PDA

View Full Version : Teal discovered ??!!



Pages : 1 2 [3]

Kale
11-28-2015, 07:20 PM
You may want to brush up on how to read Dstats and the significance of the Z score, as some of the stats you posted are insignificant (Z<3)

You interpreted this wrong...they are ALL insignificant, because age is an insignificant factor. If age WAS a factor, all those stats would be significant (or near significant) toward the older genome in the comparison, but they are not.

Megalophias
11-28-2015, 07:41 PM
Does anyone have access to the Anzick-1 genome? That would be ideal to test the role of age, I think.

e.g. Primate, Ust'-Ishim; Karitiana, Anzick-1.

K33
11-28-2015, 08:23 PM
Does anyone have access to the Anzick-1 genome? That would be ideal to test the role of age, I think.

e.g. Primate, Ust'-Ishim; Karitiana, Anzick-1.
Anzick-1 is available here (http://www.y-str.org/p/ancient-dna.html). It's Gedmatch kit# F999919

K33
11-28-2015, 08:32 PM
The paper's wrong. Lots of papers that get published are wrong. Nothing new in that.
Could you please elaborate on this statement? It seems like there has been a sizeable cro-magnid-esque component to Eurasian populations, peaking in S-C Asia, that every calculator was unable to identify. And now this piece appears to be an almost perfect fit to that puzzle.

CHG doesn't appear as an intermediate between say, ENF and EHG on the PCA plot. So why would you reflexively assume that is is a hybrid?

Personally I believe they will find more isolated HG populations in the future... in particular in the Altai mountains, possibly associated with Turkic-Tungusic peoples. But again I fail to see how that invalidates anything in the paper re: CHG.

Anabasis
11-28-2015, 08:46 PM
..... and north via the Caucasus to the steppe, taking "Teal" (? and M269) with them

Teal <> step ancestry or something like ANE. M269 have not any connection with Teal. Its better to look EHG for seeking the ancestry of so called PIE. Jones study should end that "ANE" mess...

Gravetto-Danubian
11-28-2015, 10:40 PM
Teal <> step ancestry or something like ANE. M269 have not any connection with Teal. Its better to look EHG for seeking the ancestry of so called PIE. Jones study should end that "ANE" mess...

Your statement doesn;t really follow. Autosomal components like 'CHG' are complex mosaics which don't have one to one relationships with Y haplogroups (although there are some trends) - at least by the Copper Age statement I was referring to.

So whilst I agree that its 90% + probably that M269 'originated on the steppe, anyone who is familiar with the Palaeolithic and Mesolithic materials from the Caspian regions, Russia, central Asia etc will be acutely aware of the interactions taking place between north and south, probably along an East Caspian corridor (before it dried up too much). This explains several similarities and diffusions found from Uzbekistan to NW Russia, and it goes toward explaining the presence of 'odd' haplogroups like J and Q in eastern European mesolithic (as we know). So there is also good reason to entertain the possibility that there were also R1b groups in the south/ southeast Caspian area.

And let's keep in mind that the current samples found by the Jones study aren;t a perfect fit for the "Teal' in Yamnaya. So perhaps later samples and/or from different regions might prove a better fit.

You seem preoccupied with compartmentalizing things into mutually exclusive domains "that belongs there, this here". That's not reality.

Kurd
11-28-2015, 11:06 PM
You interpreted this wrong...they are ALL insignificant, because age is an insignificant factor. If age WAS a factor, all those stats would be significant (or near significant) toward the older genome in the comparison, but they are not.

Hey you were the one who cited those 5 stats to corroborate your position that age is an insignificant factor. When I pointed out that you did not read your stats properly, you then switched to Z scores to save your position (BTW lets not forget I was the one that pointed to your Z scores as insignificant). I do thank you though for stimulating an interesting discussion, honest :)

Ok, back to the issue at hand, you are incorrect again. Of course age is a significant factor. If it was not for age and mutations we would all still be apes! amount of change is proportional to mutation rate x time put simply, but I think what you mean is that 10000 years or 20000 years of non-shared drift is not significant enough to be picked up by analysis using stats. Correct?

Ok, I'll admit it is not easy to prove this because we would have to compare samples from different ages on one side of the split, and we don't have too many of those. So absent anything better, I picked BedouinA and Anatolia_Neolithic, since they are both substantially on the Basal Eurasian side of the split away from Ust. Neither appears to have received more or less ANE or other admixture from the other side of the split (Ust's non Basal Eurasian side) to skew the analysis. Here is what I got:

number of blocks for jackknife: 709
nrows, ncols: 81 508601
result: BedouinA Anatolia_Neolithic Ust_Ishim Gorilla -0.0305 -14.875 14031 14915 284891

I attribute most of this much greater similarity of the Anatolian farmer to him being closer to the split than Bedouin, and thus having less non-shared drift with Ust.

Kale
11-28-2015, 11:22 PM
[SIZE=3]Hey you were the one who cited those 5 stats to corroborate your position that age is an insignificant factor. When I pointed out that you did not read your stats properly,
I read them just fine, you read them incorrectly by assuming I was posting stats that show significance. I was doing in fact the opposite. Insignificant stats reflect an insignificant factor (age).

you then switched to Z scores to save your position
I didn't switch anything. They are both linked...z-score is d-score and overall quantity of comparisons.

(BTW lets not forget I was the one that pointed to your Z scores as insignificant).
Let's also credit you with discovering the sky is blue.


Of course age is a significant factor. If it was not for age and mutations we would all still be apes! amount of change is proportional to mutation rate x time put simply, but I think what you mean is that 10000 years or 20000 years of non-shared drift is not significant enough to be picked up by analysis using stats. Correct?
Yes, I already said that.


Ok, I'll admit it is not easy to prove this because we would have to compare samples from different ages on one side of the split, and we don't have too many of those. So absent anything better, I picked BedouinA and Anatolia_Neolithic, since they are both substantially on the Basal Eurasian side of the split away from Ust. Neither appears to have received more or less ANE or other admixture from the other side of the split (Ust's non Basal Eurasian side) to skew the analysis.
And what are your sources on this?
Chimp Loschbour Mbuti Ust-Ishim 0.3263 58.776 40774 20711 466345
Chimp Loschbour Mbuti Anatolia_Neolithic 0.3905 100.000 501921
There is inordinate sharing between Anatolia_Neolithic and Loschbour beyond Ust-Ishim.
Chimp Anatolia_Neolithic Mbuti Bichon 0.3856 100.000 390026
Chimp Anatolia_Neolithic Mbuti Hungary_HG 0.3977 100.000 430538
Chimp Anatolia_Neolithic Mbuti Iberia_Mesolithic 0.3857 100.000 514490
Chimp Anatolia_Neolithic Mbuti Loschbour 0.3893 100.000 501921
Differential sharing between HG's reinforces this...Those 6,000 years between Bichon and his descendents should have pulled his descendents away from Anatolia_Neolithic according to you, but in fact, they're closer.

K33
11-29-2015, 12:52 AM
I believe that populations similar to BA Armenians ( which are different from modern Armenians) in combination with Eastern Hunter Gatherer types contributed to the creation of Yamanya typeDavidski specifically tested Bronze Age Armenians as a stand-in for the "Teal" component of Yamanya, and it did not work. The algorithm instead pointed directly at modern Mingrelians as the closest match to the non-EHG component of Yamnaya. Mingrelians presumably have lower ENF admixture than Georgians in general.

TreeMix is very specific and precise about this. In my analyses, based on a couple of different methods, the Mingrelians are the only population chosen as a source for the Near Eastern-related ancestry in the Yamnaya.

Keep in mind, this is an unsupervised test and the algorithm has an infinite number of choices, because migration edges can run from any part of the tree, and yet it chooses the Mingrelians. By the way, if anyone's wondering, I did also try the Bronze Age Armenians, to no avail.
http://eurogenes.blogspot.com.au/2015/10/yamnayas-exotic-ancestry-kartvelian.html

Kurd
11-29-2015, 01:03 AM
Davidski specifically tested Bronze Age Armenians as a stand-in for the "Teal" component of Yamanya, and it did not work. The algorithm instead pointed directly at modern Mingrelians as the closest match to the non-EHG component of Yamnaya. Mingrelians presumably have lower ENF admixture than Georgians in general.

http://eurogenes.blogspot.com.au/2015/10/yamnayas-exotic-ancestry-kartvelian.html

Since I was not able to pin it to the BA Armenians in my database, I used the qualifier "similar". In hindsight that statement (prediction) was not that bad considering that it was made before the Jones paper came out before I even had the CHG genomes :)

Kurd
11-29-2015, 02:40 AM
Chimp Loschbour Mbuti Ust-Ishim 0.3263 58.776 40774 20711 466345
Chimp Loschbour Mbuti Anatolia_Neolithic 0.3905 100.000 501921
There is inordinate sharing between Anatolia_Neolithic and Loschbour beyond Ust-Ishim.
Chimp Anatolia_Neolithic Mbuti Bichon 0.3856 100.000 390026
Chimp Anatolia_Neolithic Mbuti Hungary_HG 0.3977 100.000 430538
Chimp Anatolia_Neolithic Mbuti Iberia_Mesolithic 0.3857 100.000 514490
Chimp Anatolia_Neolithic Mbuti Loschbour 0.3893 100.000 501921
Differential sharing between HG's reinforces this...Those 6,000 years between Bichon and his descendents should have pulled his descendents away from Anatolia_Neolithic according to you, but in fact, they're closer.

I don't see how the 1st stat is relevant to anything here. The 2nd simply shows Anatolian farmer much more similar to Loschbour than Mbuti is, again I don't see how this is relevant. The 3rd shows Bichon much more similar to Anatolian than Mbuti, again no relevancy to this matter. The 4th shows Hungary HG much more similar to Anatolian than Mbuti, ditto. The 5th is the same. All of these not relevant.

If your point is that Anatolian has connections with the non Basal Eurasian side of the split (specifically WHG) that Bedouin does not have, thereby giving Anatolian an unfair advantage, I will even go ahead and stipulate to that because to be fair, Anatolian probably does have more WHG than Bedouin. But I still don't think enough to fully counteract the significant stat I posted (Z=-14.85).

I have highlighted the 3 genomes we associate with WHG in the following table. How would you explain the older ones (La Brana & Bichon) have more similarity to Ust.




OUTGROUP
POP1
TARGET
OUTGROUP
D
Z
SNP


Chimp
MA1
Ust_Ishim
Gorilla
-0.9593
-100
205307


Chimp
LaBrana1
Ust_Ishim
Gorilla
-0.9593
-100
271054


Chimp
Han
Ust_Ishim
Gorilla
-0.9586
-100
286939


Chimp
Dai
Ust_Ishim
Gorilla
-0.9586
-100
286939


Chimp
Papuan
Ust_Ishim
Gorilla
-0.9584
-100
286939


Chimp
KOTIAS
Ust_Ishim
Gorilla
-0.9582
-100
252477


Chimp
Kostenki14
Ust_Ishim
Gorilla
-0.9579
-100
269687


Chimp
Bichon
Ust_Ishim
Gorilla
-0.9578
-100
191261


Chimp
Yamnaya
Ust_Ishim
Gorilla
-0.9575
-100
285906


Chimp
Karelia_HG
Ust_Ishim
Gorilla
-0.957
-100
277029


Chimp
Stuttgart
Ust_Ishim
Gorilla
-0.957
-100
282363


Chimp
Loschbour
Ust_Ishim
Gorilla
-0.9566
-100
284422


Chimp
LBK_EN
Ust_Ishim
Gorilla
-0.9559
-100
286504


Chimp
BedouinB
Ust_Ishim
Gorilla
-0.9555
-100
286939


Chimp
Anatolia_Neolithic
Ust_Ishim
Gorilla
-0.955
-100
284891


Chimp
Yoruba
Ust_Ishim
Gorilla
-0.94
-100
286939

Ryukendo
11-29-2015, 02:44 AM
The motivation for these stats was to check CHG's position vs. ENF's position with regards to ANE. If the assumption is that CHG's affinity to ANE is limited to any drift it shares with ANE, and there was no subsequent ANE input after CHG's split from Basal, then CHG & ENF should have approximately the same affinity to MA1. I say approximately because realistically CHG would appear a little more similar to MA1 than ENF, for the simple reason that the CHG genomes are older than the ENF genomes, and would thus be a little closer to the split of Basal from the rest of the non African groups, resulting in a little less non-shared drift with MA1, and thus should appear a little more similar than ENF to MA1,

I used 2 out groups to gauge absolute similarity to MA1, because for a Dstat of the form D(W, MA1, TARGET, OUT), W will affect the outcome if there is any similarity with the target, that is above and beyond any similarity the other targets share with W.

I first tried using Mbuti as outgroup W, but what I found that some targets had more affinity with Mbuti than others, not because there had been any genefow between Mbuti and the targets subsequent to OOA, but simply because some targets were older, and thus had less non-shred drift with Mbuti. So I opted for chimp gorilla outgroups.

Unfortunately my 2nd Karelian sample did not work out because it only had about 30K overlapping markers. I removed any samples with less than 130K marker overlap, because I found out they were at a disadvantage. Case at point Pashtun_Afghan had only 60K markers overlap, and because of this its affinity to MA1 was much less than Pathan. In fact its position was similar to Bedouin. So for an accurate comparison it is imperative that all samples have similar marker overlaps.

My conclusions based on this table are:

1- CHG shows more similarity to MA1 than ENF, and although it is likely that it is due to actual ANE input post split, there is a small possibility it may be due to CHG being older, and thus having less non-shared drift than ENF with regards to MA1.

2- S and SC Asians have recieved an ANE boost from the steppe post CHG, otherwise they would have appeared much farther from MA1 than CHG because they have much more non-shared drift with MA1 than CHG due to the 9000 or 10000 years that separate the two.

3- Kalash's position is interesting. It is closer to MA1 than other SC Asians, which begs the question why?

4- Somalis Ethiopian Jews and Yoruba were included for member Awale.

The table is sorted with the most similarity to MA1 on top.

Hmm Kurd and Kale,

Not doubting that you discovered weird things with Mbuti here, most definitely you did, and indeed D stats produce weirdness when Chimp and mbuti are included through the 'phantom chimp effect', but I think we have to clarify that, theoretically at least, it isn't true that the direction of D-stats are affected by branch length private to a particular population... at least, a logical reading of the original paper makes this very clear. But perhaps the magnitude of the D-stat can be affected.

Given four populations W, X, Y, Z,
D stats are simply a comparison whether allele frequencies BABA or ABBA predominate, a.k.a whether or not W+Y and/or X+Z share more alleles (BABA), or X+Y and/or W+Z share more alleles (ABBA). Across all marker sites all the cases where we see BABA or ABBA configurations of alleles are recorded, and then the total number of each is compared to see if there is an excess of BABA or ABBA, then the entire thing is divided over the total number of ABBA+BABA, aka

(Sum (BABA) - Sum (ABBA))/(Total number of BABA+ABBA occurrences)

Lets say we have four populations, Chimp, Loschbour, Han, Dai,
D(Chimp Loschbour Han Dai)
This produces zero because Chimp does not share an excess of alleles with Han or Dai, aka the number of B(?)BA and A(?)BA occurrences are equal, and Loschbour likewise does not share an excess of alleles with Han or Dai, aka the number of (?)BBA or (?)BAB occurrences are also equal. But for the following stat:
D(Chimp Han Loschbour Dai)
Chimp contributes a zero term again because Chimp shares an equal number of alleles with Loschbour and Dai, aka there is not an excess of either B(?)BA or A(?)BA, but Han shares an extremely large excess of alleles with Dai over Loschbourr, aka there is a *lot* of (?)BAB over (?)BBA, and so the statistic is very positive.

This should make it clear that if we have an extremely drifted population, let's say population Y in the third position, that should manifest itself as an excess of allele configurations of the type AABA or BBAB, but such configurations of alleles are not utilized in the calculation at all... In other words, suppose a population Y drifted extremely rapidly after all of its interactions with W, X, and Z were completed; this drift path should not bias the stat in any direction.

The reverse of this should also apply... if a population Y for some reason or another has less drift private to itself than is normal, whether because it is an old sample or due to extremely large population size, this should manifest itself as a deficiency of AABA/BBAB configurations, but once again this should not affect the statistics.

Nevertheless Kurd has produced compelling statistics... I think this what has happened is that AABA/BBAB-type mutations destroy a certain fraction of the BABA/ABBA conffigurations of alleles, aka whenever a population has mutation private to itself, it reduces the total number of ABBA/BABA configurations available for the entire calculation, which reduces the significance and magnitude, but does not change the direction, of the D-stats; so it ends up being the case that older populations produce more significant statistics with each other than the younger ones. Kurd, perhaps you would like to try tree-building with the ancient genomes directly, instead of rank orders with moderns?

About the Mbuti, there might be an issue with ascertainment here, because all our snp sets are focused on markers that have mutated in modern Eurasian populations, and this creates problems. The Mbuti have an extremely different drift path from Eurasians, the tree is:

Chimp (Mbuti (Yoruba (Eurasians)))

There are the following sets of mutations from the human-chimp ancestor:
(Chimp)
(Mbuti (Yoruba (Eurasians)))
(Mbuti)
(Yoruba (Eurasians))
(Yoruba)
(Eurasians)

SNPs focusing on Eurasians will capture the following sets of mutations
(Mbuti (Yoruba (Eurasians)))
(Yoruba (Eurasians))
(Eurasians)

but it will not capture the set (mbuti) or (Yoruba), so the sites on which Africans differ from chimpanzee (but Eurasians do not) are not fully captured, while the sites on which Eurasians differ from Chimp (but Africans do not) will manifest fully, which will tend to make Africans look biased towards chimps, which is indeed what we observe.

Eurasians all have extremely similar drift paths however, which makes me really doubt that this effect can operate in Eurasians.

Ryukendo
11-29-2015, 03:14 AM
Could you please elaborate on this statement? It seems like there has been a sizeable cro-magnid-esque component to Eurasian populations, peaking in S-C Asia, that every calculator was unable to identify. And now this piece appears to be an almost perfect fit to that puzzle.

CHG doesn't appear as an intermediate between say, ENF and EHG on the PCA plot. So why would you reflexively assume that is is a hybrid?

Personally I believe they will find more isolated HG populations in the future... in particular in the Altai mountains, possibly associated with Turkic-Tungusic peoples. But again I fail to see how that invalidates anything in the paper re: CHG.

Hi K33, CHG are the same distance from Dai as EEF are, and this is inconsistent with even the author's own models about the composition off CHG and EEF.

The authors show that CHG split off immediately from a Basal Eurasian population, which also contributed, together with WHG, into Early European Farmers, but this is not logical because WHG shares quite some drift with Dai to the exclusion of Basal, which you can see in the diagram, which should logically pull EEFs closer to Dai as a result. The fact that CHG is the same distance from Dai as EEF are therefore means that at least some of the ancestry of CHG shares drift with Dai to the exclusion of Basal, though we cannot place this ancestry precisely except that it is majority comprised of (ANE, WHG, EHG) and is closer to (ANE, WHG, EHG) than to (Kostenki).

Ryukendo
11-29-2015, 03:22 AM
Do you think that when we do discover an ancient genome from Southwest Asia, that this will decrease CHG and CHG-like ancestry in some Middle Easterners substantially considering that CHG still contains a good amount of Basal Eurasian?

Hi Zephyrous,

Thank you for your question. I suspect, like Davidski, that EEF and CHG are both admixed manifestations of an even more Basal Eurasian population that was found further south in the Middle East in ancient periods.

Kale
11-29-2015, 03:48 AM
I don't see how the 1st stat is relevant to anything here. The 2nd simply shows Anatolian farmer much more similar to Loschbour than Mbuti is, again I don't see how this is relevant. The 3rd shows Bichon much more similar to Anatolian than Mbuti, again no relevancy to this matter. The 4th shows Hungary HG much more similar to Anatolian than Mbuti, ditto. The 5th is the same. All of these not relevant.
Those stats I posted are essentially just the double outgroup thing you are doing. If mine aren't relevant, neither are yours. The second outgroup being Mbuti is actually more relevant, because, like we both agreed, age IS a factor after millions of years, and with Chimp AND Gorilla, we're looking at many millions of years of random drift that will undoubtedly show spurious results here and there.


I have highlighted the 3 genomes we associate with WHG in the following table. How would you explain the older ones (La Brana & Bichon) have more similarity to Ust.
Science isn't about highlighting the 3 things that support your theory while ignoring the 8 things that don't.
First and foremost... La Brana is younger than Loschbour!!! (though not by much)...get your facts straight man!
Here's what the order should look like according to your theory
1) Kostenki (people like to cling to the idea that he has basal Eurasian in the same way Neolithic farmers do, he doesn't)
2) MA1
3) Bichon
4) Loschbour
5) LaBrana
6) Karelia
The level basal Eurasian plays could muddle up this middle a bit, but these will be the next block, in no particular order
Han, Dai, Papuan, Kotias, Yamnaya, Stuttgart, LBK_EN, Anatolia_Neolithic
And to round out the bottom
15) BedouinB
16) Yoruba

EDIT: Thank you Ryukendo for the detailed explanation :)
Kurd, I think after Ryu's explanation it seems we were both right in our observations.

Kurd
11-29-2015, 04:42 AM
double post

Kurd
11-29-2015, 04:46 AM
Those stats I posted are essentially just the double outgroup thing you are doing. If mine aren't relevant, neither are yours. The second outgroup being Mbuti is actually more relevant, because, like we both agreed, age IS a factor after millions of years, and with Chimp AND Gorilla, we're looking at many millions of years of random drift that will undoubtedly show spurious results here and there.


Science isn't about highlighting the 3 things that support your theory while ignoring the 8 things that don't.
First and foremost... La Brana is younger than Loschbour!!! (though not by much)...get your facts straight man!
Here's what the order should look like according to your theory
1) Kostenki (people like to cling to the idea that he has basal Eurasian in the same way Neolithic farmers do, he doesn't)
2) MA1
3) Bichon
4) Loschbour
5) LaBrana
6) Karelia
The level basal Eurasian plays could muddle up this middle a bit, but these will be the next block, in no particular order
Han, Dai, Papuan, Kotias, Yamnaya, Stuttgart, LBK_EN, Anatolia_Neolithic
And to round out the bottom
15) BedouinB
16) Yoruba

EDIT: Thank you Ryukendo for the detailed explanation :)
Kurd, I think after Ryu's explanation it seems we were both right in our observations.

Ok, I took out the samples that are Basal Eurasian admixed, except for the ones believed to be majority Basal (to see the pattern). LaBrana seems to be the only one outside the curve ( You are correct, I don't know what I was thinking earlier with LaBrana). Also, based on my experience, I would say that if I evened the playing field by doing the comparison based all comparisons on Bichon's 191K overlapping SNPs, I am very confident his position and MA1's would get closer to Ust. Here is what we are left with:



OUTGROUP
POP1
TARGET
OUTGROUP
D
Z
SNP


Chimp
MA1
Ust_Ishim
Gorilla
-0.9593
-100
205307


Chimp
LaBrana1
Ust_Ishim
Gorilla
-0.9593
-100
271054


Chimp
Bichon
Ust_Ishim
Gorilla
-0.9578
-100
191261


Chimp
Karelia_HG
Ust_Ishim
Gorilla
-0.957
-100
277029


Chimp
Loschbour
Ust_Ishim
Gorilla
-0.9566
-100
284422


Chimp
LBK_EN
Ust_Ishim
Gorilla
-0.9559
-100
286504


Chimp
BedouinB
Ust_Ishim
Gorilla
-0.9555
-100
286939


Chimp
Anatolia_Neolithic
Ust_Ishim
Gorilla
-0.955
-100
284891


Chimp
Yoruba
Ust_Ishim
Gorilla
-0.94
-100
286939



I am not sure about you, but personally I can see a general pattern with regards to the older ones on top (LaBrana excepted, and Karelia marginally excepted. with majority Basal and African on the bottom following a pattern based on amount of Basal or African and age of sample)

EDIT: I have filtered out transition SNPs to enable a more accurate result (I have previously posted the reasons why transitions are more susceptible to errors during sequencing of ancient genomes

Kale
11-29-2015, 05:00 AM
I think Ryukendo's post contains the information needed to satisfy both of our concerns. I think any future double outgroups stats you run though, that are going to be measuring Eurasian differences only, you should use maybe Gorilla and Mota.

Silesian
11-29-2015, 05:05 AM
Teal <> step ancestry or something like ANE. M269 have not any connection with Teal. Its better to look EHG for seeking the ancestry of so called PIE. Jones study should end that "ANE" mess...

Since I'm not the sharpest tool in the shed and like graphics, have a look at this 3d plot posted by OpenGenomes from posting on Eurogenes

Blogger Open Genomes said...

Here is a 3D Global9 PC plot of PC1-PC2-PC3:

Global9 3D PC plot with ancient DNA samples

We can see that it's really critical to show *all* samples, including Africans. The presence of Africans, Oceanians, East Asians, and Native Americans changes the picture completely for Eurasia.

Rather than "compressing" Eurasians, the presence of Africans shows us some fascinating things: The WHG-SHG-EHG group trends downward toward MA-1, who in turn leads to the Inuit and Na-Dene, and distantly to the Native Americans. However, Ust'-Ishim leads off on a separate upper South-Asian / Austroasiatic edge towards a vertex consisting of Japanese and Taiwanese Aboriginals. These in turn on another edge leads downward through Paleo-Siberians (Chukchi, etc.) to Native Americans.

A closer examination of the upper left reveals that the Early Farmers (EF) are nowhere near the Bedouin B, who appear to be admixed with Sub-Saharans, but rather, represent their own separate Eurasian vertex, today only populated by WHG-admixed Sardinians. The Kebaran Levantine Hunter-Gatherers would be even more isolated, beyond KO2 (Starcevo_EN) and the Anatolian Neolithic.

The lesson here is that all ancient and modern genomes need to be plotted together, and only then can we "zoom in" on a particular region of interest, knowing which way the drift is headed. The real "projection bias" (or rather "biased projection" ;) is when certain regions are left out, and and arbitrary 2-dimensional projection leaves out key variation that makes samples appear to be "related" when in fact they are not.

Looks like it cluster's the Levantine/WHG/CHG/ENF IJ-M429 cluster from EHG & R*<MA-1>
http://www.open-genomes.org/images/Global9%20PCA%201-2-3%203D%20plot.png
http://eurogenes.blogspot.ca/

Kurd
11-29-2015, 05:24 AM
I think Ryukendo's post contains the information needed to satisfy both of our concerns. I think any future double outgroups stats you run though, that are going to be measuring Eurasian differences only, you should use maybe Gorilla and Mota.

He is referring to the comparisons many people make using chimp and mbuti as outgroups. I was the one who months ago pointed out that I had observed problems using Mbuti and Chimp together as outgroups, as there appeared to be geneflow between the 2, whereas realistically there should not be. That is why you see me using Gorilla instead of Chimp. My conclusion was that the fixed sites for Mbuti and Eurasians differed, and the use of the Eurasian ascertained panels created this bias.

So when I need one outgroup I use Gorilla, since Chimp seems to affect the results of those with substantial African admixture. However, if I need 2 outgroups I use Gorilla and Chimp. Mota in lieu of chimp would be very problematic because in D(Mota, pop1, pop2, Gorilla) geneflow between Mota and pop2 would selectively affect the results for anyone with substantial African and Basal Eurasian admixture, because of geneflow, whereas those lacking Basal Eurasian and African would not be affected. Using chimp instead of Mota mitigates this problem of geneflow except for the small effect for African admixed individuals due to the ascertainment bias, but this would be far less than any affinity between Mota and pop2, where pop2 is Basal or African admixed

Ryukendo
11-29-2015, 06:02 AM
Since I'm not the sharpest tool in the shed and like graphics, have a look at this 3d plot posted by OpenGenomes from posting on Eurogenes

Blogger Open Genomes said...

Here is a 3D Global9 PC plot of PC1-PC2-PC3:

Global9 3D PC plot with ancient DNA samples

We can see that it's really critical to show *all* samples, including Africans. The presence of Africans, Oceanians, East Asians, and Native Americans changes the picture completely for Eurasia.

Rather than "compressing" Eurasians, the presence of Africans shows us some fascinating things: The WHG-SHG-EHG group trends downward toward MA-1, who in turn leads to the Inuit and Na-Dene, and distantly to the Native Americans. However, Ust'-Ishim leads off on a separate upper South-Asian / Austroasiatic edge towards a vertex consisting of Japanese and Taiwanese Aboriginals. These in turn on another edge leads downward through Paleo-Siberians (Chukchi, etc.) to Native Americans.

A closer examination of the upper left reveals that the Early Farmers (EF) are nowhere near the Bedouin B, who appear to be admixed with Sub-Saharans, but rather, represent their own separate Eurasian vertex, today only populated by WHG-admixed Sardinians. The Kebaran Levantine Hunter-Gatherers would be even more isolated, beyond KO2 (Starcevo_EN) and the Anatolian Neolithic.

The lesson here is that all ancient and modern genomes need to be plotted together, and only then can we "zoom in" on a particular region of interest, knowing which way the drift is headed. The real "projection bias" (or rather "biased projection" ;) is when certain regions are left out, and and arbitrary 2-dimensional projection leaves out key variation that makes samples appear to be "related" when in fact they are not.

Looks like it cluster's the Levantine/WHG/CHG/ENF IJ-M429 cluster from EHG & R*<MA-1>
http://www.open-genomes.org/images/Global9%20PCA%201-2-3%203D%20plot.png
http://eurogenes.blogspot.ca/

Hi Silesian,

Any analysis that uses mostly modern samples, such as this PCA, and many ADMIXTURE runs, will end up parsing the ancients in terms of the moderns, instead of the other way round. E.g. the most diverged populations in that PCA appear to be Japanese, Native Americans, and Sardinians, with ancient genomes a combination of these, but that isn't true.

In the case of that PCA, the algorithm finds it most important to sort out the differences between present-day west eurasians, east asians, and amerinds, as there are so many of these modern samples compared to a few of the ancients only, so W Eurasian vs E Asian vs Amerind takes priority over any other variation. Ancients that contributed to modern Europeans end up clustering around Europeans/W Eurasians no matter how far they may be from each other in the absolute sense.

Hi Kurd,

Looks like the 'phantom chimp' phenomenon has been bugging multiple people. :beerchug:

I noticed from your fst stats that the 'WHG' cluster was around 30% closer to the EEFs than to the E Asians, the EHG cluster was ~20% closer to the EEFs than the E Asians, but the CHG cluster was ~50% closer to the EEFs than to the East Asians. Alone among most ADMIXTURE runs your components behave as though they reflected the relationships of ancient ghost populations, although of course some other stats, such as the E Asian ones, show us that the segregation is not completely clean. Nevertheless, this convinces me that your EEF and CHG clusters are especially basal-rich.

Could you post the rest of the fsts?

Kurd
11-29-2015, 06:32 AM
Hi Silesian,

Any analysis that uses mostly modern samples, such as this PCA, and many ADMIXTURE runs, will end up parsing the ancients in terms of the moderns, instead of the other way round. E.g. the most diverged populations in that PCA appear to be Japanese, Native Americans, and Sardinians, with ancient genomes a combination of these, but that isn't true.

In the case of that PCA, the algorithm finds it most important to sort out the differences between present-day west eurasians, east asians, and amerinds, as there are so many of these modern samples compared to a few of the ancients only, so W Eurasian vs E Asian vs Amerind takes priority over any other variation. Ancients that contributed to modern Europeans end up clustering around Europeans/W Eurasians no matter how far they may be from each other in the absolute sense.

Hi Kurd,

Looks like the 'phantom chimp' phenomenon has been bugging multiple people. :beerchug:

I noticed from your fst stats that the 'WHG' cluster was around 30% closer to the EEFs than to the E Asians, the EHG cluster was ~20% closer to the EEFs than the E Asians, but the CHG cluster was ~50% closer to the EEFs than to the East Asians. Alone among most ADMIXTURE runs your components behave as though they reflected the relationships of ancient ghost populations, although of course some other stats, such as the E Asian ones, show us that the segregation is not completely clean. Nevertheless, this convinces me that your EEF and CHG clusters are especially basal-rich.

Could you post the rest of the fsts?


The following clusters are based on ancients:

1- Anatolian Farmers
2- EHG
3- CHG
4- WHG





Papuan
Neolthic_Anatolian_Farmer
S_Indian
Eastern_Hunter_Gatherer
Kalash
E_Asian
Amerindian
E_African
Caucuses_Hunter_Gatherer
W_African


Neolthic_Anatolian_Farmer
0.222




















S_Indian
0.162
0.115


















Eastern_Hunter_Gatherer
0.231
0.123
0.122
















Kalash
0.198
0.106
0.082
0.112














E_Asian
0.182
0.154
0.087
0.155
0.123












Amerindian
0.289
0.248
0.193
0.234
0.219
0.175










E_African
0.209
0.114
0.11
0.14
0.116
0.145
0.241








Caucuses_Hunter_Gatherer
0.2
0.087
0.089
0.107
0.074
0.129
0.22
0.102






W_African
0.229
0.175
0.144
0.189
0.163
0.172
0.267
0.132
0.154




WHG
0.216
0.109
0.107
0.122
0.101
0.142
0.23
0.114
0.084
0.174

Padre Organtino
11-29-2015, 09:26 AM
You do realize that CHG is a composite? I highly doubt the Basal Eurasian ancestry in CHG is native to the Caucasus or even that all of the CHG present in Non-Caucasians is reflective of in situ migrations, we have still yet to uncover ancient genomes from the lowland valleys. You can't draw conclusions about migrations from one or two genomes, we still need ancient genomes from South-Central Asia or elsewhere from the Middle East before we can deduce who migrated where. Formal stats demonstrated that Satsurblia and Kotias didn't contribute directly to South-Central Asians and if that's the case, then it's likely there was no direct gene flow to Middle Easterners either. Which implies that there may be an alternate source by which CHG in these populations was acquired and it very well could be a population in the lowlands.

Bronze Age Middle East saw a number of invaders not lowlands from various northern directions. As an ethnic Assyrian you probably now that an ethnonym Chaldean actually comes from Urartians who used it as a self-reference. Urartians were hardly the sole Caucasus-Zagross invading tribe, though. Hurrians, Kassites, Guitars, Kaskians, Hatts and others ravaged the Middle East of that time.

ZephyrousMandaru
11-29-2015, 10:35 AM
Bronze Age Middle East saw a number of invaders not lowlands from various northern directions. As an ethnic Assyrian you probably now that an ethnonym Chaldean actually comes from Urartians who used it as a self-reference. Urartians were hardly the sole Caucasus-Zagross invading tribe, though. Hurrians, Kassites, Guitars, Kaskians, Hatts and others ravaged the Middle East of that time.

I don't know about Chaldeans being a derivative of the Urartians, the Chaldeans were a Semitic-speaking group. As I stated previously, J1's presence in Satsurblia strengthens the hypothesis of J1's possible origin being in the Caucasus. However, I still have my doubts about this, mainly because much of CHG harbors a tremendous amount of Basal Eurasian. Something which the neighboring WHG and EHG genomes in the region lack, if CHG is from the Caucasus, why is Basal Eurasian missing in similar amounts in populations proximal to where CHG was discovered?

Something tells me that the answer to this question lies further south.

Padre Organtino
11-29-2015, 10:55 AM
I don't know about Chaldeans being a derivative of the Urartians, the Chaldeans were a Semitic-speaking group. As I stated previously, J1's presence in Satsurblia strengthens the hypothesis of J1's possible origin being in the Caucasus. However, I still have my doubts about this, mainly because much of CHG harbors a tremendous amount of Basal Eurasian. Something which the neighboring WHG and EHG genomes in the region lack, if CHG is from the Caucasus, why is Basal Eurasian missing in similar amounts in populations proximal to where CHG was discovered?

Something tells me that the answer to this question lies further south.

CHG is obviously not "native" to the Caucasus in the strict sense as just like ENF it's a mix of Basal and Crown Eurasians. I simply made a point that given that we see two very distinct pops in ancient Near East it's safe to assume that the two have started mixing strongly not that long ago. Bronze Age invasions of of Euphratus Valley and Fertile Crescent from Zagross and Caucasus look like a good source of CHG-like ancestry in the modern Middle East.

royking
11-30-2015, 01:35 AM
OGF via Ted Kandell did a nice 3D interactive graphic for the world PC1 vs PC2 vs PC3 furnished by Davidski:

http://www.open-genomes.org/analysis/PCA/Eurogenes_Global9_PC_1-2-3_plot_with_aDNA.html

Kurd
11-30-2015, 03:50 AM
Trying to get a handle on the source populations that formed these ancient humans that have been discovered based on the handful of ancient samples we have is not unlike solving a jigsaw puzzle. Case in point CHG. Not only are we limited to a handful of ancients, but to make matters worse only a couple of those predate Satsurbila. We are truly pioneers in this area.

I have decided to take a little time to run various Dstats aimed at piecing together this CHG puzzle. I don't know if I will have enough pieces to figure the puzzle out, but I will give it a try. Each table I post should be considered a clue. Maybe if we put all the clues together we may be in a better position than we are at now, especially since we have quite a few enthusiasts at this forum

Clue #1: CHG is less WHG shifted and more MA1 shifted than the Neolithic farmers and Bedouin.




OUTGROUP
POP1
TARGET
OUTGROUP
D
Z
SNP


MA1
Bichon
Loschbour
Gorilla
-0.2185
-22.447
137538


MA1
Bichon
Iceman
Gorilla
-0.0727
-7.383
138307


MA1
Bichon
LBK_EN
Gorilla
-0.0547
-7.075
138747


MA1
Bichon
Stuttgart
Gorilla
-0.0499
-5.479
136616


MA1
Bichon
Anatolia_Neolithic
Gorilla
-0.043
-5.69
137813


MA1
Bichon
Srubnaya
Gorilla
-0.0311
-3.808
137565


MA1
Bichon
RISE_baAndrov
Gorilla
-0.028
-3.477
138628


MA1
Bichon
BedouinB
Gorilla
-0.0255
-3.482
138824


MA1
Bichon
BedouinA
Gorilla
-0.0229
-3.176
138824


MA1
Bichon
Kostenki14
Gorilla
-0.0204
-2.043
131942


MA1
Bichon
SATSURBILA
Gorilla
-0.0195
-1.691
99620


MA1
Bichon
KOTIAS
Gorilla
-0.019
-1.974
123638


MA1
Bichon
Georgian
Gorilla
-0.0153
-2.067
138824


MA1
Bichon
Scythian_IA
Gorilla
-0.0101
-1.03
129210


MA1
Bichon
RISE_baArm
Gorilla
-0.0045
-0.457
97513


MA1
Bichon
Yamnaya
Gorilla
-0.0012
-0.152
138548


MA1
Bichon
Samara_HG
Gorilla
0.0024
0.211
84502


MA1
Bichon
Ust_Ishim
Gorilla
0.0065
0.712
138571


MA1
Bichon
Kalash
Gorilla
0.0096
1.287
138824


MA1
Bichon
Karelia_HG
Gorilla
0.0148
1.458
135554


MA1
Bichon
Samara_Eneolithic
Gorilla
0.0188
1.893
102628


MA1
Bichon
Nganasan
Gorilla
0.0231
2.962
138824


MA1
Bichon
Pima
Gorilla
0.0556
6.663
138824


MA1
Bichon
Karitiana
Gorilla
0.0655
7.645
138824

Kurd
11-30-2015, 03:54 AM
Table to be updated

Kurd
11-30-2015, 03:58 AM
Table to be updated

K33
11-30-2015, 04:04 AM
@Kurd, not trying to derail you, but could you possibly run the d-stats to see if Saudis have significantly more affinity with the CHG samples than with Neolithic Farmers? Since it now seems the "West Asian" component detected by Dienekes etc is highly correlated with CHG ancestry, my shot-in-the-dark theory was that the curious "Southwest Asian" component picked up on the same admixture model might represent band of hunter-gatherer/herders who migrated from the Caucasus to the Arabian desert and drifted enough to form their own cluster.

Saudis show up with ~ 71% "Southwest Asian" on the dv3 calculator, the highest of any modern pop....

Thanks in advance.

EDIT: I see now that you've done exactly this test above, only with Bedouins. Very interesting! But Bedouins only scored ~41 SW Asian in the admixture model.... could you plz see if the CHG affinity is even more pronounced with Saudis?

Kurd
11-30-2015, 04:05 AM
CLUE: Kotias is more MA1 shifted than Bedouin, Saudi, and slightly more MA1 shifted than Anatolian Farmers. Bichon (WHG) is significantly MA1 shift, indicating that WHG admixture in EF inferred ANE. With Kotias being slightly more MA1 shifted than Anatolian, would Kotias have acquired his ANE through admixture with WHG or through some other event. To be continued tomorrow with more Dstats relating to WHG.



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Karelia_HG
Anatolia_Neolithic
MA1
Gorilla
0.1038
15.122
197895


Bichon
Anatolia_Neolithic
MA1
Gorilla
0.0427
6.335
137813


KOTIAS
BedouinB
MA1
Gorilla
0.032
5.473
181609


KOTIAS
Saudi
MA1
Gorilla
0.027
4.553
181609


Kostenki14
Anatolia_Neolithic
MA1
Gorilla
0.0243
3.355
192439


Georgian
Anatolia_Neolithic
MA1
Gorilla
0.0171
6.007
204242


Iceman
Anatolia_Neolithic
MA1
Gorilla
0.0168
2.722
203051


KOTIAS
Anatolia_Neolithic
MA1
Gorilla
0.0108
1.769
180326


LBK_EN
Anatolia_Neolithic
MA1
Gorilla
0.0041
1.379
204003


Stuttgart
Anatolia_Neolithic
MA1
Gorilla
0.0003
0.054
200938


Saudi
Anatolia_Neolithic
MA1
Gorilla
-0.0167
-5.837
204242


BedouinB
Anatolia_Neolithic
MA1
Gorilla
-0.0223
-8.08
204242


BedouinA
Anatolia_Neolithic
MA1
Gorilla
-0.03
-12.761
204242


Ust_Ishim
Anatolia_Neolithic
MA1
Gorilla
-0.0351
-5.23
203850


Hadza
Anatolia_Neolithic
MA1
Gorilla
-0.2728
-62.853
204238


Yoruba
Anatolia_Neolithic
MA1
Gorilla
-0.2851
-78.64
204242

Kurd
11-30-2015, 04:31 AM
Most similar to E African Hadza on top



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Iceman
Anatolia_Neolithic
Hadza
Gorilla
0.0208
4.745
283568


LBK_EN
Anatolia_Neolithic
Hadza
Gorilla
0.0047
2.474
285060


BedouinB
Anatolia_Neolithic
Hadza
Gorilla
0.0045
2.401
285459


Saudi
Anatolia_Neolithic
Hadza
Gorilla
0.0039
2.061
285459


Georgian
Anatolia_Neolithic
Hadza
Gorilla
0.0036
2.013
285459


KOTIAS
Anatolia_Neolithic
Hadza
Gorilla
0.001
0.265
251180


Stuttgart
Anatolia_Neolithic
Hadza
Gorilla
0.001
0.296
280877


BedouinA
Anatolia_Neolithic
Hadza
Gorilla
-0.0003
-0.179
285459


Bichon
Anatolia_Neolithic
Hadza
Gorilla
-0.0034
-0.755
190259


Karelia_HG
Anatolia_Neolithic
Hadza
Gorilla
-0.0067
-1.448
275643


Kostenki14
Anatolia_Neolithic
Hadza
Gorilla
-0.0095
-2.002
268308


MA1
Anatolia_Neolithic
Hadza
Gorilla
-0.012
-2.362
204238


Ust_Ishim
Anatolia_Neolithic
Hadza
Gorilla
-0.0239
-5.723
284886


Yoruba
Anatolia_Neolithic
Hadza
Gorilla
-0.0699
-26.626
285459

Kurd
11-30-2015, 04:40 AM
@Kurd, not trying to derail you, but could you possibly run the d-stats to see if Saudis have significantly more affinity with the CHG samples than with Neolithic Farmers? Since it now seems the "West Asian" component detected by Dienekes etc is highly correlated with CHG ancestry, my shot-in-the-dark theory was that the curious "Southwest Asian" component picked up on the same admixture model might represent band of hunter-gatherer/herders who migrated from the Caucasus to the Arabian desert and drifted enough to form their own cluster.

Saudis show up with ~ 71% "Southwest Asian" on the dv3 calculator, the highest of any modern pop....

Thanks in advance.

EDIT: I see now that you've done exactly this test above, only with Bedouins. Very interesting! But Bedouins only scored ~41 SW Asian in the admixture model.... could you plz see if the CHG affinity is even more pronounced with Saudis?




Edit: Tables to be updated

Kurd
11-30-2015, 04:57 AM
I know that there are not enough clues yet, but does anyone have any ideas yet?

What about perhaps a proto Bichon population mixing with the Basal Eurasians in the Caucuses to form CHG. Could perhaps explain WHG affinity as well as ANE affinity(just guessing at this point)

VOX
11-30-2015, 05:13 AM
Just speculation but the mitochondrial DNA haplogroups, M1 and U6 is seen as being a back migration to Africa from the Middle East in published literature. Given what we know about CHG, it might not be a stretch if the M1/U6 people carried an autosomal signature partially related to CHG.

Ryukendo
11-30-2015, 05:25 AM
Hmmm Kurd, I understand that using the two outgroups creates an f3-like effect, but why do you use the methodology you use here? Surely it would be more profitable to compare the genomes directly?

E.g. for whether or not Africans have an attraction to Kotias

Primate_Gorilla Hadza LBK_EN Kotias -0.007 -1.707 271786

Here, Hadza are if anything closer to LBK_EN, though the comparison is not significant.

The statistic is almost significant/significant for other Africans though:
Primate_Gorilla Mbuti LBK_EN Kotias -0.0072 -1.896 271793
Primate_Gorilla Somali LBK_EN Kotias -0.025 -6.813 271793


For west Eurasians, we have the comparisons

Chimp Kotias Kostenki14 Loschbour 0.0263 3.598 343553
Chimp Kotias Kostenki14 Karelia_HG 0.0218 2.528 219554
Chimp Kotias Kostenki14 MA1 0.0055 0.708 271405

There are many others which David has run at the community's behest which compare and contrast the samples directly.

DMXX
11-30-2015, 05:33 AM
I know that there are not enough clues yet, but does anyone have any ideas yet?

What about perhaps a proto Bichon hybrid WHG/ANE population mixing with the Basal Eurasians in the Caucuses to form CHG. Could perhaps explain WHG affinity as well as ANE affinity. Maybe this hybrid population is yet to be discovered, went extinct, or somehow got absorbed (just guessing at this point)

Also, what could explain the contrast in affinity to Africans between CHG and Anatolian Farmers?

Assuming the stats hold up, one possible scenario...

1. EEF and CHG are both heavy in the residual "Basal Eurasian" (perhaps we should term it "Residual Eurasian" in light of that? I'll defer to Ryu for consideration if it's valid) that may have remained in the Near-East after the major Eurasian clades formed and separated via hunter-gatherers.
2. However, CHG contains more "Basal Eurasian" than EEF. This accounts to the closer affinity to Africa in CHG versus EEF. Let us recall the paper from earlier this year that found West Eurasian-like ancestry across Africa via a prehistoric back-migration. That could well be from "Residual Eurasian".
3. CHG contains some concealed ANE which is absent in EEF.
4. CHG's affinity to WHG is a by-product of its' concealed ANE (WHG-EHG-ANE forms a macro-clade, does it not?).

In sum, CHG would be more "Basal/Residual Eurasian" than EEF, with actual concealed ANE, with the African affinity being due to the previously found old West Eurasian admix purportedly found in Africa.

Feel free to appraise the scenario as required, ladies and gents.

parasar
11-30-2015, 05:36 AM
I know that there are not enough clues yet, but does anyone have any ideas yet?

What about perhaps a proto Bichon hybrid WHG/ANE population mixing with the Basal Eurasians in the Caucuses to form CHG. Could perhaps explain WHG affinity as well as ANE affinity. Maybe this hybrid population is yet to be discovered, went extinct, or somehow got absorbed (just guessing at this point)

Also, what could explain the contrast in affinity to Africans between CHG and Anatolian Farmers?

To be sure we need more ancient DNA.
My own thinking was that the population movement was from parts of Europe to western Asia and the Mediterranean where there was contact with an African type population.

VOX mentioned M1, for which there is this strange disconnect - M1's closest relative M20 and M51 are found in E Asia. So it not clear where M1’20’51 originated.

Kurd
11-30-2015, 06:01 AM
Hmmm Kurd, I understand that using the two outgroups creates an f3-like effect, but why do you use the methodology you use here? Surely it would be more profitable to compare the genomes directly?

E.g. for whether or not Africans have an attraction to Kotias

Primate_Gorilla Hadza LBK_EN Kotias -0.007 -1.707 271786

Here, Hadza are if anything closer to LBK_EN, though the comparison is not significant.

The statistic is almost significant/significant for other Africans though:
Primate_Gorilla Mbuti LBK_EN Kotias -0.0072 -1.896 271793
Primate_Gorilla Somali LBK_EN Kotias -0.025 -6.813 271793


For west Eurasians, we have the comparisons

Chimp Kotias Kostenki14 Loschbour 0.0263 3.598 343553
Chimp Kotias Kostenki14 Karelia_HG 0.0218 2.528 219554
Chimp Kotias Kostenki14 MA1 0.0055 0.708 271405

There are many others which David has run at the community's behest which compare and contrast the samples directly.

Just in the interest of saving time, and making a quick ranking of various populations, otherwise, I also do direct comparisons like the table I posted on the previous page comparing Bichon and MA1 to Kotias. Thanks for pointing out the discrepency though. I just ran it myself and got

Kotias LBK_EN Hadza Gorilla -0.0043 -1.070 11036 11131 252643

The slight difference being attributed to me filtering out transition SNPs

Theoretically there should not be a problem in running with 2 outgroups, but apparently some phenomena is skewing the result. I guess no more short-cuts till I can figure out if there is something with the selection of the outgroup or whatever else is causing it. I'll see if Nick has any thoughts on it.

I will re-run with only 1 outgroup the ones previously ran with 2 outgroups

Ryukendo
11-30-2015, 06:18 AM
Hi Kurd,

Yeah, a calculation showed La Brana having more shared drift with Bedouin than Anatolia_EN, and this will almost certainly be falsified in a direct comparison

Gorilla Bedouin Anatolia_EN La_Brana

Likewise, the 'f3-like' stats show that Kotias has more absolute affinity with WHG than European Neolithic Farmers, when a f3 shared drift stat from the paper's fig 2 already contradicts this head-on.

So sorry to throw a spanner into the works, but I really suspect that comparisons of the type

Outgroup1, Ancient1, Ancient2, Outgroup2

Where the outgroups are non-human are all that reliable. My suspicion is that branch length is having an effect, and, given lists of comparisons with large numbers of ancient populations for each calculation with different branch lengths, we cannot correct for this.

Thank you for updating your priors so rapidly.

Kurd
11-30-2015, 06:45 AM
The comparisons to Yoruba are updated using only 1 outgroup. Iceman and modern SW Asians show more affinity to Yoruba than Anatolian Farmers. Kotias shows slightly more affinity than Anatolian to Yoruba, but the stat is not significant. The Anatolians however, show more affinity to W African Yoruban than E African Hadza does



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Iceman
Anatolia_Neolithic
Yoruba
Gorilla
0.0244
6.144
283573


BedouinB
Anatolia_Neolithic
Yoruba
Gorilla
0.0084
5.138
285464


BedouinA
Anatolia_Neolithic
Yoruba
Gorilla
0.0082
5.831
285464


Saudi
Anatolia_Neolithic
Yoruba
Gorilla
0.0075
4.345
285464


Stuttgart
Anatolia_Neolithic
Yoruba
Gorilla
0.0055
1.761
280882


LBK_EN
Anatolia_Neolithic
Yoruba
Gorilla
0.0044
2.446
285065


KOTIAS
Anatolia_Neolithic
Yoruba
Gorilla
0.0016
0.456
251185


Bichon
Anatolia_Neolithic
Yoruba
Gorilla
0.0015
0.381
190263


Somali
Anatolia_Neolithic
Yoruba
Gorilla
0.0011
0.57
285464


Kostenki14
Anatolia_Neolithic
Yoruba
Gorilla
-0.0063
-1.478
268313


MA1
Anatolia_Neolithic
Yoruba
Gorilla
-0.0065
-1.451
204242


Ust_Ishim
Anatolia_Neolithic
Yoruba
Gorilla
-0.0211
-5.547
284891


Hadza
Anatolia_Neolithic
Yoruba
Gorilla
-0.0465
-16.586
285459

Ryukendo
11-30-2015, 09:42 AM
Wow Kurd! O_o

Hadza generally breaks off after Yoruba in Treemix, so this closeness of Anatolia Neolithic to Yoruba to the exclusion of Hadza is an impressive finding. Let's see if David can replicate this stat itself, as well as reproducing the signal in Treemix.

Kurd
11-30-2015, 02:06 PM
Wow Kurd! O_o

Hadza generally breaks off after Yoruba in Treemix, so this closeness of Anatolia Neolithic to Yoruba to the exclusion of Hadza is an impressive finding. Let's see if David can replicate this stat itself, as well as reproducing the signal in Treemix.

Although I am not as surprised since the Hadza appear to be an isolate. Maybe they have been for a very long time, but again, did they branch off that far back, or was the impact of back to Africa migrators so much greater on Yoruba than on Hadza? I guess this raises a few questions

ZephyrousMandaru
11-30-2015, 02:12 PM
Hmm, I'm not so sure that prehistoric back-migrations would necessarily explain this affinity of Anatolian and CHG genomes with Africans. The non-African portion of ancestry in Africans is about 23,000 years old, several thousands of years older than Kotias and Satsurblia. I think that the Eurasian-like segments that characterizes African genetic diversity may not necessarily be an indication of direct gene flow from Eurasians, but that this part of their ancestry appears to be so "uncharacteristically" African that it seems Eurasian. And that it exhibits a close genetic kinship to populations which possess a significant amount of Basal Eurasian ancestry.

I think we need to be cautious about making overreaching inferences based off of ancient genomes, much of the Middle East remains severely understudied.

ren
11-30-2015, 02:19 PM
Begining with the Raghavan paper, the model has become more and more elaborate to include ever more hypothetical populations and hypothetical scenarios. Now you have 3 Basal populations that are less related to the European-East Eurasian population: Basal Eurasian, Kostenki, and CHG. The far easier explanation is that WHG-EHG and Mal'ta are just admixed with a proto-Native American population and Kostenki, SW Asians, CHG are pure W. Eurasian populations.

No mental gymnastics and flowering flow charts are needed in this explanation and it does the samething mathematically.

Kurd
11-30-2015, 02:32 PM
Hmm, I'm not so sure that prehistoric back-migrations would necessarily explain this affinity of Anatolian and CHG genomes with Africans. The non-African portion of ancestry in Africans is about 23,000 years old, several thousands of years older than Kotias and Satsurblia. I think that the Eurasian-like segments that characterizes African genetic diversity may not necessarily be an indication of direct gene flow from Eurasians, but that this part of their ancestry appears to be so "uncharacteristically" African that it seems Eurasian. And that it exhibits a close genetic kinship to populations which possess a significant amount of Basal Eurasian ancestry.

I think we need to be cautious about making overreaching inferences based off of ancient genomes, much of the Middle East remains severely understudied.


Actually, you got this backwards. If some of NAF's Basal ancestors returned to Africa, and mixed with other African groups who mixed with Yoruba's ancestors, some of those Basal Eurasians alleles would get incorporated into Yoruba, so when a Dstat allele by allele comparison is made between NAF (descendants of those Basal Eurasians) and Yorubans, there would be similarities shared between Yorubans and NAF to the exclusion of Hadza (assuming the back to Africa migrators did not contribute to them, or they contributed less to Hadza ancestors than Yoruba ancestors) at certain loci. These similarites would of course also exist if Yorubans ancestors contibuted to the Out of Africa groups ancestral to NAF.

ZephyrousMandaru
11-30-2015, 03:01 PM
Actually, you got this backwards. If some of NAF's Basal ancestors returned to Africa, and mixed with other African groups who mixed with Yoruba's ancestors, some of those Basal Eurasians alleles would get incorporated into Yoruba, so when a Dstat allele by allele comparison is made between NAF (descendants of those Basal Eurasians) and Yorubans, there would be similarities shared between Yorubans and NAF to the exclusion of Hadza (assuming the back to Africa migrators did not contribute to them, or they contributed less to Hadza ancestors than Yoruba ancestors) at certain loci. These similarites would of course also exist if Yorubans ancestors contibuted to the Out of Africa groups ancestral to NAF.

But if this back-migration occurred 2,700-3,300 years ago, wouldn't this present a temporal discrepancy between Basal Eurasian ancestry that is contributed relatively recently and the age of that ancestry itself? I think there would be a difference between the antiquity of Basal Eurasian as an ancestral population, and when some of their descendants expanded. I think we'd need some more ancient genomes from Africa in order to determine exactly what Basal Eurasian is.

Kurd
11-30-2015, 03:11 PM
But if this back-migration occurred 2,700-3,300 years ago, wouldn't this present a temporal discrepancy between Basal Eurasian ancestry that is contributed relatively recently and the age of that ancestry itself? I think there would be a difference between the antiquity of Basal Eurasian as an ancestral population, and when some of their descendants expanded. I think we'd need some more ancient genomes from Africa in order to determine exactly what Basal Eurasian is.

I don't follow :confused: where did I write anything about a migration occuring 2700-3300 years ago?

alan
11-30-2015, 05:05 PM
I enjoy endlessly speculating on things but I feel some of the questions people are trying to answer on the earliest west Eurasian dispersals c. 45000-30000BC simply cannot be done due to lack of sufficient Palaeolithic datapoints in terms of ancient DNA. The upper Palaeolithic of western/central Eurasia is over 35000 years in duration and includes some incredible displacements caused by climatic oscillation. Goes without saying Eurasia is vast too. We barely have a handful of datapoints for the whole of Eurasia. The sample is so small for this early period I would still feel archaeology is more useful than genetics at this point in time. That will change as more samples come in but we need vastly more samples to cover that huge time/space. There are some staggering gaps in the sampling. Far too many hunter samples are very late Mesolithic ones. AFAIK we dont have any European Gravettian ancient yDNA samples at all.

ZephyrousMandaru
11-30-2015, 11:10 PM
I don't follow :confused: where did I write anything about a migration occuring 2700-3300 years ago?

I am referring to the back-migration to Africa that occurred around that time period. Here's a link to the article.

https://www.newscientist.com/article/dn24988-humanitys-forgotten-return-to-africa-revealed-in-dna/#.UvBVqfmSw9Y

Kurd
12-01-2015, 12:55 AM
With regards to using 2 outgroups in Dstats, I just received the following message from Nick Patterson:

"Ancient DNA with multiple outgroups has its problems, especially with transitions". Nick also seems to be of the oppinion that transition SNPs can be problematic with ancient DNA, precisly why I try to filter them in my analysis.

ZephyrousMandaru
12-01-2015, 01:56 AM
Kurd, upon testing with your latest calculator, I noticed some rather large differences in terms of what different East Assyrian subgroups such as Nestorians and Chaldeans score for CHG. I myself registered around 51%, whereas most of the Chaldeans scored around 40-45%. A 5-10% difference, is this a function of the number of SNPs use or is it something else?

Kurd
12-01-2015, 02:24 AM
Kurd, upon testing with your latest calculator, I noticed some rather large differences in terms of what different East Assyrian subgroups such as Nestorians and Chaldeans score for CHG. I myself registered around 51%, whereas most of the Chaldeans scored around 40-45%. A 5-10% difference, is this a function of the number of SNPs use or is it something else?

I think this is within the normal variation between Chaldeans Nestorians and Assyrians. Also the calculator sources on transversion SNPs, which mutate at half the rate of transitions

Kurd
12-01-2015, 04:09 AM
Now we have the Dstat outgroup issue squared away, we can forge ahead with more clues to help solve CHG's ancestry

CLUE: Whereas the farmers are more similar to WHG than EHG, Kotias is positioned half way between EHG and WHG, with a slight EHG preference. Other interesting observations are Kalash more similar to EHG than WHG. Yoruba and Somalis are more similar to WHG than EHG



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Karelia_HG
Bichon
Samara_Eneolithic
Gorilla
0.0676
7.69
139314


Karelia_HG
Bichon
Karitiana
Gorilla
0.0621
8.425
186747


Karelia_HG
Bichon
MA1
Gorilla
0.061
6.292
135554


Karelia_HG
Bichon
Samara_HG
Gorilla
0.0551
5.307
114802


Karelia_HG
Bichon
Pima
Gorilla
0.0551
7.593
186747


Karelia_HG
Bichon
Yamnaya
Gorilla
0.0364
5.402
186604


Karelia_HG
Bichon
Nganasan
Gorilla
0.0309
4.464
186747


Karelia_HG
Bichon
Kalash
Gorilla
0.0203
3.167
186747


Karelia_HG
Bichon
RISE_baArm
Gorilla
0.0128
1.511
130886


Karelia_HG
Bichon
Scythian_IA
Gorilla
0.0117
1.281
175115


Karelia_HG
Bichon
Srubnaya
Gorilla
0.0081
1.199
185267


Karelia_HG
Bichon
RISE_baAndrov
Gorilla
0.008
1.133
186489


Karelia_HG
Bichon
Georgian
Gorilla
0.0039
0.609
186747


Karelia_HG
Bichon
KOTIAS
Gorilla
0.0036
0.43
166214


Karelia_HG
Bichon
Yoruba
Gorilla
-0.0083
-1.383
186747


Karelia_HG
Bichon
BedouinB
Gorilla
-0.0086
-1.325
186747


Karelia_HG
Bichon
Ust_Ishim
Gorilla
-0.0095
-1.167
186378


Karelia_HG
Bichon
Somali
Gorilla
-0.0124
-2.074
186747


Karelia_HG
Bichon
Anatolia_Neolithic
Gorilla
-0.0148
-2.249
185418


Karelia_HG
Bichon
LBK_EN
Gorilla
-0.0177
-2.605
186723


Karelia_HG
Bichon
Kostenki14
Gorilla
-0.0247
-2.738
177434


Karelia_HG
Bichon
Stuttgart
Gorilla
-0.0281
-3.513
183784


Karelia_HG
Bichon
Iceman
Gorilla
-0.0421
-4.65
186068


Karelia_HG
Bichon
LaBrana1
Gorilla
-0.1165
-13.025
179555


Karelia_HG
Bichon
Loschbour
Gorilla
-0.1669
-18.809
185033

gihanga.rwanda
12-01-2015, 04:22 AM
Interesting results, but the Hadze are generally closer to OOA than the Yoruba, even despite the fact that a portion of their ancestry derives from a divergent hunter-gatherer population with affinities to either Mbuti or Ju'hoansi; the remainder of their ancestry is some sort of East African, possibly similar to the ancient Ethiopian Mota genome. Is it possible that the Hadze's divergent AHG (African Hunter-Gatherer) ancestry is skewing the results?

Also, I am not sure if you saw these Treemix results on Eurogenes...

http://eurogenes.blogspot.com/2015/11/treemix-graphs-with-ancient-ethiopian.html?m=1

But ~25% of the Mota genome's ancestry seems to derive from an ancient African population at the root of the AMH lineage, more divergent than even Ju'hoansi. This is interesting since Mota is already closer to the OOA than all other non-East African populations, even Dinka. Polako also detected a migration edge from Mota into CHG, which might say something about "Basal Eurasian," which has only been described as "Mbuti>X>Crown Eurasian." Also notice that it appears that the Mbuti and Biaka appear to possess minor at ~2% admixture from an archaic African population.

IMHO Basal Eurasian is either North African or ancient Arabian.

gihanga.rwanda
12-01-2015, 04:30 AM
That study detected long suspected East African (i.e. South Cushitic) ancestry in Khoe-speaking pastoralist communities in Southern Africa, who are known to have migrated to the region from the area of Kenya/Tanzania, just south of the Horn, several thousand years ago. The Khoe are also suspected to be related to Sandawe, who are also known to possess ancestry from both an ancient AHG population and East African pastoralists. E1b1b/E-M293 is one of the most frequent yDNA lineage among the Khoe-Sandawe (upwards of ~50%), a proposed linguistic phylum. The hunter-gatherer San, such as the Ju'hoansi and Xun, generally lack such ancestry.

Kurd
12-01-2015, 04:39 AM
Farmers prefer Samara very slightly over Karelian, Kotias however prefers Karelian over Samara. Kotias is again less WHG shifted than the farmers.



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Samara_HG
Bichon
Karelia_HG
Gorilla
0.0813
4.097
21443


Samara_HG
Bichon
Samara_Eneolithic
Gorilla
0.0568
2.599
16836


Samara_HG
Bichon
Karitiana
Gorilla
0.0449
2.709
21746


Samara_HG
Bichon
Pima
Gorilla
0.0327
1.937
21746


Samara_HG
Bichon
Yamnaya
Gorilla
0.0267
1.678
21738


Samara_HG
Bichon
MA1
Gorilla
0.0238
0.973
15819


Samara_HG
Bichon
Nganasan
Gorilla
0.0115
0.697
21746


Samara_HG
Bichon
Kalash
Gorilla
-0.0014
-0.091
21746


Samara_HG
Bichon
Georgian
Gorilla
-0.0084
-0.554
21746


Samara_HG
Bichon
Scythian_IA
Gorilla
-0.0095
-0.428
20703


Samara_HG
Bichon
RISE_baArm
Gorilla
-0.0097
-0.427
15265


Samara_HG
Bichon
Srubnaya
Gorilla
-0.0112
-0.694
21543


Samara_HG
Bichon
Yoruba
Gorilla
-0.0113
-0.734
21746


Samara_HG
Bichon
KOTIAS
Gorilla
-0.012
-0.619
20482


Samara_HG
Bichon
RISE_baAndrov
Gorilla
-0.012
-0.708
21706


Samara_HG
Bichon
BedouinB
Gorilla
-0.0199
-1.271
21746


Samara_HG
Bichon
Somali
Gorilla
-0.0202
-1.34
21746


Samara_HG
Bichon
Ust_Ishim
Gorilla
-0.0234
-1.276
21733


Samara_HG
Bichon
Anatolia_Neolithic
Gorilla
-0.0322
-2.028
21551


Samara_HG
Bichon
Stuttgart
Gorilla
-0.0323
-1.815
21546


Samara_HG
Bichon
Kostenki14
Gorilla
-0.0371
-1.808
20657


Samara_HG
Bichon
LBK_EN
Gorilla
-0.0423
-2.568
21743


Samara_HG
Bichon
Iceman
Gorilla
-0.0628
-3.11
21667


Samara_HG
Bichon
LaBrana1
Gorilla
-0.1337
-6.437
20969


Samara_HG
Bichon
Loschbour
Gorilla
-0.1932
-10.62
21627

Ryukendo
12-01-2015, 07:58 AM
With regards to using 2 outgroups in Dstats, I just received the following message from Nick Patterson:

"Ancient DNA with multiple outgroups has its problems, especially with transitions". Nick also seems to be of the oppinion that transition SNPs can be problematic with ancient DNA, precisly why I try to filter them in my analysis.

Thank you for clarifying this, Kurd.

MfA
12-01-2015, 10:29 AM
Kurd, upon testing with your latest calculator, I noticed some rather large differences in terms of what different East Assyrian subgroups such as Nestorians and Chaldeans score for CHG. I myself registered around 51%, whereas most of the Chaldeans scored around 40-45%. A 5-10% difference, is this a function of the number of SNPs use or is it something else?

It's simple; you are part of the original run and they are not.

Kurd
12-01-2015, 12:09 PM
It's simple; you are part of the original run and they are not.

The effect you are referring to is what is known as the calculator effect. The Eurasia 11 calculator, however, has an almost 0 calculator effect. You can check for yourself. Here are the "run" results of some of the project members. You can compare them to what they posted as their calculator results. You should notice almost no difference.



IND
ID
Papuan
Neolthic Anatolian
S_Indian
Eastern Hunter Gatherer
Kalash
E_Asian
Amerindian
E_African
Caucuses Hunter Gatherer
W_African
WHG


.Awale
ID001
0.53%
0.00%
0.44%
0.00%
0.00%
0.00%
0.00%
62.09%
0.00%
35.40%
1.53%


.Bol_Nat
ID001
1.56%
2.81%
42.02%
10.58%
6.05%
0.43%
0.00%
0.00%
34.57%
0.00%
1.98%


.Bored
ID001
0.79%
0.00%
42.99%
8.60%
6.23%
3.70%
1.33%
0.00%
28.38%
0.00%
7.98%


.Dluffy
ID001
0.74%
0.00%
43.65%
8.06%
6.47%
0.48%
0.89%
0.00%
32.70%
0.40%
6.61%


.DMXX
ID001
2.10%
16.39%
2.69%
6.81%
5.40%
5.87%
0.52%
9.23%
46.54%
0.89%
3.54%


.Farid
ID001
0.00%
1.88%
21.06%
8.76%
8.16%
0.00%
0.00%
9.69%
47.63%
2.82%
0.00%


.Hanna
ID001
0.00%
24.83%
0.00%
1.90%
2.12%
5.93%
1.52%
5.57%
48.59%
0.00%
9.53%


.Jesus
ID001
0.84%
11.68%
9.16%
5.40%
13.80%
0.36%
3.36%
10.69%
39.30%
3.38%
2.03%


.Kandhari
ID001
1.61%
4.38%
20.65%
11.85%
1.06%
4.03%
1.80%
1.09%
45.82%
0.00%
7.70%


.Kenji
ID001
0.72%
2.17%
39.80%
10.20%
11.21%
2.24%
2.24%
2.38%
25.23%
0.00%
3.82%


.Khana
ID001
0.71%
5.16%
35.70%
9.63%
6.84%
4.06%
1.74%
2.22%
31.67%
0.00%
2.26%


.Passa
ID001
0.58%
30.42%
0.00%
3.14%
1.44%
0.00%
0.00%
8.15%
29.85%
0.00%
26.43%


.Rukha
ID001
1.00%
4.46%
20.21%
13.36%
7.14%
1.25%
4.22%
0.00%
39.14%
1.78%
7.44%


.Sein
ID001
0.00%
4.17%
30.80%
9.91%
11.08%
4.21%
0.87%
0.00%
33.98%
0.60%
4.38%


.Varun
ID001
0.00%
0.39%
52.70%
2.47%
9.55%
4.30%
0.00%
0.00%
26.11%
0.00%
4.48%


.Zara
ID001
0.00%
7.11%
15.87%
6.96%
9.56%
0.00%
0.37%
7.08%
49.19%
3.85%
0.00%


.Zephyrous
ID001
1.53%
24.91%
2.42%
3.35%
0.00%
0.89%
0.00%
12.86%
51.31%
0.00%
2.73%

gihanga.rwanda
12-01-2015, 12:22 PM
That study detected long suspected East African (i.e. South Cushitic) ancestry in Khoe-speaking pastoralist communities in Southern Africa, who are known to have migrated to the region from the area of Kenya/Tanzania, just south of the Horn, several thousand years ago. The Khoe are also suspected to be related to Sandawe, who are also known to possess ancestry from both an ancient AHG population and East African pastoralists. E1b1b/E-M293 is one of the most frequent yDNA lineage among the Khoe-Sandawe (upwards of ~50%), a proposed linguistic phylum. The hunter-gatherer San, such as the Ju'hoansi and Xun, generally lack such ancestry.

@ZephyrousMandaru

ZephyrousMandaru
12-01-2015, 12:38 PM
It's simple; you are part of the original run and they are not.

I'm not so sure about that, in the original K14 run I scored around 33% for CHG. 27% less than in Eurasia K11 CHG-NAF.

Edit: Never mind, Kurd posted the non-calculator values for it, it doesn't appear that the calculator effect is the culprit here. So there must be something else that's causing these differences.

Kurd
12-01-2015, 01:14 PM
CLUE: Kotias CHG is more ANE shifted than Anatolian Farmers, even though he predates them by about 1500 years. Notice the increase in ANE shift from Satsurbila (13kya) to Kotias (9.7kya). I believe that if the comparison were to be made with Anatolian Farmers of Kotias's time, if they existed, the ANE shift would have been even greater (speculation). K14 is substantially more WHG shifted than Kotias in D(Karelia, Bichon, K14, Gorilla), and marginally more ANE shifted than Kotias in this table



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Anatolia_Neolithic
Karelia_HG
MA1
Gorilla
-0.1128
-8.796
36617


Anatolia_Neolithic
Samara_Eneolithic
MA1
Gorilla
-0.0857
-6.235
27218


Anatolia_Neolithic
Yamnaya
MA1
Gorilla
-0.0714
-9.037
37723


Anatolia_Neolithic
Scythian_IA
MA1
Gorilla
-0.0626
-4.604
35307


Anatolia_Neolithic
Karitiana
MA1
Gorilla
-0.0581
-6.259
37851


Anatolia_Neolithic
Bichon
MA1
Gorilla
-0.0551
-4.336
25954


Anatolia_Neolithic
RISE_baAndrov
MA1
Gorilla
-0.0532
-6.552
37739


Anatolia_Neolithic
Pima
MA1
Gorilla
-0.0531
-6.464
37851


Anatolia_Neolithic
LaBrana1
MA1
Gorilla
-0.0501
-3.926
35835


Anatolia_Neolithic
Loschbour
MA1
Gorilla
-0.0485
-4.353
37642


Anatolia_Neolithic
Srubnaya
MA1
Gorilla
-0.0479
-6.59
37719


Anatolia_Neolithic
Kostenki14
MA1
Gorilla
-0.0314
-2.37
35480


Anatolia_Neolithic
RISE_baArm
MA1
Gorilla
-0.031
-2.217
25120


Anatolia_Neolithic
Kalash
MA1
Gorilla
-0.027
-4.373
37851


Anatolia_Neolithic
KOTIAS
MA1
Gorilla
-0.0206
-1.822
35030


Anatolia_Neolithic
Georgian
MA1
Gorilla
-0.018
-3.278
37851


Anatolia_Neolithic
Nganasan
MA1
Gorilla
-0.0114
-1.444
37851


Anatolia_Neolithic
Stuttgart
MA1
Gorilla
-0.0084
-0.855
37517


Anatolia_Neolithic
Iceman
MA1
Gorilla
-0.0066
-0.54
37614


Anatolia_Neolithic
LBK_EN
MA1
Gorilla
-0.0017
-0.271
37797


Anatolia_Neolithic
SATSURBILA
MA1
Gorilla
0.0097
0.654
27818


Anatolia_Neolithic
BedouinB
MA1
Gorilla
0.0265
4.843
37851


Anatolia_Neolithic
Ust_Ishim
MA1
Gorilla
0.0266
2.588
37830


Anatolia_Neolithic
Somali
MA1
Gorilla
0.1773
31.211
37851


Anatolia_Neolithic
Yoruba
MA1
Gorilla
0.2886
45.552
37851

Padre Organtino
12-01-2015, 01:25 PM
Continuing the topic of Assyrians and CHG: Urartians did call themselves something like Khaldini as their main deity was named Khaldi. It does look like there was an active genetic exchange between Urartians and Assyrians as the latter are the most "Caucasus-like" Semitic group and the former have become more "Southern" than Georgians genetically while being much alike North Caucasians in the Bronze Age.

parasar
12-01-2015, 03:18 PM
Kotias CHG is more ANE shifted than Anatolian Farmers, even though he predates them by about 1500 years. Notice the increase in ANE shift from Satsurbila (13kya) to Kotias (9.7kya). I believe that if the comparison were to be made with Anatolian Farmers of Kotias's time, if they existed, the ANE shift would have been even greater (speculation). K14 is substantially more WHG shifted than Kotias in D(Karelia, Bichon, K14, Gorilla), and marginally more ANE shifted than Kotias in this table...

This is seen vis a vis Stuttgart too:
http://www.anthrogenica.com/showthread.php?5833-Teal-discovered-!!&p=121497&viewfull=1#post121497

slight gradation towards ANE from Stuttgart to Sarsurblia to Kotias...

D(Yoruba,MA1;Satsurblia,Stuttgart)= D:-0.0096, Z:-1.423
D(Yoruba,MA1;Kotias,Stuttgart) = D:-0.0153, Z:-2.380


Do you think this is due the Ancient North Eurasian in MA1/ANE or the non-Ancient North Eurasian portion in MA1/ANE or the Ancient North Eurasian in Karelian and Karitiana? One way to check would be by comparing the Neolithics and CHG to the Karitiana since a branch of Ancient North Eurasian went into both the Karitiana and Karelians and another into MA1/ANE. (Comparing with Karelian will cause a problem since both the Neolithics and Karelians have affinity via WHG related ancestry.)

"ANE ancestry in Karelia_HG is derived from the branch of “Ancient_North_Eurasian” that goes into the Karitiana Native Americans, rather
than the MA1 branch ... Karelia_HG share more alleles with Native Americans due to sharing additional common genetic drift (and thus Native Americans should share more alleles with Karelia_HG than with MA1) ... Karelia_HG has WHG-related ancestry which dilutes this affinity to Native Americans ... (EHG, WHG, and ANE) cannot be related to each other by a simple tree, and at least one of them must be admixed ..."

Kurd
12-01-2015, 04:48 PM
This is seen vis a vis Stuttgart too:
http://www.anthrogenica.com/showthread.php?5833-Teal-discovered-!!&p=121497&viewfull=1#post121497


Do you think this is due the Ancient North Eurasian in MA1/ANE or the non-Ancient North Eurasian portion in MA1/ANE or the Ancient North Eurasian in Karelian and Karitiana? One way to check would be by comparing the Neolithics and CHG to the Karitiana since a branch of Ancient North Eurasian went into both the Karitiana and Karelians and another into MA1/ANE. (Comparing with Karelian will cause a problem since both the Neolithics and Karelians have affinity via WHG related ancestry.)

"ANE ancestry in Karelia_HG is derived from the branch of “Ancient_North_Eurasian” that goes into the Karitiana Native Americans, rather
than the MA1 branch ... Karelia_HG share more alleles with Native Americans due to sharing additional common genetic drift (and thus Native Americans should share more alleles with Karelia_HG than with MA1) ... Karelia_HG has WHG-related ancestry which dilutes this affinity to Native Americans ... (EHG, WHG, and ANE) cannot be related to each other by a simple tree, and at least one of them must be admixed ..."


Well, this can't be answered with the Dstats I have run so far, so we have to recruit Karitiana into this as you have suggested (thanks for the suggestion). I will run these in about 10 hrs, starting off with D(MA1, Karitiana, Karelia, OUT) just to get the background, which should verify the longer drift branch between Karelia & Karitiana as opposed to Kaelia & MA1. It should verify assuming Karelia & MA1 don't share anything above and beyond ANE, which is not shared between Karelia & Karitiana.

Next, I will do D(X, Y, Karitiana, OUT) to compare samples X & Y to Karitiana. Specifically, what would you like for X & Y. Also, what other comparisons would you like to see in addition to Karitiana?

parasar
12-01-2015, 10:19 PM
Well, this can't be answered with the Dstats I have run so far, so we have to recruit Karitiana into this as you have suggested (thanks for the suggestion). I will run these in about 10 hrs, starting off with D(MA1, Karitiana, Karelia, OUT) just to get the background, which should verify the longer drift branch between Karelia & Karitiana as opposed to Kaelia & MA1. It should verify assuming Karelia & MA1 don't share anything above and beyond ANE, which is not shared between Karelia & Karitiana.

Next, I will do D(X, Y, Karitiana, OUT) to compare samples X & Y to Karitiana. Specifically, what would you like for X & Y. Also, what other comparisons would you like to see in addition to Karitiana?

Thanks, yes it would be nice to have a confirmation of higher drift sharing between Karelia and Karitiana vs. MA1.


D(Yoruba,Karitiana;Satsurblia,Stuttgart)
D(Yoruba,Karitiana;Kotias,Stuttgart)
D(Yoruba,Karitiana;Satsurblia, Anatolia_Neolithic)
D(Yoruba,Karitiana;Kotias, Anatolia_Neolithic)

would helpful to compare Karitiana vs MA1 (from prior stats).


D(Yoruba,Ongee;Satsurblia,Stuttgart)
D(Yoruba,Ongee;Kotias,Stuttgart)
D(Yoruba,Ongee;Satsurblia, Anatolia_Neolithic)
D(Yoruba,Ongee;Kotias, Anatolia_Neolithic)

To see if the trend towards ANE is also a trend towards the Ongee.

Perhaps a more distant outgroup can be used if you think that the Yoruba is not a true outgroup vis a vis EEF and CHG.

Kurd
12-02-2015, 04:18 AM
Thanks, yes it would be nice to have a confirmation of higher drift sharing between Karelia and Karitiana vs. MA1.


D(Yoruba,Karitiana;Satsurblia,Stuttgart)
D(Yoruba,Karitiana;Kotias,Stuttgart)
D(Yoruba,Karitiana;Satsurblia, Anatolia_Neolithic)
D(Yoruba,Karitiana;Kotias, Anatolia_Neolithic)

would helpful to compare Karitiana vs MA1 (from prior stats).


D(Yoruba,Ongee;Satsurblia,Stuttgart)
D(Yoruba,Ongee;Kotias,Stuttgart)
D(Yoruba,Ongee;Satsurblia, Anatolia_Neolithic)
D(Yoruba,Ongee;Kotias, Anatolia_Neolithic)

To see if the trend towards ANE is also a trend towards the Ongee.

Perhaps a more distant outgroup can be used if you think that the Yoruba is not a true outgroup vis a vis EEF and CHG.

I was quite puzzled and surprised when I saw Yoruba being used as an outgroup in Jones et al. Yoruba as an outgroup is a poor choice because some of these populations being studied have Yoruba related admixture, and thus the results will be somewhat inaccurate.

It appears that MA1 is more similar to Karelia than Karitiana is.



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


Karitiana
Samara_Eneolithic
Karelia_HG
Gorilla
-0.1016
-7.586
37269


Karitiana
MA1
Karelia_HG
Gorilla
-0.0616
-4.263
36954


Karitiana
Yamnaya
Karelia_HG
Gorilla
-0.0466
-5.23
51521


Karitiana
Bichon
Karelia_HG
Gorilla
-0.0389
-3.141
35156


Karitiana
Loschbour
Karelia_HG
Gorilla
-0.0359
-3.224
51282


Karitiana
RISE_baAndr
Karelia_HG
Gorilla
-0.034
-3.639
51433


Karitiana
Scythian_IA
Karelia_HG
Gorilla
-0.0291
-2.261
48144


Karitiana
Srubnaya
Karelia_HG
Gorilla
-0.0217
-2.48
51025


Karitiana
LaBrana1
Karelia_HG
Gorilla
-0.0148
-1.117
48855


Karitiana
Pima
Karelia_HG
Gorilla
0.0052
0.721
51581


Karitiana
Surui
Karelia_HG
Gorilla
0.0078
0.97
51581


Karitiana
RISE_baArm
Karelia_HG
Gorilla
0.0085
0.661
34150


Karitiana
KOTIAS
Karelia_HG
Gorilla
0.0149
1.309
47688


Karitiana
Kalash
Karelia_HG
Gorilla
0.0155
1.984
51581


Karitiana
LBK_EN
Karelia_HG
Gorilla
0.0244
2.767
51565


Karitiana
Iceman
Karelia_HG
Gorilla
0.025
2.086
51271


Karitiana
Georgian
Karelia_HG
Gorilla
0.0277
3.568
51581


Karitiana
Stuttgart
Karelia_HG
Gorilla
0.0312
2.978
51097


Karitiana
Anatolia_Neolithic
Karelia_HG
Gorilla
0.037
4.517
51102


Karitiana
Kostenki14
Karelia_HG
Gorilla
0.0432
3.415
48379


Karitiana
Nganasan
Karelia_HG
Gorilla
0.0452
5.851
51581


Karitiana
SATSURBILA
Karelia_HG
Gorilla
0.0475
3.447
37905


Karitiana
BedouinB
Karelia_HG
Gorilla
0.0619
7.765
51581


Karitiana
Ust_Ishim
Karelia_HG
Gorilla
0.1031
9.567
51556

Kurd
12-02-2015, 04:22 AM
The following show Kotias to be slightly more Karitiana shifted Anatolian Farmers and LBK
To be continued tomorrow......



POP2
POP1
TARGET
OUTGROUP
D
Z
SNP


KOTIAS
Surui
Karitiana
Gorilla
-0.29
-36.596
49376


KOTIAS
Pima
Karitiana
Gorilla
-0.2389
-29.847
49376


KOTIAS
Nganasan
Karitiana
Gorilla
-0.1085
-13.09
49376


KOTIAS
MA1
Karitiana
Gorilla
-0.0823
-5.999
35349


KOTIAS
Karelia_HG
Karitiana
Gorilla
-0.0727
-5.928
47688


KOTIAS
Samara_Eneolithic
Karitiana
Gorilla
-0.0561
-4.416
35150


KOTIAS
Scythian_IA
Karitiana
Gorilla
-0.0512
-4.263
45670


KOTIAS
Yamnaya
Karitiana
Gorilla
-0.0415
-4.781
49204


KOTIAS
RISE_baAndrov
Karitiana
Gorilla
-0.039
-4.372
49254


KOTIAS
Bichon
Karitiana
Gorilla
-0.0251
-2.102
33770


KOTIAS
Loschbour
Karitiana
Gorilla
-0.0207
-1.973
49101


KOTIAS
RISE_baArm
Karitiana
Gorilla
-0.0207
-1.58
32844


KOTIAS
Kalash
Karitiana
Gorilla
-0.0205
-2.677
49376


KOTIAS
Srubnaya
Karitiana
Gorilla
-0.017
-2.027
48752


KOTIAS
LaBrana1
Karitiana
Gorilla
-0.0131
-1.125
46830


KOTIAS
Kostenki14
Karitiana
Gorilla
-0.0098
-0.832
46375


KOTIAS
Georgian
Karitiana
Gorilla
-0.0043
-0.573
49376


KOTIAS
Iceman
Karitiana
Gorilla
0.0003
0.029
49092


KOTIAS
SATSURBILA
Karitiana
Gorilla
0.003
0.216
36000


KOTIAS
Stuttgart
Karitiana
Gorilla
0.0039
0.398
48942


KOTIAS
LBK_EN
Karitiana
Gorilla
0.009
1.083
49297


KOTIAS
Anatolia_Neolithic
Karitiana
Gorilla
0.0168
2.095
48923


KOTIAS
Ust_Ishim
Karitiana
Gorilla
0.0198
1.911
49350


KOTIAS
BedouinB
Karitiana
Gorilla
0.0317
4.092
49376

VOX
12-02-2015, 03:56 PM
Kurd, would it be possible to test if parts of North West Africa have have CHG related admixture. Looking to see if mtDNA U6/H rich areas such as Algeria, have influential effects from something related to CHG.

alan
12-05-2015, 10:00 PM
David has done some very interest calculations on Eurogenes. Sort of suggestive of a post-Neolithic spread of Caucasus types with a lot of CHG but low EHG through southern Europe. What strikes me about this is that this really brings to mind the recent paper which controversially suggested Caucuses migration through southern and south Alpine Europe c. 3200BC or so. I cannot recall the name of the paper. It pointed to single graves, barrows, dagger symbols etc. The cultures included Remedello II.

My question for David is do Remedello people have CHG?

Tolan
12-06-2015, 11:11 AM
My question for David is do Remedello people have CHG?

I do not think so!
Remedello is, instead, poor with caucasus / Baloch components compared to LBK.

Remedello is not on gedmatch, I tried with DIY Dodecad with the calculator harappa:

http://gen3553.pagesperso-orange.fr/ADN/harappabarre.png
The Italians today (among others) have perhaps CHG component.

Rise 423 (armenian) seems to be a mixture CHG + Europeans
Armenians today also have a higher caucasian than their ancestor rise 423.

Yamnaya not seem to be a mixture CHG + EHG, because very low of caucasian.

Generalissimo
12-06-2015, 01:55 PM
Yamnaya not seem to be a mixture CHG + EHG, because very low of caucasian.

Very funny. Thanks for sharing.

Tomenable
12-06-2015, 03:55 PM
RISE423 Armenian looks like a mix of Karelia + Satsurblia + small bits of Mediterranean and SW Asian.

sweuro
12-06-2015, 05:04 PM
I do not think so!
Remedello is, instead, poor with caucasus / Baloch components compared to LBK.

Remedello is not on gedmatch, I tried with DIY Dodecad with the calculator harappa:

http://gen3553.pagesperso-orange.fr/ADN/harappabarre.png
The Italians today (among others) have perhaps CHG component.

Rise 423 (armenian) seems to be a mixture CHG + Europeans
Armenians today also have a higher caucasian than their ancestor rise 423.

Yamnaya not seem to be a mixture CHG + EHG, because very low of caucasian.
Yamnaya has a very different ratio of Baloch-Caucasus than Satsurblia, if they are a mix of CHG + EHG it should be similar.

Tomenable
12-06-2015, 05:08 PM
But which Yamnaya sample(s) did Tolan use?

There were some internal autosomal differences between various Yamnaya groups.
(see for example Yamnaya Kalmykia versus Yamnaya Samara)