GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-05 > 1178011986


From: "Dienekes Pontikos" <>
Subject: Re: [DNA] The new J2 Cohen-Levi Modal Haplotype (so-called)
Date: Tue, 1 May 2007 02:33:06 -0700
References: <463404D5.3050204@comcast.net><f3f05ce80704301313l14e17462ob2a95805c3be592@mail.gmail.com><46367FC2.2090201@ucl.ac.uk>
In-Reply-To: <46367FC2.2090201@ucl.ac.uk>


Dear James,

It is possible that CMHs among Cohens fall mostly in J1; my point is
that this is undocumented, not that it is false. Your claim that a
second CMH/J2 cluster is statistically unlikely would indeed be valid
if the public database you refer to was a representative sample of the
Cohen Y-DNA gene pool. However, I do not have much confidence that
this is the case for two main reasons: (a) it does not consist of a
random sample of world Jews, since as you know American Ashkenazi Jews
from Eastern Europe make the bulk of the people taking genetic tests,
(b) the publicity surrounding the CMH and the suggestion that the
CMH-12 is an "updated" version of the CMH may have discouraged non-CMH
or non-CMH-12 Cohens from releasing their information.

Therefore, while it may be true that many of the 6-marker Cohen CMHs
reported in the literature may be in J1 or related to the extended
CMH-12, that still needs to be demonstrated.

On 4/30/07, James Heald <> wrote:
> Dienekes,
>
> The frequency of the CMH in Jews as a whole is not very relevant,
> because, as all the Cohen studies have demonstrated, the Cohen
> population is very distinct, with a very different balance of haplotypes.
>
> It is a fact, albeit admittedly with a small sample size, that if you
> look up Cohen-type names in the public DNA databases, ones with a CMH
> haplotype have extended haplotypes that are tightly clustered and
> characteristic of J1.
>
> Could there be a second Cohen cluster with the CMH haplotype but in J2 ?
> While it's not completely impossible, it is statistically unlikely. It
> flies in the face of the rigorous Bayesian formulation of Occam's razor.
>
> Suppose, a-priori, one were to consider it equally likely that there
> were two big Cohen clusters, or that there was one. But then one learns
> that the Cohens in the big cluster/s are all associated with the CMH.
> If there is only one big cluster, it is no surprise to find one main
> haplotype. But if there were two big clusters of independent descent,
> then why out of all the haplotypes in the world, should they both fetch
> up at the same one? It's not impossible, but the odds against it are
> very long.
>
> So despite the a-priori evenhandedness, a-posteriori of learning that
> there is only one main haplotype, it is much more probable that there is
> a single cluster not two different independent ones.
>
> It's a Bayesian take on Occam's razor. If the data needs two big
> clusters for its explanation, then it is justified to introduce them.
> But if the data does not, then the introduction of additional entities
> are not justified. (And one can quantify just how unjustified, by
> considering the unlikeliness of the second cluster haplotype exactly
> matching the first, out of the full distribution of possible haplotypes
> it could otherwise have taken).
>
>
> The CMH Cohens are much more likely to be in J1 than J2.
>
> -- James.
>
>
>
>
> Dienekes Pontikos wrote:
>
> > Among Ashkenazi Jews, the CMH is found at a sligtly higher frequency
> > in J1 (6.3%) compared to J2 (5.7%). In Anatolians, it occurs equally
> > in J1 and J2. I did not check these figures since my 2005 post, so let
> > me know if I tabulated wrong:
> >
> > http://dienekes.blogspot.com/2005/11/what-was-really-aarons-lineage-cohen.html
> >
> > In a different study, it was said that "Haplogroup membership was
> > determined for 266 samples matching at ≥5 of the CMH STR alleles,
> > defined as the Cohen Modal Haplogroup (CMHg). The bulk of the CMHg
> > chromosomes were observed in J1 (53.0%) and J2 (43.2%), with a small
> > portion falling outside of haplogroup J (3.8%)."
> >
> > http://dienekes.blogspot.com/2005/12/more-on-cohen-modal-haplotype.html
> >
> > In any case, I don't see any evidence that the CMH occurs at a
> > substantially higher rate in a J1 compared to a J2 background. The
> > existing published data on Cohen haplotypes are certainly insufficient
> > to demonstrate whether or not the majority of CMH-bearing Cohens are
> > in J1 or J2, or whether the majority of J-haplogroup Cohens are in J1
> > or J2.
> >
> > On 4/28/07, Bonnie Schrack <> wrote:
> >
> >>I have no time for this, but again, I worry that people will be misled
> >>by Sasson's postings.
> >>
> >>He wrote:
> >>
> >>> I just have counted how many of the Cohen and Levite haplotypes in Behar
> >>>data are estimated to be J1 or how many - J2. There are much more haplotypes
> >>>in J2.
> >>>
> >>
> >>Apparently he means, estimated by Whit's Predictor. I'd like Whit to
> >>come on himself and comment on this, but I think there's something fishy
> >>about the statistics quoted. For example, the CMH itself:
> >>
> >> >(14,16,23,10,11,12) 3 2 off the CLMH J2=49.4% CMH
> >>
> >> >The probability of
> >> >being J2 (obtained with older version of Haplotype predictor) is shown for
> >> >some of the haplotypes
> >>
> >>I have a hard time believing there has been any version of the Predictor
> >>that gave a higher probability for J2 than J1, for the CMH! Currently
> >>there are two versions of the Predictor, one with 15 haplogroups and the
> >>more recent Beta version with 21.
> >>
> >>Here are the actual estimates given by both versions of the Predictor
> >>program, and in each case I give the result with the Mediterranean
> >>region selected, or with Equal Priors (this refers to the Bayesian
> >>algorithm).
> >>
> >>For the CMH (12-23-14-10---16---11)
> >>Whit's Haplogroup Predictor scores for Mediterranean and for Equal
> >>Priors, latest version with 21 haplogroups:
> >>
> >>Haplo-------Fitness---Prob. Med.---Prob. Eq. Pr.
> >>J1------------100-----87.0---------74.6
> >>J2a1b----------81------9.6---------17.7
> >>J2a1k----------49------0.3----------1.3
> >>J2a(xJ2a1-bk)--70------3.0----------6.2
> >>J2b------------31------0.1----------0.1
> >>
> >>Predictor scores for Mediterranean and Equal Priors, version with 15
> >>haplogroups:
> >>Haplo---Fitness---Prob. Med.-----Prob. Eq. Pr.
> >>J1-------100-------84.6-----------89.9
> >>J2--------83-------15.4-----------10.0
> >>
> >>As can be seen from this, the estimates given by the Predictor for the
> >>CMH are far higher for J1 than for J2. So where did that "49.4% J2"
> >>estimate for the CMH come from?
> >>
> >>Whit is has updated the Predictor various times as he expands his data
> >>set and gains a better understanding of the haplogroups and clades, with
> >>some assistance from me in the case of the J haplogroup. Since these
> >>updates reflect improved knowledge, it would make no sense to go back to
> >>any earlier versions, if these are what Sasson is referring to.
> >>
> >>Now, on the Levites, I had written:
> >>
> >>>In all of Behar's data, I can see only eight Levites in haplogroup J, who have a total of four haplotypes
> >>
> >>And Sasson wrote:
> >>
> >>>Actually, there are 11 Levite haplotypes, not just four!
> >>
> >>Yes, there are 11. Now I can see why I said that -- at that moment I
> >>was looking at a file of Behar's data which Al had prepared, in which he
> >>had segregated all the J haplotypes that Cohanim had at the top, and far
> >>below, 105 rows down, after all the Cohanim that belong to other
> >>haplogroups, and all the data for BR*(xDE,JR) and E*(xE3a), there were
> >>the other J haplotypes that none of the Cohanim had. That's where the
> >>other 7 Levite haplotypes were found, which I had temporarily
> >>overlooked. Of these, two can be recognized as J1, three as probably J2,
> >>and the other two are indeterminate.
> >>
> >>Now let's go back to Sasson's statement:
> >>
> >>
> >>> I just have counted how many of the Cohen and Levite haplotypes in Behar
> >>>data are estimated to be J1 or how many - J2. There are much more haplotypes
> >>>in J2.
> >>>
> >>
> >>Notice how tricky and vague this is. "Cohen and Levite haplotypes" are
> >>lumped together, and a general statement is made about *all of them
> >>together.* But in reality, they are two different groups with different
> >>distributions.
> >>
> >>Furthermore, the key fact is the number of Cohen or Levite individuals
> >>in a given haplogroup, not "how many haplotypes are estimated to be J1
> >>or . . . J2." There are 71 Cohanim in the Behar dataset who share the
> >>CMH haplotype. Then there are a lot of haplotypes that have just one or
> >>a few Cohanim. Does it matter if a fair number of these singletons are
> >>J2? A total of 47 Cohanim have these other J haplotypes. Thus, even if
> >>many of those were by some chance J2, it's hard to see any way that it
> >>could be argued that J2 was predominant among Cohanim, given the 71 who
> >>have the CMH.
> >>
> >>Oh yes -- my statements assume what many of us find obvious -- that the
> >>Cohanim who have the CMH are J1.
> >>
> >>Sasson -- if you believe that there are Cohanim who have the CMH who are
> >>J2, why don't you email some of the J2 people who have the CMH, and see
> >>if you can find any of them who will tell you they have a tradition of
> >>being Cohanim. If you do, it will be a first. If you are truly
> >>interested in this as a fair scientific inquiry, why not do that, and
> >>also write to some of those in J1 who have the CMH. Ask if they are
> >>Cohanim, and I'll bet you will quickly find many who are.
> >>
> >>But if you did some of this and what you found didn't support your
> >>theory, I wonder whether that would find its way to the list.
> >>
> >>For my part, I have never had a fixed theory about the Cohanim before
> >>studying the data, but have only observed and listened to what the data
> >>tell, as I hope to contribute to a clearer and more objective
> >>understanding of this question.
> >>
> >>Bonnie
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>-------------------------------
> >>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
> >>
> >
> >
> >
>
>
>
> -------------------------------
> 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


--

Dienekes' Anthropology Blog
http://dienekes.blogspot.com

This thread: