GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2010-06 > 1276738192


From: "Ken Nordtvedt" <>
Subject: [DNA] My a-posteriori STR mutation rates
Date: Wed, 16 Jun 2010 19:29:52 -0600


I finished a run through about 70 or so clades in haplogroup I --- representing several thousand haplotypes --- and evaluated a-posteriori RELATIVE mutation rates for the STRs I use routinely in Generations5.

You can view and download an excel file of the a-posteriori rates versus assumed initial rates in file "A-posteriori-M.xls" at

http://knordtvedt.home.bresnan.net

This is preliminary. I have to run through the output table to look for any anomalies. These sometimes happen when super outliers are not handled well. Super outliers are probably due to either multi-step mutations or clerical errors of entry into ysearch, etc.

There are some significant adjustments of well known STRs. DYS388 is the most dramatic, I think. Other folks had argued for years on this list that they saw DYS388 behaving with higher mutation rates than standard accepted rate; so I apparently see it, too, in haplogroup I. Is the low 12 repeats for DYS388 in the dominant R haplogroup a dna configuration of unusual mutation suppression?

The sum of aposteriori rates is the same as the sum of the old rates. So these changes are relative.

But changes in the relative rates will be important if others use a truncated set of the STRs as compared to the STRs I use in Generations5. Because it could affect the ratio of the sum of marker rates for one person's set of employed STRs as compared to another's sum of marker rates. Sum of marker rates is the important calibrator which converts sum of STR variances into the clade age estimate.

The use of 70 or so clades in my methodology is to have as many independent productions of STR mutations from arbitrary clade founding values as possible as input. Then the formula for aposteriori relative m(i) was used:

m(i)* = Sum over c of { Var(c,i) ] / Sum over c of { G(c)* }

With G(c)* = Sum over i of { Var(c,i) } / M

Var(c,i) = self variance of STR i in clade c
M = Sum over i of m(i)

This thread: