GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2008-05 > 1211221129


From: "Ken Nordtvedt" <>
Subject: [DNA] S21/S28 Split+m223 stuff
Date: Mon, 19 May 2008 12:18:49 -0600


Taking some S21 and S28 databases from others which hopefully are reliably pure haplogroup collections, I stripped them down to 67 marker ones, leaving 130 "S21" and 80 "S28" haplotypes. So from other information we have about the structure of the tree downstream of M269, there seems to have been an explosion of that tree into a number of different surviving branch lines about 3780 years ago.



I get an age back to the common MRCA for S21 and S28 to be 3780 years. The S28 MRCA is almost immediately after that event, while the S21 MRCA does not occur until about 500 years later, 3270 years ago.

For my "I" friends, I find the common ancestor for M223-Roots and P78-Continental3 clades to be 7440 years ago. If you recall, Roots and P78 MRCAs seemed the oldest within the M223 complex.

I am going to take a break from running more databases through the existing programmed excel spreadsheet, and program another spreadsheet to fully do the interclade node time estimates. They are really less subject to pesty factors than are the MRCAs for the clades themselves, because the former have no correction terms to estimate like self variance of a clade needs:

SVar = M [ G - Sum over contributors c of f(c)^2 ]

When you are estimating the age of a node from which separate branch lines go to two different clades, you are guaranteed that every line connecting a haplotype from clade A and a haplotype from clade B must have a length of 2G. That's why there are no corrections to the formal estimator and why the structure of that interclade formal estimator is so intuitive:

2MG = (Sum over a in clade A)(Sum over b in clade B) [ {r(a) - r(b)}^2 ] / N(A)N(B)

Ken


This thread: