January 30, 2012

Egyptian autosomal genetics in the regional context (quick 'Admixture' run)

[Important caveat: apparently both Egyptian samples are from the Delta region, the one most affected historically by Eurasian influence. The one labeled Egypt or Egypt (1) is Henn's sample (n=18), while the one labeled  egyptan (sic) or Egyptian (2) is Behar's (n=12)].

Some readers questioned whether the strong Iberian affinity apparently found in Egypt in the previous Admixture run focused on North Africans was actually masked Highland West Asian or otherwise non-Peninsular Arab West Asian influences. I was initially skeptic because I had expected by default that Saudi Arabs would represent better all possible West Asian influences than Iberians.

I was mostly wrong as shown here.

Methods: I just run Admixture for K=4, K=6 and K=8 on a selection from the 1000 Genomes sample, following GNXP's instructions and using the following populations: both Egyptian samples (n=18 and n=12), plus 10 individuals from each of the following populations: Spaniards, Moroccans, Maasai, Ethiopians, Saudi Arabs, Palestinians, Turks and Kurds. The selection of Maasai and Ethiopians to represent Tropical Africans was made because previous research (mine and Henn's) showed these two being, of the available samples, the ones to best represent ultra-Saharan influence in Egypt specifically. 

Results: I am showing only the K=8 results because the lower K levels do not seem overly informative (if anyone wants them, feel free to ask).

1. The K=8 graph:


2. The K=8 numerical apportions (same as above but in figures, minimally edited by me to improve visualization):



3. The K=8 ADMIXTURE summary, showing Fst distances between components ('pops'), minimally edited to improve quick understanding (component 'ethnic' labels):



Highlights:

Egypt:

There is (and I could eventually detect) an Egyptian-specific component, of West Eurasian affinity (look at the Fst table), what implies that it's surely descendant of the pre-Neolithic Egyptians of Asian origin. Paleolithic Egyptians that I presume existed based on other genetics (mtDNA X1, M1 and such), as well as Eurasian-like iconography like the Qurta rock art, similar to materials from SW Europe and Anatolia (but admittedly the Egyptian Paleolithic, with a few exceptions, is not well known on archaeological grounds being such a sedimentary and then desertic area overall, and also because archaeology in Egypt has been largely focused on the quite impressive pharaonic period). 

This Egyptian-specific component represents 29% of one sample but only 19% of the other one, being also of some relevance in Ethiopia (9%). This and other differences between the two samples suggest some structure to be unveiled within Egypt but I lack the means (diverse enough samples) to do it. Anyhow the two samples are only somewhat different.

Besides this component, Egyptians show a diverse array of external influences, possibly Neolithic immigrants (?). The most important ones are the Kurdish or Highland West Asian component (17-20%) and the two Arab components together (14-25%) but others (Berber, Palestinian, East African) are also quite influential. The Iberian influence was largely a mirage (although still weights 4-8%).


East Africans:

Among the other populations, the most interesting finding of this run is that the Maasai appear, unlike in other research, to be 96% themselves (but still less distant from Eurasians than the average Tropical African, which is in the Fst=0.2 range), with at most residual admixture from Eurasians (mostly Egyptian/Palestinian). Ethiopians in turn appear here as somewhat admixed Maasai, with North African (mostly Egyptian) and peninsular Arab influences. However my previous relevant exercise showed that, at sufficient K-depth (or with a different sample strategy), Ethiopians eventually converge in their own specific and dominant genetic cluster (91%), which, as in the Fulani case, is similarly distant (and not too distant) to West Eurasians and Tropical Africans, indicating (I understand) very ancient homogenized intercontinental admixture. It surely requires an specifically designed run to understand these matters well enough. 


Arabian-Egyptian (Arab 2) very distant component:

Also notice the Arab 2 extremely distant Fst values, in the >0.2 range. On first impression I thought they were the Maasai component for that reason but nope. We may be here before another OoA remnant, which is very relevant in Arabia peninsula and also in the second Egyptian sample (c. 12% in both cases) and totally absent in Iberia instead.


In any case, it is again evident that different sample strategies can produce quite different results and therefore it is good to look at these matters with an open mind and many complementary perspectives.



Update: K=4 and K=6 graphs, for the record and because some kind of speculation may have some use for them:




Update (Feb 1, after realizing that both samples are from the Delta):

The finding of an Egyptian-specific component may be even more relevant further South. If some areas of the Delta have retained some 30% of this component, it's probable that it'd be even better preserved towards the interior. On the other hand I'd also expect more Tropical African influence further South but that should be at least balanced by a significant decrease of (post-)Neolithic West Asian influences.

Of course only real samples will provide real answers.

47 comments:

  1. IIRC, the Maasai sample has some close relatives, which makes the sample have its own cluster. If these relatives are pruned the Maasai appear to be a 2-way mixed population of Nilotic + Bantu

    ReplyDelete
  2. "Bantu" is probably not any real cluster. The Luhya (who are Bantu but East African) clustered with the Maasai and distinctly from West Africans in Henn 2012. In Patin 2009, Mozambican Bantus showed clear distinctiveness and clustered with the Twa and, less clearly, also the Bakola Pygmies (in the "non-Pygmy"(?) fraction of the genome). People insists in identifying languages and genes but that's not correct (at least not in many cases).

    Now, the exact alignment of Maasai is not what this exercise was designed to test anyhow and their component's Fst towards Eurasians is in any case lower than your usual Tropical African (but higher than your usual Eurasian), so it may still accept an admixture scenario.

    This should be tested on its own parameters, possibly an exercise with East African populations (and a few outsider controls).

    ReplyDelete
  3. Also why do you think that the presence of Ethiopians would influence the formation of the Maasai cluster? It shows up similarly at K=4, K=6 and K=8, which is what I tested.

    In Henn 2012, at K=4 they appear to be mostly identical as the Luhya (a Bantu but East African population), while the Bulala (a Nilo-Saharan speaking pop. from East Chad) actually cluster closer to the West Africans (Afroasiatic Hausa and Niger-Congo Yoruba are not distinct at these levels instead), although showing some "Maasai" affinity (before forming their own distinct cluster at K=8).

    In turn, when the Luhya and Maasai diverge, these do indeed show some Luhya influence (and vice versa) but show up as mostly their own distinct cluster which must be the same as in here. Only tiny apportions of "Arab" influence remained (which may be compared with the minimal apportions of Eurasian influence here).

    Other than the Luhya there was no "close relative" of the Maasai in Henn's analysis.

    That one is actually more convincing than this exercise in fact (but the results are fully comparable anyhow). However before the Maasai diverge from the Luhya, they do show up some "Basque" and "Berber" affinity, what is indicative, together with their somewhat intermediate Fst values, of ancient admixture later homogenized. There's an array of such groups indeed through the Sahelian belt.

    ReplyDelete
  4. Here is my Speculation:

    The Red (K1), The Light Cyan (K5) , the Blue (K6) and the light green (K3) are all affiliated with some type of an Afrasan connection, with K1 most affiliated with berber or maybe even some type of chadic included, K5 affiliated with Cushitic (Ancient), K6 associated with pre-semitic or cannanite/levantine and K3 affiliated with Post-Semitic, peninsular pre-Arabic or maybe even Sabaean/South Semitic (Now extinct) affinity.

    The light orange? Or K2 maybe affiliated with some type of Nilotic or Nilo Saharan in Egypt, however it is not much present in the Maasai, who are known to have Nilotic affinity as well, so it maybe a chadic or chadic-saharan affiliated cluster as well.

    The Dark green or K4, another ambiguous one, not sure what I'd associate that with.

    The Pink or K8, associated with Euro, may represent ice age enclave people mixing with Neolithic.

    The Purple or K7, associated with Ancient Indo European or Anatolian.


    ^ All total speculation really....

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. Sorry but your speculation is so Afrocentric influenced. The light orange isn't Chadic-Saharan at all, if you look at the figures well,you will know that Turks and Kurds have it and it is impossible that Turks have even 1% Chadic. The minor Sub-Saharan admixture in Egyptians is supposed to be Ethiopian or Somali. Bottom of line from most of DNA studies i saw Egyptians cluster between North Africans and South levantines ( Palestinians and Jordanians ) and i think that it also clear on their phenotype.

      Delete
  5. I posted the K=4 and K=6 graphs for the record and to serve as counterpoint to your speculation, Etyopis. What you're saying "some kind of Afrasian" is 3/4 and 4/6 components (true that most samples are of Afroasiatic language in any case).

    But, as I said to Conroy, I think that the genes=languages way of thinking is very limiting. These clusters may have coalesced deep in the Upper Paleolithic (or chronological equivalent), while no recognizable linguistic family is that old (Afroasiatic barely Epipaleolithic and is one of the oldest recognized groupings). There could be closer links with Y-DNA, maybe, because of the peculiars of male reproduction (effective polygyny and such) but probably not with the overall genetics or mtDNA.

    As I suggested in previous entries, if Fst>0.2 is a divergence of approx. OoA age, then Fst of c. 0.1 is half the OoA age (very roughly, I know), what is maybe 40 Ka ago, what is the time frame of the colonization of West Eurasia by H. sapiens (Aurignacian and such) and surely also that of the first or main Eurasian backflow into Africa (mtDNA M1, X1, U6...)

    So Afroasiatic flows would have remixed these components indeed but there does not seem to be one single component linked to that ethno-linguistic expansion process.

    My opinion anyhow.

    ReplyDelete
  6. "The Dark green or K4, another ambiguous one, not sure what I'd associate that with".

    That's the most distant component, notice that it has Fs>0.2 to all other components. It is more distant to any other component than Maasais are to Spaniards (or whatever other pair). For me it is another remnant of OoA tribes absorbed by the backflow.

    I guess it can all be an error but I have no reason other than the anomaly itself to think so, and I've already found such kind of components at least once before among South Moroccans, where it was expected (the 'Slovenian' component is more doubtable however).

    ReplyDelete
  7. Erratum: I just wrote: "notice that it has Fst>0.2 to all other components" and that's not exact, the actual figure is Fst>0.18.

    ReplyDelete
  8. "I think that the genes=languages way of thinking is very limiting."

    Offcourse genes are not always associated to languages or macro-language groups but it has been shown that genetics has a correlation to both geography and languages, sometimes to both, and other times just to either one of those criteria.
    Good Idea showing the other K sequences.
    So what about K=2,3,5 and 7? The progression of which group splits first, second, third etc..tells you alot about the dataset. Like for instance we can see that @ K=4,
    East Africans, North West Africans, Peninsular Arabians and west Eurasians have split off already/only.

    ReplyDelete
  9. "The progression of which group splits first, second, third etc..tells you alot about the dataset".

    That depends a lot on sample size and distribution. If I compare 50 Chinese, 50 French and 10 Bushmen, it's likely that at K=2 the Bushmen split between the Chinese and French components (at whichever apportion - never tried), while we know for a fact that Bushmen diverged before French and Chinese did.

    I trust more Fst distances. I have not tried here but in my first Admixture incursion a few weeks ago, a generic West Eurasian analysis, I made a Venn diagram based on Fst distances (scroll down for the update) that suggests structure of divergence - in that case with the Highland West Asian and NE European components in the center and the rest diverging in what looks like early branching out (except Basque and Chuvash components that are a bit anomalous).

    I suggest that you read it, including the "post-statement", where I build a tentative tree or divergence history, because that way you can probably better understand my line of thought.

    In this case it may be harder to recreate a comparable narrative because mixed inter-continental components are around, complicating the understanding, but excluding the Maasai-Ethiopian and the Arab2 component, it is still possible to make a partial reconstruction like the one I did then (it won't be identical however because the European divisions are not apparent).

    I could do it quite fast with pen and paper: the core in this case is Ibe+Kur and there are four components hanging at about the same Fst distances from this core (and slightly larger to each other): Berber, Egyptian, Arab1 and Palestinian components.

    As said before, Maasai and Arab2 components were excluded (too distant and surely complicated to integrate).

    ReplyDelete
  10. ''That one is actually more convincing than this exercise in fact (but the results are fully comparable anyhow). However before the Maasai diverge from the Luhya, they do show up some "Basque" and "Berber" affinity, what is indicative, together with their somewhat intermediate Fst values, of ancient admixture later homogenized. There's an array of such groups indeed through the Sahelian belt.''

    The Maasai are essentially Horn Africans who absorbed Nilotic and some Bantu elements. They speak a Nilotic language but most of their ancestry actually derives from the Horn rather than South Sudan.

    According to Tishkoff's study on Africans their primary genetic component is Cushitic rather than Nilotic.

    Also, see this:

    ''5.1. The migrations of the Maasai

    lt is a general belief that the pastoralists of Kenya, especially the Cushitic elements among them, are recent arrivals in the area. However, history now illustrates that these people preceded the currently populous and dominant Bantu in the settling of East Africa. Although the Cushitic population makes up over 80% of the population of Ethiopia, Djibouti, Somalia, and Eritrea, and Cushitic peoples extend to eastern parts of Egypt, it is clear that the ancestry of Cushitic people lies in the East African Rift Valley. For at least three millennia, Cushitic people lived in the East African Rift Valley and held sway over all the East African highlands prior to the migration of the Bantu people to the area between 500 and 1000 years ago.

    Among the first groups of pastoralists in East Africa were the Southern Cushites whose herds grazed the East African highlands down to central parts of Tanzania. The Cushitic people (who were also called Hamites) and the Nilotic people interacted significantly in pre-historic times. The result of the mixture of the two peoples has been the so-called 'Nilo Hamitic' peoples, which include the Kelenjin, the Maasai, and the Samburu.
    ''

    ReplyDelete
  11. The exercise is not designed to discern among East Africans, so I cannot really discuss what you say, Eze.

    It is plausible however. Not just the Maasai but also the Tutsi of Rwanda-Burundi have phenotype affinities to Horners indeed.

    ReplyDelete
  12. I know, I was just letting you know the overall affinities of that group. You can read more about their population structure in Tishkoff et al:

    http://www.sciencemag.org/content/324/5930/1035.abstract

    In order to discern Upper Nile region ancestry in Egyptians they may not be the best samples. You'll need South Sudanese for that.

    Also, the available Bulala samples from Central-East Chad may be an interesting group to use in analyses concerning North Africans as they may show Chadic-Saharan ancestry in them. Because the Tropical African ancestry in Egypt is not 100% East African as is shown in this analysis, it also has more Central affinities.

    ReplyDelete
    Replies
    1. Do you have a blog or email one can get into with you. I'm interested in tropical African ancestry in Egypt. What are your thoughts on the Ramses e1b1a

      Delete
    2. Do you have a blog or email one can get into with you. I'm interested in tropical African ancestry in Egypt. What are your thoughts on the Ramses e1b1a

      Delete
  13. Tishkoff's paper is PPV and I don't buy anything online (I just don't trust e-commerce security). So if you really want me to read it send me a copy, thanks in advance.

    I can judge the analysis as done in Henn 2012, and there Egyptians show, among Tropical African connections, mostly Maasai affinity (unlike NW Africans who show mostly Luhya affinity). They also show very minor other trans-Saharan affinities, in decreasing order: Luhya, Bulala and Nigerian (Hausa+Yoruba).

    "the available Bulala samples from Central-East Chad may be an interesting group to use in analyses concerning North Africans"

    Henn did and they are not too important (scattered and very minor affinity).

    I chose the Maasai and the Ethiopians for this exercise not blindly but because they were the ones scoring highest affinity with Egyptians in previous tests (mine and Henn's). Of course all this probably just reflects Nubian (or otherwise Sudanese) influence but we do not have Nubian samples to work with.

    ReplyDelete
  14. This comment has been removed by the author.

    ReplyDelete
  15. ''Tishkoff's paper is PPV and I don't buy anything online (I just don't trust e-commerce security). So if you really want me to read it send me a copy, thanks in advance.''

    Just sent it to you. It's a very useful study in understanding African population structure.

    ''Henn did and they are not too important (scattered and very minor affinity).

    I chose the Maasai and the Ethiopians for this exercise not blindly but because they were the ones scoring highest affinity with Egyptians in previous tests (mine and Henn's). Of course all this probably just reflects Nubian (or otherwise Sudanese) influence but we do not have Nubian samples to work with.
    ''

    Well, the Tropical African ancestry in Egypt is not completely East African. They also have nontrivial amounts of Western/Central ancestry (likely from the Sahara in the case of Egyptians). Without using samples from those areas it might skew overall results.

    http://3.bp.blogspot.com/-m5emkUFdgys/TgRIBluaAmI/AAAAAAAAAic/i3y0AGPK3Gw/s1600/ADMIXTURE_10.png

    ReplyDelete
  16. Thanks for the paper: it's a must have indeed - and incidentally shows again that Basques are not "inbred" as Dienekes claims (somewhat low diversity for Europe but higher than any East Asian, never mind Melanesians or Native Americans).

    However your notion is not based on it but on that graph, which I imagine comes from Dienekes' studies. Dienekes has pre-selected some "zombie" components and does supervised Admixture runs based on them. IMO that may be useful to run for individual samples or for police identification drills. But for our purpose, which is understanding real relationships between real populations, it is always much better to compare these real populations with each other and let affinities show up themselves naturally (unsupervised for the win!)

    Running unsupervised analysis you do not get those effects or rather they are less marked. In my North African test, Egyptians got 2.2% Mandenka and 0.5% Fulani, totaling less than 3% West African affinity. In the Henn study similarly they only show tiny bits of Bulala and Nigerian affinity, so it stands regardless of the control samples chosen.

    ReplyDelete
  17. Actually the link I posted earlier was an unsupervised run/with no zombie populations. This was one of his old runs.

    The W/C component in Delta Egyptians (the Behar et al. Egyptians are from the Delta) is about 4-5%. This is a nontrivial amount, especially when Ethiopians have 0% of it and suggests Egyptians also have some ancestry from further Southwest, albeit minor.

    This may be a very old component in Egypt, since STR Analyses on the Amarna mummies also linked them to West-Central populations.

    http://dnatribes.com/dnatribes-digest-2012-01-01.pdf

    ReplyDelete
  18. The figures for the mummies are best for the Great Lakes area (generally East Africa) and occasionally even South Africa. I do not take DNA tribes seriously at all but for whatever is worth that's no indication o "central-Africanness" nor are the mummies surely representative of the average Egyptian, ancient or modern (they are monarchs!)

    "the Behar et al. Egyptians are from the Delta"

    That's very interesting, thanks. I'm not sure which of the two samples is that one but when I find out, it may help with the understanding.

    In any case I don't think it's too important: you say 4.5% and I say 2.5% (Henn and my North African run).

    ReplyDelete
  19. question :
    where is the Arab2 sample from ?

    -it seems that both Egyptian samples are representative of the most desired "real estate" in Egypt , the Delta region, that was swarmed by the Neolithic waves...
    the ancestral Egyptian component is probably not showing up early, because :
    -the lack of an Egyptian sample with a relative ancestral Egyptian majority which could be obtained from the regions of Mid-Egypt , Upper Egypt,El Fayoum oasis, non-Delta Copts..etc

    -if someday a "strong" Ancestral Egyptian sample is obtained,it is most probable that the 'Southwest Asian' component outside the Arabian sphere in the Middleast,(West Asian Highland , North Africa and Europe.. ) will emerge as Egyptian rather than Arabian...

    ReplyDelete
  20. "This Egyptian-specific component represents 29% of one sample but only 19% of the other one, being also of some relevance in Ethiopia (9%)....possibly Neolithic immigrants (?). The most important ones are the Kurdish or Highland West Asian component (17-20%)"

    Pure speculation, based on my mostly geographical rules:

    1) Layer of foragers everywhere mainly low pop. density but high in a few favoured spots (like deltas?)
    2) What i think of as the gobekli component hopping along rivers and coasts focused on those spots that were best suited to their agricultural package.
    3) Once settled the gobeklis creating ripple effects among the surrounding populations.

    My rules and the map would predict that any basal Egyptian population would be densest in the delta. I think the delta area itself might have been too wet for the original farmer package and the foragers too dense to fight so i picture the Gobeklis maybe taking a few sites on the coast but mainly sailing down the Nile to the area around Cairo southwards and settling there.

    If so the basal Egyptian population would decrease along a North-South cline and the Gobeklis the reverse, denser in the south, more diluted in the delta i.e. the opposite to what you might expect.

    (With the possible exeption of the coast itself which i imagine has had a very complex history.)

    I also wonder - and this is very wild speculation i know - if there's any genetic connection between the Marsh Arabs of southern Iraq and a basal Egyptian component - if it can be found - again on the basis that high pop. density foragers could have fought off early farmers long enough to have settled into trade and diffusion of technology.

    Later of course numerous layers from Arabia etc.

    ReplyDelete
  21. There's no Arab2 sample: there's an Arab2 component (at K=8), actually shared (specially) between the Saudi and one of the Egyptian samples.

    When I first noticed it I just thought most important among Saudis and only after the labeling I realized it was almost as much important among one of the Egyptian samples. Otherwise I might have worked out a better name like "Ancient Red Sea" or something.

    "it seems that both Egyptian samples are representative of the most desired "real estate" in Egypt , the Delta region, that was swarmed by the Neolithic waves"...

    Are you sure? One is (said Eze above) but I was hoping the other one was from Cairo. There's some difference between both clades in the Egyptian and Arab2

    All the rest you say, may indeed make sense but we would have to test it to be reasonably sure.

    Still I haven't found any Egyptian centrality in other genetic markers, excepting maybe mtDNA X1 and the African clade of R1b (although maybe Sudan is more central in that case).

    Do you know it for sure? I was hoping at least one of them was from Cairo (more representative of the whole country, I guess). Actually by default I thought both would be from Cairo but I'm learning it's not the case.

    ReplyDelete
  22. @Grey: all too speculative IMO

    "My rules and the map would predict that any basal Egyptian population would be densest in the delta".

    The opposite is true in my mind, the Delta being more exposed to all Asian, Berber and even transmediterranean invasions. Upper Egypt is the stronghold: it has always been that way.

    ReplyDelete
    Replies
    1. Personally I don't think that invasions have big impact on Egyptian population, Iraq and Palestine were invaded more than Egypt and i think you know about cruel Mongol invasion of Iraq but i don't see any Mongol influence on Iraqis. Also Ancient Egyptians invaded many of their surrounding areas and occupied it for long time. During Thutmose III Egyptians expanded north till the southern borders of Modern turkey ,South till Ethiopia and West till Eastern Libya, this is the map of the Egyptian Empire at that time http://omarmohamed2222.imgur.com/all/?third_party=1.


      Delete
  23. For the record: the sample labeled 'Egypt' is from Henn 2011/12, while the one labeled 'egyptans' (sic) is from Behar (date?), and according to Eze, from the Delta. Henn's sample is also marked on the map as being from the Delta.

    If both samples are from the Delta, then it's easy to explain why they are so overwhelmed by Eurasian blood (and have comparatively low Tropical African one) but it's harder to understand why they differ as they do (maybe Behar's sampling location is closer to Asia?)

    Of course, we'd need to get other more interior samples (Cairo, Tebas, etc.) in order to judge properly. There's also a huge sampling hole in the Sudans, before partition the largest state of Africa and one of the largest on Earth, that should be relevant.

    ReplyDelete
    Replies
    1. I am Egyptian and i can tell you that Northern Egypt( Delta and its surrounding ) is far more populated than the rest of Egypt. Actually Northern Egypt is far far more populated than upper Egypt.

      Delete
    2. I don't doubt that, Knight, but it's also clear, I believe that there are important historical and paleo-historical differences between both regions, right? And also that the Delta has been more interactive with nearby regions such as West Asia, Libya or even across the Mediterranean sometimes, while the Upper Egypt area is more isolated and, if anything had contacts with Nubia and a few desert nomads east and west. So it's clear that having various points of reference would be interesting, as I would not expect Egyptians (modern or ancient) to be homogeneous across the board (even two Delta samples show some differences, so what else is hiding further inland?)

      Delete
    3. This comment has been removed by the author.

      Delete
    4. This comment has been removed by the author.

      Delete
    5. Egypt is a country located in North Africa and part of it is in West Asia ( Sinai peninsula ) and actually Ancient Egyptians looked something between Berbers and Semitics (the Christian Copts as the pure descendants of Ancient Egyptians and they resemble them). The difference between Upper and lower Egyptians is too exaggerated, actually North Upper Egyptians and Middle Upper ones look predominately Caucasoid and don't look much different from Northern Egyptians. The ones who look different are Nubians and Bejans and some of their surrounding. Actually the borders of Ancient Egypt is different from the Modern one, Modern Egypt include parts of Nubia, Also Nubians in the last century were moved by the governement from their area to Aswan and parts of Luxor because of Building the High Dam.

      This study of 15 autosomal STR loci in Upper Egyptians shows that The Upper Egyptian population is closer to the Middle Eastern, North African, South Asian and European populations than others areas.
      https://www.researchgate.net/publication/24006858_Genetic_variation_of_15_autosomal_STR_loci_in_Upper_Southern_Egyptians

      i would like you also to see this study
      https://www.ncbi.nlm.nih.gov/pubmed/9164042

      I would like you to see also this study about Genomic Ancestry of North Africans from Egypt till Morocco
      http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1002397

      Delete
    6. It's not about looks or "racial categories", Egyptians, like other North Africans are dominated by Y-DNA E1b, which is clearly of African origin, and also carry a sizable fraction of L(xM,N) mtDNA, which should also be of direct African origin one way or the other. They also carry West Asian lineages and maybe even European derived ones, so it's a melting pot since very ancient times. Sure: in autosomal DNA (which indicates the overall ancestry) the Eurasian one dominates no doubt, but it's not alone and there is at least some sizable Tropical African ancestry (8-11% in the samples available), which is color cyan (light blue) in the K=8 of this exercise and (maybe more apparent) color red in the K=4.

      As for there being some structure in Egypt in regard to this or other elements, we won't know until someone performs this kind of analysis or similar with better identified samples. My bet is that there should be some structure but unsure of what type. Then there's also Nubians, who used to live in what is now Southern Egypt (south of the 1st cataract) but were displaced by the construction of the Aswan dam.

      Your first link is not too informative for what I can see: only a way-too-global PCA is provided and the SSA sample is not identified (probably Yoruba YRI, which is not too relevant here). A proper unbiased analysis would use populations from the area, like Dinka, Maasai or Ethiopians (or ideally Nubians if available) to represent the possible sources of trans-cataractan admixture in Egyptians (and vice-versa). It would also restrict the Eurasian populations to a reasonable distance (what's the point in comparing with SE Asians?!)

      The second link is old (1997) and generic, although probably correct in what I can see in the abstract, at least in the rough outline.

      The third link (Henn's paper) is very similar in approach and results to what I did on my own on NW Africa, within this mini-series of ADMIXTURE "experiments", so there should not be any major contradiction. Anyhow I discussed it here if you're interested.

      Delete
  24. ''In any case I don't think it's too important: you say 4.5% and I say 2.5% (Henn and my North African run).''

    See this:

    https://docs.google.com/spreadsheet/ccc?key=0ArJDEoCgzRKedEY4Y3lTUVBaaFp0bC1zZlBDcTZEYlE#gid=0

    In the latest K12b Dodecad run the Delta group collectively had 12.2% East African and 4.8% 'Sub-Saharan' (a generic West-Central based cluster), while Ethiopian Jews and Somalians had 0% of the West-Central based cluster.

    So I don't think 4.8% is nontrivial, and it may be higher (8-10%) in the more Southern parts of Egypt. This suggests a Southwest Saharan substratum in Egypt.

    ReplyDelete
    Replies
    1. this study said that Jodranians have 2.2 Sub-Saharan (a generic West-Central based cluster) and that Yemenis have 8.8 and Algerians 7.5, i guess this compenent isn't native at all to the area, i think it is recent maybe because of Black Slave trade.

      Delete
    2. Know your prehistory, please. There's a clear northwards migration in the Mesolithic from the area of Nubia/Sudan which leads to Natufian in the Levant and to Capsian in NW Africa, and in general should be associated to the spread of Northern Afroasiatic languages (incl. Semitic, Egyptian/Coptic, Berber), although nowadays I'm not excluding at all that some East Sudanic (Nubian) languages were also involved. In any case, this is also apparent in the important presence of African-rooted lineage E1b1 in the Levant (also some African mtDNA in Neolithic Syria) and its subsequent spread to Europe already in the Neolithic.

      So I would not load the explanation primarily on the slave trade but rather, at least in part, on prehistoric migrations, which were pretty much color-blind for all we know. Nubia was an important Mesolithic cultural center (cereal gathering, very old pottery), it may get more peripheral later on but there was a time in the early Holocene when it was a major cultural and genetic center, and we do see their legacy today in all North Africa but also in West Asia and even in Europe.

      Delete
  25. This is what I and Henn coincidentally have obtained in unsupervised runs. I don't really trust Dienekes' "zombie" methodology.

    So I remain unmoved, sorry.

    The results Dienekes gets are very different but the methodology is also very different, so there probably lays the problem (as the samples are the same).

    If you don't trust my method/results/whatever, you can replicate or make a variant of the same test easily and I'll be glad to help you all I can. Essentially you just need a Linux OS (dual boot is fine and easy to install nowadays) and follow Razib's instructions (I'd start with step 3: removing unnecessary populations, so all the rest goes faster).

    You tell me what you find (actually I'll be glad to publish your results if different in any aspect, as long as the methodology is correct).

    ReplyDelete
  26. If it were for this or my other study you could say that there's lack of data (as I ignored Nigerians or the Bulala or the Luhya). But in this case I did because of Henn's study, which includes them all up to the point that they clutter the K-slots.

    And still nothing like what you/Dienekes claim.

    It must be his zombification doing strange things.

    ReplyDelete
  27. "there's no Arab2 sample: there's an Arab2 component (at K=8), actually shared (specially) between the Saudi and one of the Egyptian samples."
    ok, thanks

    "When I first noticed it I just thought most important among Saudis and only after the labeling I realized it was almost as much important among one of the Egyptian samples. Otherwise I might have worked out a better name like "Ancient Red Sea" or something."

    there is a possibility that the Arab2 component could be of Southern Indian origin with SouthEast Asian affinities (Melanesian-Papuan ), this specific component maybe strong in some Southern Arabian groups between Yemen and Oman..
    from my observations about the looks of the small Bedouin groups that migrated to North Africa in the 11th and 12th centuries, under the Banu Hilal 'brand' , they looked mostly Hindustani rather than Armenoid , and a few individuals are very dark and show some Australoid phenotype !
    I always wondered if they get to the Arabian Peninsula via some ancient Indian Ocean maritime connection ?

    until the 15th century Upper Egypt was largely Coptic,sampling Copts from different regions will be the best sampling strategy,for two main reasons :
    -no mixing with the Muslim Arabs
    -no role or behavioural participation in the Arabian-Muslim slave trade.
    by sampling the Copts , we will know for example the extend of the Nilothic gene flow in Egypt prior to the Islamic Era...or measure the true input of the Muslim Arabians...etc

    ReplyDelete
  28. ''It must be his zombification doing strange things.''

    He didn't use zombies in the latest runs. You can read the Technical Details on his blog. Anyway, as you said overall it is not that important, so let's not digress any longer.

    ReplyDelete
  29. @dalouh

    ''I always wondered if they get to the Arabian Peninsula via some ancient Indian Ocean maritime connection ?''

    There is little to no Indian sub-continental influence in ethnic Arabians. Otherwise the ASI/East Asian would shift them eastwards in global plots, which is not the case. They are mostly linked to West Eurasians.

    http://onlinelibrary.wiley.com/doi/10.1002/ajpa.20960/abstract

    ReplyDelete
  30. @Dalouh:

    "there is a possibility that the Arab2 component could be of Southern Indian origin with SouthEast Asian affinities (Melanesian-Papuan )"

    'Melanesian affinities' nope: there's no direct genetic connection that way. You may be here misled by the Onge approximate of Zack but it's a very questionable matter - in any case the Onge are not Melanesians in anything genetic but a distinct group of very distant South and East Asian affinities.

    The usual South Indian component should not be so distant either, more in the Maasai range of Fst distances at the most.

    That's why I suspect some other stuff, like a survivor component from the OoA (should not surprise us so much, I think). But I know I'm chasing an elusive ghost here and it's hard to prove.

    "sampling Copts from different regions will be the best sampling strategy"

    Sampling Copts could be interesting indeed but regular Muslim Egyptians should do as well. Even if Arabs (generic term) may have set colonies in Cairo and such, these would necessarily end up being swallowed by the Egyptian natural numerical superiority, based on a most productive agriculture.

    The Arab components therefore may actually relate to the Muslim conquest... BUT also to previous flows since the time of the Hyksos and even before. I have already mentioned that Arabs were known to exist in the Eastern fringes of Egypt already in Roman times, right?

    "no role or behavioural participation in the Arabian-Muslim slave trade".

    Really? Christianity did not forbid slavery, as you surely know.

    They may have some differences indeed but it would need extensive sampling to discern them properly. At the moment all that is speculation.

    My best guess is that both communities will be very similar and that the main differences will be geographic. That's because I don't think recent admixture was so important.

    But the research must be made fist.

    .......

    @Eze:

    "There is little to no Indian sub-continental influence in ethnic Arabians".

    There are "Indian" mtDNA lineages in Oman and other parts of South Arabia in sizable amounts: U2, M6 and M3 at the very least. This is an Arabian Sea connection and distinct from those via Iran.

    http://www.biomedcentral.com/1471-2156/5/26

    I still do not think it is the cause of this anomaly. These lineages are effectively non-existent in Egypt and even in Saudia they are minor (the caption area for Indian lineages is mostly Oman plus Hadramaut, and they are not even South Indian lineages specifically but generically South Asian ones in most cases).

    And do not trust too much PC analysis, or in general any single viewpoint of these genetic matters, no matter how automatically calculated it is. IF there is contradictory data, it's probable that both viewpoints have some merit and that the matter deserves further research (remains open).

    In autosomal genetics specially, only when several different exercises once and again converge in the same result we can be quite sure that it's solid. Because two different samples, two different algorithms, two different representation methods, two different starting assumptions... make for many different results potentially (unless it's a very clear-cut matter).

    ReplyDelete
  31. Actually those samples aren't taken from Delta Egyptians, it is taken from Middle Egypt area.

    ReplyDelete
    Replies
    1. Really? I took that info from Dalouh above, who I know has some knowledge and also from Eze, who was the first one to mention that one of the samples was from the Delta. Can you confirm that they are wrong?

      Delete
  32. LWK is just Bukusu, not Luhya tribes but just a cushitic clan within Luhya. My own Luhya tribe has a different origin. So you should be careful about using LWK to address 8million non related groups.

    ReplyDelete
    Replies
    1. OK, I take notice from your observation. However where did I use LWK in this analysis? The only East African population represented are Ethiopians and Masai.

      Delete

Please, be reasonably respectful when making comments. I do not tolerate in particular sexism, racism nor homophobia. Personal attacks, manipulation and trolling are also very much unwelcome here.The author reserves the right to delete any abusive comment.

Preliminary comment moderation is... ON (your comment may take some time, maybe days or weeks to appear).