February 16, 2012

First look at Turkish and Kyrgyz data from Hodoğlugil & Mahley (2012)

The authors of the recent paper on Turkish population structure were kind enough to share their data with me. I will be sure to use this data in future experiments, such as the ChromoPainter and fastIBD analysis of Balkans/West Asia, as well as a ChromoPainter analysis of Altaic speakers, following on the footsteps of my recent analysis of Afroasiatic speakers.

PCA


As a first step, after processing the new data, I carried out a PCA analysis (in smartpca with no outlier removal iterations), combined with various Turkic groups, as well as a few neighbors of Anatolian Turks, combining data from the literature and the Dodecad Project.


The Turkic cline from East to West Eurasia, observed by myself and others in various experiments is again evident.

The blowup of the above, focusing on the West Eurasian portion (top right) is easier to read:

As always, population labels are placed in the average position of each population. So, for example, the Behar et al. Iranians_19 sample is shifted to the left, because of the existence of a few African admixed individuals in this sample. The Iranian_D sample of Project participants seem to lack this admixture.

Also, note that since there is no South Asian reference in this first experiment, Iranians overlap with Turks along the first two dimensions. As we've seen in the Dodecad Project, both Iranians and Anatolian Turks are "eastward-shifted" relative to other West Eurasians, but the former have a strong South Asian- and the latter a Central Asian- tendency.

The new Kyrgyz sample falls between the Kazakh and the Altai along the cline, and is more "eastern" compared to the Uygurs and Uzbeks, and more "western" compared to Altai, Tuva, and Dolgans.

Kayseri and Istanbul Turks overlap with Behar et al. Turks as well as the Turkish_D sample. The Aydin sample appears to be more heterogenous, with a more eastern overall center of weight. More on this below.

ADMIXTURE


I also carried out a K=3 ADMIXTURE analysis of the dataset.


Below are the population portraits for the three new Turkish samples, as well as the Kyrgyz sample:


It is obvious that many Turks have low levels of Asian admixture, lacking in their geographical neighbors, but this is quite variable on an individual basis.

UPDATE (17 Feb):


I have also assessed the new data with the K12b calculator. Below are the normalized median proportions.



17 comments:

  1. It is interesting to see that Turkish Cypriots have only a slither of East Eurasian admixture, about the same amount as Greece, while Greek Cypriots appear to have zero. Thus paradoxically the EEA admixture levels in TC and Greece are closer to each other than either is to GC or Turkey. TC have lower levels of EEA admiture than any other group that shows EEA admiture, apart from Greece. Greece and Turkey appear to have zero African admixture, while TC have a slightly higher level than GC. Thus Im guessing that TC got their A admixture from Cyprus not from Turkey and got a slight EEA admixure from Turks. TC would thus appear to be Cypriots with small Turkish admixture. Possibly GC have some Greek admixture, which has lowered their levels of A admixture. Greece would appear, judging by levels of EEA to have a similar amount of Turkic admixture as TC but the EEA levels in Bulgaria perhaps raise the question of older EEA admiture in the Balkans that predates the Ottoman Empire.

    ReplyDelete
  2. Dieneke, could you repeat the same ADMIXTURE analysis this time with South Asian populations (especially from the south) and Adyghe (I want Adyghe just out of curiosity, not due to any relevance) added (you can also add several other Balkan, other Caucasian and northern Arab populations) and at a higher K to test also for the levels of South Asianness?

    Also, could you post spreadsheets that show the component percentages of the populations used in the ADMIXTURE analyses (including this one)?

    but this is quite variable on an individual basis

    This is expected due to its recency.

    but the EEA levels in Bulgaria perhaps raise the question of older EEA admiture in the Balkans that predates the Ottoman Empire

    Greeks (especially Balkan Greeks) might have some Balkan Slavic admixture, however small.

    ReplyDelete
  3. You can also add Mongols in your analysis.

    ReplyDelete
  4. Having studied and driven through SW Anatolia, there are many Yoruk (nomadic Turkmen) villages in the area of W and SW Turkey. So it is not surprising that Aydin which is in W Anatolia would carry a higher C Asian admixture than Kayseri and Istanbul. This is also a bit evident in the Y chromosome data from Cinnoglu et al.

    ReplyDelete
  5. Having studied and driven through SW Anatolia, there are many Yoruk (nomadic Turkmen) villages in the area of W and SW Turkey. So it is not surprising that Aydin which is in W Anatolia would carry a higher C Asian admixture than Kayseri and Istanbul. This is also a bit evident in the Y chromosome data from Cinnoglu et al.

    That was what I had in mind too. Looking at the results, it seems to me that they sampled the Aydin population largely from the pre-Balkan migration population of Aydin, hence the relatively high Eust Eurasian admixture. But the average genetics of Turks in regions where Yoruks are found must have significantly changed following the significant Balkan Turkish migration to western Turkey during the first half of the 20th century.

    On the other hand, interior regions like Kayseri did not experience significant migration from the Balkans, so the Kayseri samples should be fairly representative of the general genetics of Turks in Kayseri of today (as well as much of the Ottoman era).

    ReplyDelete
  6. It is not correct to attribute all Asian admixture in Turks to Yuruks.
    It well may have come through recent Balkan refugees, who have been admixed with Tartars in 18-19 century population exchange with Russia.
    It is only in rural Anatolia where Tartars and Cherkess lived separately and did not mix with locals, not in Bulgaria and Macedonia.
    See, Aydins have also higher North European admixture, which could not have come straight from Central Asia and local Anatolians.

    ReplyDelete
  7. ssas, no one is attributing all Mongoloid admixture in Turks to Yoruks. I just think that the Aydin individuals with relatively high Mongoloid input might be Yoruk or have Yoruk origin. Your Tatar theory is doubtful, as Tatars (Crimean Tatars to be more precise) settled only a small part of the Balkan Turkish areas of the Balkans and how much they mixed with them is open to question.

    Some Aydin Turkish samples have very little to no Mongoloid admixture. Maybe they are the ones who have Balkan Turkish origins. We don't know. Dienekes can help us on this issue by publishing the individual results of the above K12b analysis. This way we can see which Aydin samples have more "North European" component.

    ReplyDelete
  8. 1 Aydin 18636 14.3 6.7 2.3 0.0 12.6 12.2 0.0 0.0 10.2 5.3 36.5 0.0
    2 Aydin 18873 9.7 4.6 2.2 0.0 16.3 16.8 2.0 0.1 10.0 4.7 33.7 0.0
    3 Aydin 18784 10.3 6.6 2.5 1.3 14.5 13.0 2.5 0.0 9.7 4.8 34.9 0.0
    4 Aydin 18675 5.1 0.9 0.3 0.1 26.0 32.7 0.0 0.0 4.3 0.8 29.8 0.0
    5 Aydin 17337 18.3 0.0 0.0 0.0 10.6 3.5 0.0 0.0 11.5 1.2 54.8 0.0
    6 Aydin 18404 13.8 7.9 0.1 1.8 13.2 9.8 1.1 0.0 10.5 7.1 34.8 0.0
    7 Aydin 18314 12.1 5.4 1.6 1.3 18.8 13.9 0.0 0.0 8.6 4.5 33.8 0.0
    8 Aydin 18948 17.5 3.7 0.0 0.2 13.7 8.9 0.0 0.0 11.6 2.2 42.1 0.0
    9 Aydin 18483 14.2 7.5 0.3 0.4 11.6 11.5 0.1 0.6 9.6 8.2 36.2 0.0
    10 Aydin 18112 11.6 7.1 0.0 1.1 12.9 11.2 1.4 0.8 10.6 5.1 37.9 0.3
    11 Aydin 18116 14.0 10.4 0.7 0.0 11.7 13.6 2.9 0.0 7.7 6.0 33.1 0.0
    12 Aydin 18122 7.8 2.6 1.5 0.2 21.0 20.9 0.3 0.0 7.9 2.9 34.9 0.0
    13 Aydin 18117 7.7 1.4 0.1 0.0 25.5 28.6 0.0 0.0 7.5 0.0 29.1 0.0
    14 Aydin 18311 11.1 5.6 1.8 0.7 15.1 14.9 0.4 0.0 10.2 6.8 32.2 1.2
    15 Aydin 24783 13.8 6.9 1.7 0.2 13.6 11.0 1.4 0.7 11.0 5.4 34.1 0.0
    16 Aydin 18414 6.5 0.6 0.2 0.3 24.1 33.1 0.9 0.3 8.7 0.0 25.2 0.0
    17 Aydin 18587 10.6 5.9 1.4 0.1 15.3 13.3 0.0 0.0 8.7 7.6 37.1 0.0
    18 Aydin 18419 11.3 5.7 1.6 0.6 15.6 11.0 0.2 1.8 10.9 5.9 31.9 3.5
    19 Aydin 18596 12.8 8.3 0.0 1.3 11.9 13.7 0.4 0.0 10.4 6.6 34.6 0.0
    20 Aydin 24763 9.1 4.8 0.5 0.4 22.0 18.1 0.0 1.0 8.9 3.8 30.2 1.2

    ReplyDelete
  9. 21 Kayseri 24032 17.9 4.1 1.1 0.0 12.9 7.6 0.9 0.0 12.9 3.0 39.6 0.0
    22 Kayseri 23405 15.7 3.5 0.0 0.1 12.1 9.6 1.7 0.0 11.1 4.4 42.0 0.0
    23 Kayseri 23469 8.0 1.9 0.0 0.5 22.1 24.5 0.0 0.0 10.9 4.0 28.2 0.0
    24 Kayseri 23443 14.5 5.4 0.8 0.2 13.6 10.4 0.4 0.0 11.2 2.9 40.6 0.0
    25 Kayseri 24266 14.9 4.8 1.1 0.6 13.0 9.6 0.6 0.0 13.0 1.2 41.0 0.0
    26 Kayseri 24392 14.7 2.0 1.2 1.9 13.9 10.1 1.3 0.0 11.1 2.1 41.6 0.0
    27 Kayseri 24276 15.6 2.9 0.6 1.6 14.8 5.5 0.3 0.0 12.4 1.8 44.5 0.0
    28 Kayseri 24402 14.4 3.4 0.5 0.0 16.6 10.1 1.8 0.0 12.1 3.0 38.0 0.0
    29 Kayseri 24166 15.0 3.4 0.0 0.0 11.5 9.6 0.2 0.0 12.8 4.5 43.1 0.0
    30 Kayseri 24075 16.4 1.0 0.0 0.6 14.2 8.3 1.4 0.0 11.2 2.7 44.1 0.0
    31 Kayseri 23967 15.6 4.9 1.0 0.0 12.4 9.8 0.0 0.1 10.8 4.2 41.2 0.0
    32 Kayseri 23748 8.2 3.1 0.5 1.2 22.9 26.8 1.4 0.0 6.8 0.0 29.0 0.0
    33 Kayseri 23827 13.2 3.9 1.0 1.4 14.8 9.3 1.7 0.0 11.0 2.3 41.4 0.0
    34 Kayseri 24229 20.5 2.4 0.0 0.5 7.3 16.4 0.7 0.0 0.6 1.5 50.0 0.0
    35 Kayseri 24424 15.3 2.8 0.7 0.8 14.2 8.2 2.6 0.0 11.2 2.1 42.1 0.0
    36 Kayseri 23271 14.7 2.2 0.0 0.7 13.7 7.9 0.8 0.0 12.5 1.0 46.5 0.0
    37 Kayseri 23988 22.4 2.7 1.0 0.4 8.6 8.5 2.2 0.0 11.6 0.7 41.8 0.0
    38 Kayseri 23892 13.4 2.9 1.2 0.5 13.0 9.7 1.8 0.0 12.4 4.2 41.0 0.0
    39 Kayseri 23242 14.0 3.4 2.2 0.0 15.0 9.9 0.9 0.0 9.7 4.2 40.6 0.0
    40 Kayseri 23547 14.5 2.5 0.3 0.0 16.3 8.8 2.4 0.0 13.3 5.6 36.4 0.0
    41 Kayseri 23315 14.9 4.3 0.4 2.2 13.0 8.4 0.0 0.0 12.4 4.7 39.6 0.0
    42 Kayseri 23549 16.0 4.4 0.3 0.3 10.8 8.6 0.2 0.3 14.5 4.7 39.9 0.0
    43 Kayseri 23989 16.2 3.6 2.7 1.2 9.0 11.3 2.4 0.0 13.5 1.9 38.1 0.0

    ReplyDelete
  10. 44 Istanbul 25081 13.2 2.6 0.0 2.2 14.4 17.7 0.5 0.0 10.9 2.7 35.8 0.0
    45 Istanbul 25098 5.2 0.0 0.4 1.1 28.1 30.3 0.7 0.0 7.2 0.8 26.3 0.0
    46 Istanbul 25095 7.6 3.3 0.0 0.0 21.3 29.7 0.8 0.0 7.8 4.3 25.3 0.0
    47 Istanbul 17778 15.9 5.6 0.0 1.3 10.9 9.8 1.4 0.0 9.4 3.8 42.0 0.0
    48 Istanbul 19819 19.5 3.6 0.0 0.0 8.9 10.8 0.0 0.0 10.7 7.6 39.0 0.0
    49 Istanbul 19854 16.6 6.4 1.2 1.0 7.9 8.6 0.4 0.0 8.9 6.6 42.4 0.0
    50 Istanbul 19853 23.7 1.0 0.7 0.3 10.8 7.2 2.7 0.5 14.0 1.2 37.8 0.0
    51 Istanbul 19881 13.0 5.3 0.0 0.0 14.8 12.2 1.2 0.0 8.5 8.5 36.5 0.0
    52 Istanbul 20010 17.3 4.8 1.0 1.2 12.5 6.7 2.5 0.0 12.1 0.8 41.1 0.0
    53 Istanbul 20040 22.0 1.1 2.2 0.0 7.2 7.7 2.7 0.0 12.7 1.6 42.8 0.0
    54 Istanbul 20030 13.4 1.7 1.9 0.0 16.8 18.2 0.0 0.6 7.9 3.9 35.7 0.0
    55 Istanbul 20074 12.8 1.5 0.1 1.5 15.6 9.9 0.0 0.0 10.9 3.8 43.8 0.0
    56 Istanbul 15781 8.1 2.8 1.0 0.5 22.6 24.7 1.0 0.4 9.6 0.8 28.3 0.4
    57 Istanbul 17803 16.2 0.0 0.3 0.0 10.7 11.4 0.0 0.0 12.5 0.5 48.0 0.3
    58 Istanbul 19185 13.5 3.2 0.0 0.0 14.3 12.2 1.5 0.0 8.4 5.5 41.4 0.0
    59 Istanbul 17857 13.5 3.2 0.5 1.4 11.6 6.1 0.0 0.0 10.1 3.1 50.5 0.0
    60 Istanbul 20092 20.7 0.8 1.0 0.0 9.6 9.6 2.9 0.0 14.1 0.8 40.4 0.0
    61 Istanbul 19708 15.2 5.9 0.0 2.5 10.0 10.3 2.1 0.0 10.4 3.7 40.0 0.0
    62 Istanbul 19810 15.3 2.0 0.0 0.4 14.3 10.9 3.5 0.0 7.4 2.5 43.6 0.0
    63 Istanbul 25046 14.7 2.8 0.1 0.7 11.2 10.6 0.1 0.0 12.2 3.6 44.0 0.0

    ReplyDelete
  11. Thanks, Dieneke.

    Consistent with what I said and in contradiction with what ssas said, the Aydin Turkish samples with the highest "North European" + "Atlantic Med" components and the lowest "Caucasus" + "Southwest Asian" + "Gedrosian" components have the lowest Mongoloid admixture among the Aydin Turkish samples. The samples with the the highest "North European" + "Atlantic Med" components and the lowest "Caucasus" + "Southwest Asian" + "Gedrosian" components are almost certainly partially or totally descended from the Balkan immigrants. The only exception to this rule is Aydin 5; guessing where he/she is originally from (autochthonous? from somewhere else in Anatolia?) would be too speculative at the moment, but his/her genetics somewhat reminds me of the genetics of Northeast Anatolians, maybe he/she is partially or fully descended from migrants from Northeast Anatolia.

    ReplyDelete
  12. Thanks for the K12b results of all samples from Aydin, Istanbul and Kayseri. I am wondering if you could also provide K12b results of all individuals and all reference populations in a large spreadsheet?

    ReplyDelete
  13. I ran my aunt, an average Aydinli Turk, with mixed mode Oracle for K12a. This new Oracle includes Turkmens. Here are her results:

    [1,] "53.7% Greek_D + [B]46.3% Turkmens_Y" [/B] "2.9313"
    [2,] "56.1% Ashkenazy_Jews + [B]43.9% Turkmens_Y" [/B] "3.7916"
    [3,] "55.1% Ashkenazi_D + [B]44.9% Turkmens_Y" [/B] "4.2851"
    [4,] "52.8% Sicilian_D + [B]47.2% Turkmens_Y" [/B] "4.8908"
    [5,] "52.9% S_Italian_Sicilian_D + [B]47.1% Turkmens_Y"[/B] "4.9797"
    [6,] "51.6% S_Italian_D + [B]48.4% Turkmens_Y" [/B] "5.0098"
    [7,] "49% C_Italian_D + [B]51% Turkmens_Y" [/B] "5.4416"
    [8,] "64.5% Ashkenazi_D + 35.5% Tajiks_Y" "5.4652"
    [9,] "61.1% S_Italian_D + 38.9% Tajiks_Y" "5.6985"
    [10,] "65.6% Ashkenazy_Jews + 34.4% Tajiks_Y" "5.7723"

    I think it may be fair to categorize her as half Greek Anatolian and half Turkmen from Turkmenistan. Well it has been 600 years since we came to Anatolia. This approximation may be true after all. Before Turkmens were in the sample she was coming up 1/3 Uzbek and 2/3 Ashkenazi. I don't know if this Ashkenazi proxies for Roman/Anatolian or if it is genuine Turkic/Ashkenazi admixture. Half Greek half Turkmen maybe more common sense.

    So maybe [B]this suggests to us that at least to Aydin area the immigration was large enough to bring a Turkmen population equal to that of the local Anatolian population.[/B] 50-50 admix resulted in Turkish dominance. [B]It's not like some elite dominance with 1/10th the population happened.[/B]

    I think this is a very important result. It kind of makes sense that Anatolian Turkmens are half Turkmenistan Turkmens. So an Anatolian Yuruk/Turkmen is like half Anatolian and half Turkmenistani.

    [B]We are Turkmen[/B], I've been saying it and nobody believed me.

    Pre Fatih Ottoman map is a 30 to 50% Turkic immigration population.

    The Green area minus Balkans minus Crimea was 30 to 50% Turkmen in 1481. I have no doubt.

    [url]http://en.wikipedia.org/wiki/File:Territorial_changes_of_the_Ottoman_Empire_1481.jpg[/url]

    Aydin is just one of the better sociologically better preserved cities (until the mass Kurdish immigration of the last 20 years, which has changed it immensely)

    30 to 50% Turkmen areas ca. 1481 Anatolia
    [url]http://oi42.tinypic.com/1gj635.jpg[/url]

    ReplyDelete
  14. Anatolian Turkmen,

    The numbers you gave are erroneous and irrelevant. As I have told you countless times, Turkmenistani Turkmens of today can not represent the Turkmens who invaded Anatolia ~1000 years ago. What is now Turkmenistan was invaded by Turkmens during the same centuries with the Turkmen invasion of Anatolia and before that what is now Turkmenistan was an entirely Iranic-speaking land. Turkmens who invaded what is now Turkmenistan so much mixed with and Turkified the Iranic speaking locals during the last few centuries that the Iranic languages of what is now Turkmenistan entirely disappeared in favor of the Turkmen language. Turkmens who simultaneously invaded Anatolia and what is now Turkmenistan invaded those lands from what is now Kazakhstan. So the best modern proxy for the genetics of the Turkmens who invaded Anatolia seems to be Kazakhs. Of course, the only conclusive evidence for the genetics of the Turkmens who invaded Anatolia seems to be found in the ancient DNA data from Turkmen invaders of 1000 years ago, which is unfortunately lacking for now. I think over the next few decades, with the impovements in ancient DNA extraction and sequencing technologies, genomes of the invading Turkmens can be investigated and thus can shed light on the mechanisms of the Turkification of Anatolia.

    ReplyDelete
  15. Onur these are all aproximations. If you accept Oracle overall then this is just one of another output of Oracle.

    Ancient DNA is the surest path but this is the best tool we have now. The results are very persistent.

    ReplyDelete
  16. Anatolian Turkmen,

    Oracle cannot calculate the percentages of the ethnic origins of Anatolian Turks, much less with the populations you use. As I already made clear, modern Turkmenistani Turkmens and Tajiks (and probably also Uzbeks) are not representative of genetics of the Turkmens who invaded Anatolia ~1000 years ago. Balkan Greeks (Dodecad Greeks have overwhelmingly Balkan origins), Italians and Ashkenazi Jews (these are all European populations as opposed to Anatolian, even Ashkenazi Jews have significant European admixture) are not representative of the genetics of the pre-Turkmen invasion Anatolia; a mix of Anatolian Greeks, Armenians and Assyrians would be much more representative. For the genetics of the invading Turkmens, I advise you to wait for the ancient DNA research. I am sure over the next few decades their DNA will be investigated.

    ReplyDelete
  17. Dienekes can you please report the y-DNA and mtDna of the Aydin samples?
    Thank you.

    ReplyDelete

Stay on topic. Be polite. Use facts and arguments. Be Brief. Do not post back to back comments in the same thread, unless you absolutely have to. Don't quote excessively. Google before you ask.