The first dimension is surprising as I would expect there to be greatest haplotype variation in the dataset The first dimension is not to my suprise between Africans and non-Africans but strangely enough Europeans cluster by themself peaking among northern Europeans while Africans appear to show similarity with the East-Asians. Notice here that the PCA distance between Africans and East-Asians appear rather small.
Dimension 1 - brown Finns/Saami - blue Africans
The second dimension dimension is also surprising as it instead of showing African vs non-African variation instead shows a common African-European variation vs East Asians and Siberians.
Dimension 2 - brown - Africans/Europeans - blue - East Asians/Siberians
This gradient map also shows a very striking similarity to dimension 2 in the previous Euroasian analysis. As we can see here this dimension does not only seperate Europeans and East-Asians/Siberians but seperate Europeans-Africans from East-Asians-Siberians. I am very unsure about the interpretation but as dimension 1 it appear to be ancient.
These two dimensions can be summed up into two dimensions and as we can see this plot is identical to the PCA plot dimension 1 and 2 in a previous analysis investigating relationship between Europeans, East Asians and Siberians. This may sugggest that the previous indication of East-Asian ancestry among Southern Europeans may be due to shared Papuan-Melanasian ancestry among East-Asians.and Southern Europeans.
PCA Dimension 1 (horizontal) and 2 (vertical) Overview
PCA Dimension 1 (horizontal) and 2 (vertical) Overview Europe
PCA Dimension 1 (horizontal) and 2 (vertical) Overview Europe individual results
The third dimension finally appear to be a true African vs non-African dimension. The PCA coordinate distance between Africans and non-Africans is very large and outside Africa the haplotype variation appears rather uniform in comparison suggesting a bottleneck or/and foundereffect after leaving Africa. The fact that this dimension first appear as number tree suggesting this variation to be less than the previous dimensions make the previous dimensions intriguing. Maybe its just the effect of oversampling from the European region or maybe its traces of ancient migrations or mixing from earlier than out of Africa events.
Dimension 3 - blue - Africans, brown - Non-Africans.
This maybe make dimension 2 and 3 best for investigation of African and East-Asian/Siberian minority ancestry among Europeans. As we can see Spanish and Sardinians appear to have the most African like minority admixture while Saamis, Mordovians and Vologda Russians shows the most East-Asian or Siberian admixture.
PCA Dimension 2 (horizontal) and 3 (vertical)
PCA Dimension 2 (horizontal) and 3 (vertical) Europe Overview
European zoomed gradient maps:
Dimension 1 - Europe
Dimension 2 - Europe (Note Saamis should be blue)
Dimension 3 - Europe
The Dim 2/3 PCA is much like the typical admixture PCA's of Europe.
SvarSlettPapuans pull slightly towards Africa compared to southeast asians in other PCA's of worldwide variation, so what you said about them seems possible.
If you mean the "typcail V shape" in this project it is not. In this analysis Sardinians/Spanish form one branch whille the Saamis, Mordovians and Vologda Russians form the other. In the European analysis the Saamis together with Finns and Scandinavians form one branch while the eastern populations like Lithuanians, Mordovians, Vologda Russians, Russians, Belorussians form the other.
SlettI meant just in the sense of some northeastern european groups pulling towards Asia and Iberians towards Africa.
Slett"The first dimension is surprising"...
SvarSlettSampling strategy. You can do almost anything with the wrong sampling strategy and this is a good example: Europeans are clearly oversampled relative to the rest, so component 1 becomes European (vs non-European).
"The second dimension dimension is also surprising"...
Same thing: not the first time that undersampled Africans cluster with Europeans vs. East Asians, for example in Behar 2010 (or is it 2011?) at K=2.
"The third dimension finally appear to be a true African vs non-African dimension."
With a good sampling strategy it should be the first one. Just try reducing the three continental samples to similar numbers each... what should create an Africa vs. East Asia gradient which behaves variedly in West Eurasia (which should appear intermediate but rather tending to East Asia) depending on algorithm.
Still they seem to serve your initial purpose of detecting East Asian and African influences in Europe. So all is good.
Thats why I wrote "Maybe its just the effect of oversampling from the European region ".
SlettBut even its oversampled it still provide information of how the "European" variation continues outside Europe given that variation.
Maju wrote:
"Europeans are clearly oversampled relative to the rest,"
Here's something interesting from a recent article concerning newly sequenced genome of a 12,600 years old Clovis culture individual.
SvarSlett-In f3 most Europeans show as much affinity to 40,000 years old East Asian Tianyuan individual as modern East Asians
-Finns are more related to the Clovis individual than Mordovians and North Russians despite the latter being more "eastern" overall, maybe pointing to somewhat different origins?
http://www.nature.com/nature/journal/v506/n7487/fig_tab/nature13025_SF5.html
http://www.nature.com/nature/journal/v506/n7487/fig_tab/nature13025_F2.html
Very interesting, Anonym. I can't discern much in map 2 re. Eastern Europeans, but the issue of to Tianyuan is fascinating, because that means that the ancestors of modern East Asians were not yet near Beijing. This is particularly strange considering that some 10--20 Ka later they must have been, judging on Native American ancestry, and considering that Tianyuan has mtDNA B4'5, which is important in East Easia and among Native Americans.
SlettIt is also interesting that Palestinians appear much less akin to MA-1 than all other Eurasian populations, and that Na-Dené are less akin to Anzick-1 (the early Clovis individual) than Amerinds, what implies that Clovis was an Amerindian culture and that Na-Dené surely arrived later from Siberia.
And also it seems that the pre-Inuit peoples represented by the Saqqaq individual did leave a genetic legacy in the American Arctic, especially in Greenland, after all.
SlettRegarding Palestinians (and to a greater extent Bedouins), I'd say African admixture can explain a lot.
SlettI however can't figure out why the chinese minorities Oroqen and Tujia are less related to Tianyuan than other East Asians, Europeans and even Palestinians. Melanesians are in a similar situation, but with them it's easier to understand.
Not sure but they also show much weaker affinity to Alnwick (a tendency again shared with Melanesians), what suggests that the reason should be the same. The Tujia are from Southern Inland China, possibly the area of origin of Y-DNA O3, judging on Neolithic data, so maybe it has some relation with that (assuming that the Oroqen are their offshoot, what I'm not sure about, really).
SlettOroqen are considerably more "siberian" in admixture tests than Tujia, Han or Japanese. That's why it's interesting they behave like Tujia in relation to ancient DNA.
SlettI think I can explain it. There is a "Native American" influence at the same intensity between Finns and Saami seen much less among Mordovians and Vologda Russians.
Sletthttp://fennoscandia.blogspot.no/2014/02/europeans-and-native-americans.html
Anonymu wrote:
"-Finns are more related to the Clovis individual than Mordovians and North Russians despite the latter being more "eastern" overall, maybe pointing to somewhat different origins?"
As I wrote in a earlier post earlier the Melanasian/Papuan like admixture is of course strongest among South-East Asians but it is also seen to some weaker extent among Siberians and even Native Americans. Maybe it could explain it?
SlettAnonym wrote:
"I however can't figure out why the chinese minorities Oroqen and Tujia are less related to Tianyuan than other East Asians, Europeans and even Palestinians. Melanesians are in a similar situation, but with them it's easier to understand."
[Maju from a diferent computer]:
SvarSlettWhat about the Daur and other populations from the Mongolia area which do behave as "normal" East Asians?
I've been chewing on the matter a bit since my last comment and my provisional tentative solution is to hypothesize two (or three) different Asian populations some 60 Ka ago (approx.):
Pop. A:
-> Tianyuan
-> proto-Western
-> proto-Eastern1 -> mainstream East Asians
Pop. B:
-> proto-Eastern2 -> Tujia, Oroqen
Not sure where to place Melanesians, possibly in a third "pop. C".
Possibly pop. B would have a more "highlander" origin, near Tibet, while pop. A would be more "coastal".
Slett[Maju again]
It's possible.
SlettRegardless of whether there is a connection between Oroqen and Tujia or if they are less related to Tianyuan for separate reasons, these results show the limits of Admixture analyses. If we just look at those, Tujia look almost identical to Han while Oroqen look similar to other populations of Mongolia and Amur region.
what population overlaps on the continental maps with HUT1? it's not clear. it's something like "oer". is it "Iber"? thanks
SvarSlett