 26, 2005 9:20 PM
Subject: Re: [DNA] Sometimes seeing pink


> Gareth wrote:
> > I agree with your corollary. But it also follows that as more and more
tests
> > are done, the "unlikely" cases will turn up in proportion to their
> > likelihood.
>
> Certainly, but so what? That would indeed serve as a demonstration that
> convergence *can* happen, but we already know that, and we know that it's
> unlikely to be the explanation of any given case of matching haplotypes.
>
> > Why do we need a million new SNPs to get to say 2000 BC? If the current
tree
> > takes us to about 10000 BC (and better for some groups) with a few
hundred
> > SNPs,
>
> But it doesn't. You're neglecting all the "*"s.
>
> > Does the
> > agricultural population expansion come into the equation?
>
> Absolutely. I figure that there must be at least half a million men
> who were alive 3000 years ago with descendants living today. That's
> 500,000 SNPs to tell them all apart, plus 500,000 more upstream SNPs
> to organize them into a tree, making due allowance for redundancy.
> Even if I'm wrong by a factor of ten, the qualitative picture is clear
> -- current technology has no hope of approaching the level of SNP
> specificity that seems to be expected from it. Then, too, the whole
> philosophy of SNP testing falls apart when the number of SNPs grows so
> large -- these are not really unique events after all, and you really
> do have to test the upstream SNPs to make sure you haven't got a
> "private" rerun of a tree-branching SNP. The technique for the next
> stage has to be sequencing and comparing the whole Y chromosome.
>
> John Chandler


This thread: