GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2011-08 > 1314299472


From:
Subject: Re: [DNA] The peopling of Europe and the cautionarytaleofYchromosome lineage R-M269
Date: Thu, 25 Aug 2011 15:11:12 -0400


Ken Nordtvedt published a scientific article with his demonstration. I lost the link to this work but it has been challenged many times on this list (you challenged it) and each time Ken was able to maintain his point. I don't have the same level (applying statistics isn't the same as demonstrating a new point) but I have been following these discussions and I trust his approach.

The paper is challenging the linearity between variance and age but it doesn't reach any point.

Actually there is a problem of confusion with this paper. It brings new data and produces new maps but then it focuses only on a point of interpretation of variance results. The data collected are not discussed for what they show. For example, Figure 2b, with the labeling from supplementary materials (R(M269 xR-S127) is showing a map very close to what the old p49,f RFLP ht35 group was previously found to be. Not a word on these "old" data. This map is showing a clear cut stage where "R-M269" colonized eastern Europe prior to a next migration to the west. This description in 2 stages (an interpretation but an easy way to describe the case) is completely missing. Missing for what ?
I mentioned the case on R-U106 discussed by Mike Whalen . A variance (intraclade) over 2.0 is extremely surprising (for this group).



> Neither intraclade nor interclade are unbiased.
>
> In the interclade method, it is the case that
>
> E[ASD | g] = 2*mu*g
>
> but it is NOT the case that
>
> E[g | ASD] = ASD/(2*mu)
>
> The latter is equal to an integral over all g of g * P(ASD|g) * P(g) / P(ASD)
> and hence depends on the prior for g
>
> A good way to see why the interclade method is biased is to consider
> the case of ASD=0 in which case the estimate of g=0, but the
> E[g|ASD=0] is 0 only in the contrived case where the prior P(g) is the
> Dirac delta on g=0. But, in general, E[g|ASD] depends on the shape of
> the prior P(g), which, in turn, depends on assumptions about
> population history.
>
> Actually all of the above is irrelevant for the current paper, which
> addresses a completely different problem, namely the idea that
> E[ASD|g] = 2*mu*g itself. The authors demonstrate that this is not the
> case, and that E[ASD|G] < 2*mu*g for ages beyond the range in which a
> marker behaves linearly.
>
> On Thu, Aug 25, 2011 at 7:49 PM, <> wrote:
> > Intraclade isn't unbiased as was discussed many times on this list. So, who can exclude that some of the problems noted by the authors are not because of intraclade problems.
>
> --
> Dienekes' Anthropology Blog: http://dienekes.blogspot.com
> Dodecad Ancestry Project: http://dodecad.blogspot.com
>
> -------------------------------
> To unsubscribe from the list, please send an email to with the word 'unsubscribe' without the quotes in the subject and the body of the message

This thread: