GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2007-04 > 1176977214


From: James Heald <>
Subject: [DNA] Cohen data from Behar (2003)
Date: Thu, 19 Apr 2007 11:06:54 +0100
References: <46229610.30005@comcast.net>
In-Reply-To: <46229610.30005@comcast.net>


Thanks to Al, I've been looking at the raw data from Behar (2003), which
can be considered an update of Thomas (1998).

Controversially (at least to Zoossmann-Diskin (2005)), a number of the
Cohen samples from Thomas (1998) were typed differently in Behar's later
paper, apparently because of improvements in technique. In different
cases, both SNP and Y-STR values were affected.

This seems to considerably tidy up the dataset (as well as it being
extended by new data). Many of the previous outliers are reclassified;
an effect is to make the different clusters even more concentrated.


Here is the breakdown. Total number of Ashkenazi Cohens (AC) = 76;
total number of Sephardi Cohens (SC) = 69

I hope I haven't introduced many copying errors


For what it's worth, if the split below between CMH and probable J2 is
correct, re-running the naive ASD calculation of Thomas et al (1998),
using just the members of the pure CMH cluster, this cluster is so
tightly concentrated it gives an ASD Thomas et al would interpret as
corresponding to an average of only *eleven* generations to MRCA.

I don't find that credible; and I don't think the "times 3" tweak factor
is enough to help. I think it shows we're going to need some new, much
more rigorous thinking about what range of cluster ages may be
compatible with very tightly concentrated haplotype distributions.



CMH cluster

393=12 390=23 19=14 391=10 ... 388=16 ... 392=11 AC=36 SC=36
393=12 390=23 19=14 391=10 ... 388=15 ... 392=11 AC= 2 SC= 1
393=12 390=23 19=14 391=10 ... 388=17 ... 392=11 AC= 1 SC= 1
393=12 390=22 19=14 391=10 ... 388=16 ... 392=11 AC= 1 SC= 1
393=12 390=24 19=14 391=10 ... 388=16 ... 392=11 AC= 1 SC= 1
393=12 390=23 19=14 391=10 ... 388=16 ... 392=12 SC= 1

393=12 390=24 19=14 391=10 ... 388=18 ... 392=11 SC= 1

sub-total: AC= 41 (54%) SC= 42 (61%)



19=15: Probable J2

393=12 390=23 19=15 391=10 ... 388=16 ... 392=11 AC=10
393=12 390=24 19=15 391=10 ... 388=15 ... 392=11 AC= 8
393=12 390=23 19=15 391=10 ... 388=15 ... 392=11 AC= 4
393=13 390=24 19=15 391=10 ... 388=15 ... 392=11 AC= 2
393=12 390=22 19=15 391=10 ... 388=15 ... 392=11 AC= 1
393=12 390=23 19=15 391=10 ... 388=15 ... 392=13 AC= 1

393=12 390=24 19=15 391= 9 ... 388=16 ... 392=11 SC= 4
393=11 390=24 19=15 391= 9 ... 388=16 ... 392=11 SC= 1

sub-total: AC= 26 SC= 5



J1, 388=13

393=12 390=23 19=14 391=10 ... 388=13 ... 392=11 SC= 4



DYS19=14, 392=13 70% likely J2a1b according to Athey

393=12 390=22 19=14 391=10 ... 388=15 ... 392=13 SC= 1



Haplogroups G/H/I ("BR(xDE,JR)")

393=13 390=23 19=15 391=10 ... 388=12 ... 392=11 SC= 1 G5 ?
393=12 390=24 19=14 391=11 ... 388=12 ... 392=11 AC= 1 H ?
393=14 390=23 19=15 391=10 ... 388=12 ... 392=12 SC= 1 I1b2a ?

sub-total: AC= 1 SC= 2
** according to the table in the published paper, there should be one
more AC here, one less in Haplogroup J.


Haplogroup K

393=13 390=23 19=14 391=10 ... 388=12 ... 392=13 AC= 1 K2 ?
393=13 390=24 19=15 391=10 ... 388=12 ... 392=14 SC= 1 K2 ?
393=13 390=21 19=14 391=10 ... 388=12 ... 392=14 AC= 1 K2 ?
393=12 390=22 19=14 391=10 ... 388=12 ... 392=13 SC= 1 K2 ?

sub-total: AC= 2 SC= 2



Haplogroup E3b

393=14 390=25 19=13 391=09 ... 388=12 ... 392=11 AC= 3
393=13 390=24 19=13 391=10 ... 388=12 ... 392=11 SC= 2
393=13 390=23 19=13 391=10 ... 388=12 ... 392=11 SC= 1

sub-total: AC= 3 SC= 3



Haplogroup Q

393=13 390=22 19=13 391=10 ... 388=12 ... 392=16 SC= 1
393=13 390=22 19=13 391=10 ... 388=13 ... 392=16 SC= 1

sub-total: AC= 0 SC= 2



Haplogroup R1a1

393=13 390=25 19=16 391=11 ... 388=12 ... 392=11 SC= 2
393=13 390=24 19=16 391=11 ... 388=12 ... 392=11 SC= 1
393=13 390=25 19=15 391=10 ... 388=12 ... 392=11 AC= 1

sub-total: AC= 1 SC= 3




Haplogroup R1b

393=12 390=23 19=14 391=10 ... 388=12 ... 392=14 AC= 1
393=12 390=24 19=14 391=10 ... 388=12 ... 392=14 SC= 2
393=12 390=24 19=14 391=10 ... 388=12 ... 392=13 SC= 1
393=12 390=25 19=14 391=10 ... 388=12 ... 392=14 SC= 1

393=14 390=23 19=14 391=10 ... 388=12 ... 392=10 AC= 1 G2 ???

sub-total: AC= 2 SC= 4



Data taken from Zoossmann-Diskin (2005). Haplogroup assignments beyond
the classes in Thomas et al (1998) made using the Athey predictor, which
match totals in the published table in Behar (2003).



Bonnie Schrack wrote, discussing the Cohen data from Thomas (1998):

>> I was
>> looking into the Cohanim question last night, and downloaded the
>> original data for one of the classic studies, Origins of Old Testament
>> Priests, by Thomas et al, Nature, vol. 394, 9 Jul 1998.
>>
>> It's rather fascinating data. It showed that of 106 Cohanim studied
>> (total of Ashkenazi and Sephardic), 54 had the 6-marker CMH, or 50.9%.
>> The rest were scattered in very small numbers over a lot of other
>> haplotypes. I will have to spend more time later to confirm the exact
>> haplogroups signified by the categories they used in that early study,
>> but from what I can tell, 1 Sephardic Cohen was somewhere in E (likely
>> E3b), and in the R macrohaplogroup (probably R1b, but could also include
>> R1a) there were 2 Ashkenazi and 6 Sephardic Cohanim. The rest were in a
>> large polyphyletic group, that is, made up of more than one
>> haplogroup. But it looks like perhaps 8 or 9 of the Cohanim might
>> possibly be in non-J haplogroups within that larger group (outside of E
>> and R) -- and as I said, this is only a report of my work in progress to
>> decipher the data.
>>
>> So then, among the scattered haplotypes which a few Cohanim had, there
>> were some that did look like J, but were not the "CMH." Most of these
>> were singletons, but there were a few small clusters. Now here's the
>> good part! Although the group we've just discussed
>> was present, there was only 1 Ashenkenazi Cohen in it (and one
>> Israelite, or non-priest). However! There was a cluster I had not seen
>> before, as well. This cluster had 6 Sephardic Cohanim in it. Its
>> haplotype, in this paper by Thomas, et al, is:
>> 393 - 12
>> 390 - 23
>> 19 - 15
>> 391 - 10
>> 388 - 16
>>
>> This cluster does exist on Ysearch, and three of
>> its members have the surname of Cohen, a pretty good indication that
>> they do have a Cohen tradition!
>> There are nine of them at my count, and here's their haplotype compared
>> to the larger group that I was already aware of:
>
>> ____"Pre-J2a1k"____"New" J2 Cohen cluster
>> 393 - 12 - 12
>> 390 - 23 - 23
>> 19 - 15 - 15
>> 391 - 10 - 10
>> 385a - 14 - 14
>> 385b - 17 - 17
>> 426 - 11 - 11
>> 388 - 15 - 16
>> 439 - 12 - 12
>> 389i - 13 - 14
>> 392 - 11 - 11
>> 389ii - 29 - 30
>>
>> As you know, the change from 13-29 to 14-30 at DYS389I-II is really only
>> a one-step mutation. With only two steps of genetic distance between
>> these groups' modal haplotypes, I have little doubt that they're
>> related, especially as they share the distinctive 14-17 at 385a-b.
>


This thread: