Friday, October 19, 2012

'globe4' calculator

Patterson et al. (2012) recently published evidence for admixture in northern Europeans between a population resembling modern Sardinians (and the Neolithic Tyrolean Iceman, whose genome was published earlier this year), and, surprisingly Native Americans. The authors attribute the Amerindian-like ancestry element to a North Eurasian population that spawned Native Americans, and which also contributed ancestry to northern Europeans. They propose two possibilities for the origin of this admixture: (i) the Mesolithic Europeans resembled Amerindians, or (ii) there was an influx of Amerindian-like populations from the east during late prehistory. A palimpsest of these two processes may explain parts of the observed signal of admixture.

In a recent K=4 admixture experiment, I demonstrated that ADMIXTURE software produces an Amerindian ancestral component that closely tracks the signal of admixture using the D-statistic test. I have decided to make this test available for download and use with DIYDodecad.

The test has four ancestral populations:
  • European
  • Asian
  • African
  • Amerindian
It is important to remember that some of these components track different aspects of ancestry that is better resolved at higher resolution. There are also populations that "don't fit well" in this 4-partite scheme (e.g., certain African or Australasian populations).

For example, the Amerindian component of this test may indicate (i) real recent Native American ancestry, (ii) East Eurasian ancestry found in Siberia and East Asia, (iii) the common signal of admixture differentiating most European groups from Sardinians and Near Eastern Caucasoid groups. Similarly, the Asian component may indicate Australasian, South Asian, or East Eurasian ancestry. And, the European component tracks the ancestry of individuals from West Eurasia in general, although it reaches is maximum in Sardinians.

This test may, however, be useful to Old World individuals who want to get an idea about the signal of admixture discovered by Patterson et al., so I decided to make it available. For individuals who don't suspect recent Amerindian or Siberian/East Asian ancestry, and who don't belong to populations with recent such ancestry, the Amerindian component will most likely represent the aforementioned signal.

You need to extract the contents of the RAR file to the working directory of DIYDodecad. You use it by following exactly the instructions of the DIYDodecad README, but always type 'globe4' instead of 'dv3' in these instructions. You can consult the spreadsheet for proportions of the 4 components in different world populations.

Terms of use: 'globe4', including all files in the downloaded RAR file is free for non-commercial personal use. Commercial uses are forbidden. Contact me for non-personal uses of the calculator.

17 comments:

  1. In a recent K=4 admixture experiment, I demonstrated that ADMIXTURE software produces an Amerindian ancestral component that closely tracks the signal of admixture using the D-statistic test. I have decided to make this test available for download and use with DIYDodecad.

    But ADMIXTURE is yielding much smaller percentages for the Mongoloid (Amerindians included) ancestry in Europeans than those detected by formal admixture tests (I am taking into account effects of both Mongoloid and Negroid admixture on populations in those tests). It is well known that ADMIXTURE has limitations in detecting ancient admixtures in close to full extent.

    ReplyDelete
  2. Looking at Dienekes' latest K=8 analysis reveals that while Altai, Uzbek, Uyghur people all have a lot of East Asian than other Turkic groups for example Yakuts, who live in the locale from which the Turkic language is supposed to originate, have less East Asian than all these groups. Chuvashes seem to lack it all together.

    https://docs.google.com/spreadsheet/ccc?key=0ArJDEoCgzRKedGR2ZWRoQ0VaWTc0dlV1cHh4ZUNJRUE#gid=7

    Turks have between 5% (on average in metropolitan areas) and 9.2% (in Aydin region) Siberian genes whereas Turkmens have 9.5% and Uyghurs have 23.3% suggesting that the Turkic inflow in to Anatolia is much larger than Dienekes' estimate of 1/7 depending on where you take the source population. Aydin is almost exact replica of Turkmenistan for example. It seems that Turks picked up East Asian near the Chinese border, more Siberian in the case of Chuvashes (unless they adopted Turkic later), and South Asian in the case of Turkmens and Turks who also have substantially more Near Eastern influences.

    Nevertheless if we follow the money (i.e. Siberian) one can propose 20% to 50% Turkic influence on Anatolia depending on the source population and the transposed location. East Asian, Amerindian or a summation of all "exotic" Asiatic influences are simply too unreliable for tracking Turkic movements. Siberian can unlock the key.

    http://i46.tinypic.com/2hobsig.jpg

    ReplyDelete
  3. Looking at Dienekes' latest K=8 analysis reveals that while Altai, Uzbek, Uyghur people all have a lot of East Asian than other Turkic groups for example Yakuts, who live in the locale from which the Turkic language is supposed to originate, have less East Asian than all these groups. Chuvashes seem to lack it all together.

    The commonly accepted urheimat of the Proto-Turkic language is what is now Mongolia and environs, not what is now the Sakha (Yakutia) Republic.

    Turks have between 5% (on average in metropolitan areas) and 9.2% (in Aydin region) Siberian genes

    First of all, the non-Aydin Turkish samples are no more metropolitan than the Aydin Turkish samples. You don't know how the various groups of Turkish samples were collected. Much more importantly, until the second half of the 20th century about 90% of the population of Turkey lived in the countryside. The great majority of the ancestors of today's city and town dwellers in Turkey lived in the countryside prior to the second half of the 20th century. Beginning from the 1950s, cities and towns of Turkey experienced massive immigration from the countryside due to industralization. So the differences we see between the various Turkish samples are actually the differences among the counrtyside Turks. What differentiates the Aydin Turkish samples from the other Turkish samples should be the fact that the Aydin province has a significant population of Yoruks (=nomadic Turks) while most of the other sampled regions do not have any Yoruks at all. It is natural to find more Central Asian Turkic origin in Yoruks than in village dwelling Turks. Traditionally, the great majority of the countryside Turks are village dwellers while Yoruks are a minority, so the non-Aydin Turkish samples represent the genetics of the countryside Turks and, consequently, of Turks as a whole much better than the Aydin Turkish samples.

    ReplyDelete
  4. whereas Turkmens have 9.5% and Uyghurs have 23.3% suggesting that the Turkic inflow in to Anatolia is much larger than Dienekes' estimate of 1/7 depending on where you take the source population. Aydin is almost exact replica of Turkmenistan for example. It seems that Turks picked up East Asian near the Chinese border, more Siberian in the case of Chuvashes (unless they adopted Turkic later), and South Asian in the case of Turkmens and Turks who also have substantially more Near Eastern influences.

    Nevertheless if we follow the money (i.e. Siberian) one can propose 20% to 50% Turkic influence on Anatolia depending on the source population and the transposed location. East Asian, Amerindian or a summation of all "exotic" Asiatic influences are simply too unreliable for tracking Turkic movements. Siberian can unlock the key.


    Modern-day Turkmens and Uyghurs are not the source population of the original Turkmens and other Turkic groups who invaded Anatolia during the late 11th and early 12th centuries. They came to Anatolia directly from what is now Kazakhstan and also invaded what is now Turkmenistan, Iran, the Transcaucasus and environs during the same invasion.

    Modern-day Turkmens are mainly descended from the Iranic speaking natives of what is now Turkmenistan. They were Turkified by the same original Turkmens from what is now Kazakhstan who also Turkified the Greek, Armenian and Assyrian speaking natives of Anatolia concurrently with the Turkification of what is now Turkmenistan and also of today's Azeri speaking lands.

    Turkmens possess much more "South_Asian" component than Turks, while Turks possess slightly less "South_Asian" component than Armenians and Assyrians. The great majority of Turkmens' "South_Asian" component is obviously from the pre-Seljuq Iranic natives of what is now Turkmenistan. The Dodecad Greek samples and Cypriots possess much less "South_Asian" component than Turks. The Dodecad Greek samples are ancestrally mainly from the Balkans and the Aegean islands, so they do not represent the Anatolian Greek genetics. Cypriots are an island population and have some genetic influence from the Aegean islands, in line with their Aegean Greek dialect. So there are currently no Anatolian Greek samples available to represent the Anatolian Greek genetics.

    The levels of "Atlantic_Baltic", "Near_Eastern" and "South Asian" components of Turks are intermediate between those of Greeks and Armenians/Assyrians and none of those populations carry above-noise levels of Negroid and Amerindian components. So overall, Turks are genetically pretty similar to their neighbors, chiefly Greeks, Armenians and Assyrians. What differentiates Turks from Greeks, Armenians and Assyrians is the above-noise level presence of the Asian Mongoloid components. But their levels are very small compared to their Caucasoid components.

    In summary, genetics, just like physical anthropology, is telling us that the great majority of the genes of Turks come from the Turkified natives of Anatolia and, to a lesser degree, of the Balkans. The precise proportion is still open to debate.

    ReplyDelete
  5. The "Amerindian" component appears to be highest in Uralic-speaking populations in both Asia and Europe (over and above what would be considered to be legitimate Native American admixture in the Koryak and Chukchi samples - based upon uni-parental DNAs).

    I wonder if it is possible to separate the Eurasian "Amerindian" component into two parts and, if you agree about the Uralic part, if this could be checked out. Patterson et. al. only referred to these populations as "Northern European".

    ReplyDelete
  6. Actually your K=8 pretty much separates the "Amerindian" component.

    ReplyDelete
  7. Onur you claim that all of a sudden some Turks from Kazakhistan simultaneously captured:

    -Turkmenistan
    -Uzbekistan
    -Anatolia

    This is not what happened.

    http://en.wikipedia.org/w/index.php?title=File:Seljuk_Empire_locator_map.svg&page=1

    Seljuq Turks first captured Turkmenistan.
    From this base they captured Iran.
    From Iran, Iraq and Syria as bases they captured Anatolia.

    http://en.wikipedia.org/wiki/Turkic_migration

    Seljuq capitals for the Great Seljuq Empire:

    http://en.wikipedia.org/wiki/Great_Seljuq_Empire

    Nishapur: http://en.wikipedia.org/wiki/Nishapur
    Rey: http://en.wikipedia.org/wiki/Rey,_Iran
    Isfahan: http://en.wikipedia.org/wiki/Isfahan
    Hamadan: http://en.wikipedia.org/wiki/Hamadan Western Capital in Iran
    Merv: http://en.wikipedia.org/wiki/Merv Eastern Capital in Turkmenistan


    So clearly Seljuq Turks' eastern base was in Turkmenistan until 1153.

    Then came the Anatolian Seljuqs (Seljuqs of the Rum) and other smaller Turkic tribes:

    Capitals were Iznik-Nicaea and Konya-Iconium.

    As such it is clear that Seljuqs colonized
    -First Turkmenistan
    -Then Iran and Azerbaijan
    -Finally Anatolia

    Anatolian Turks' Siberian content on average is half of Turkmenistan and Azeris are somewhere in between. There are locales such as Aydin where it is clear Turkic migration from Iran and Turkmenistan was as large as the local population.

    It is not very convincing to say Anatolian Turks descended from Kazakhstan all of a sudden when there are 2 centuries in between moving the Seljuq capital in Merv in Turkmenistan to Iznik in Anatolia.

    ReplyDelete
  8. @Anatolian Turkmen

    The Seljuqs and the original Turkmens from what is now Kazakhstan who accompanied them invaded Transoxiana, what is now Turkmenistan, Iran, the Transcaucasus and Anatolia during the same century (the 11th century). They captured Transoxiana and what is now Turkmenistan during the early 11th century, Iran and the Transcaucasus during the mid-11th century and Anatolia during the late 11th century, thus all of them during the same century. Their colonization of all those lands may have continued for some during the early 12th century, but by the mid-12th century the Great Seljuq Empire had practically collapsed and with that the colonization efforts of the original Turkmens had ended. So there is not even one century between the Turkmen invasion/colonization of what is now Turkmenistan and the Turkmen invasion/colonization of Anatolia, let alone two centuries. Thus the invading/colonizing Turkmens must be genetically the same in both regions.

    The Great Seljuq Empire was established with their invasion of Iran during mid-11th century. Their first capital was Nishapur. Merv, one of their last capitals, became capital as late as the early 12th century (in 1118 precisely).

    The Seljuq Sultanate of Rum was established in 1077, thus a few years after the Great Seljuq invasion of Anatolia in 1071. Their first capital was Nicaea (=Iznik), but after their defeat in the First Crusade they retreated from western Anatolia and moved their capital to Iconium (=Konya) in central Anatolia in 1097 and it remained their capital for the following centuries.

    Thus, Nicaea was the capital of the Sultanate of Rum, a splinter Seljuq state established in Anatolia, while Merv was a capital of the Greater Seljuq Empire. So they are capitals of two separate states that co-existed for a long time. More importantly, Nicaea was the capital of the Sultanate of Rum many decades before Merv was a capital of the Great Seljuq Empire. So you are confusing both states and chronology.

    You should learn some basic Turkish history before you post such nonsensical stuff.

    ReplyDelete
  9. while Merv was a capital of the Greater Seljuq Empire.

    while Merv was a capital of the Great Seljuq Empire.

    ReplyDelete
  10. Onur I know Turkish history very well. You have a very passive aggressive style. You have finally progressed from claiming there is not much Turkic input in Turkey to finally accepting that there is, with respective to Turkmenistan for example, significant Turkic genetic impact on Anatolia.

    Turkic immigration to Anatolia spanned 4 centuries up until the mid 1400's. The contact with Central Asia was cut off after the Safavid and Ottoman Empires became arch enemies. As such Iran blocked Anatolia off of Central Asia.

    You claim the following:

    1. Turkic people that Turkified Turkmenistan, Iran, and Anatolia to did that at the same time period. There is no proof of this.

    In fact there is a lot of counter proof. When Alp Arslan won the Manzikert Battle in 1071 he returned back to Merv, Turkmenistan. Anatolia was Turkish only on paper.

    In the late 11th and early 12th centuries Anatolia had not been Turkified yet. In fact Turkification really happened in the late 13th century.

    One could understand this from the historical architecture of Anatolia. The earliest of Seljuq architecture is the Alaeddin Camii from Konya built in 1220. Karatay Medresesi, Ince Minareli Medrese, Cifte Minareli Medrese, Gok Mederese and Buruciye Medresesi for example were all built between 1250 and 1271.


    2. You claim Turks that Turkified Turkmenistan, Iran and Anatolia were similar to today's Kazakhs or Kryghiz in genetic content. There is also no proof of this.


    What is clear is the following: There is a Turkic cline from Siberia to Anatolia and the Siberian genes seem to explain this migration event rather than East Asian genes.

    An average Uzbek has 25.4% Siberian, an average Turkmen has 9.5% Siberian, an average Aydinli Turk has 9.2% Siberian, an average Kayserili Turk has about 5.2% and an Istanbullu Turk has 5% Siberian.

    1. There is significant evidence to suggest that the Turkic impact on Anatolia has been comparable to Turkic impact on Turkmenistan

    2. Given the lack of data on the source population for Turkmenistan we do not know the genetic make up of the Oghuz tribes that reached Turkmenistan. We also do not know the exact date for that migration.

    3. We know that linguistically, culturally and genetically Turkmens are the closest Central Asian relatives of Ottoman and Azeri Turks and as such it is clear that a significant population of Turkmens from Turkmenistan and Iran moved to Anatolia between the 13th and 15th centuries. This percentage is tough to estimate but it is clear it is much higher than the 1/7 estimate looking at the Siberian segments.

    References:
    http://i46.tinypic.com/2hobsig.jpg

    https://docs.google.com/spreadsheet/ccc?key=0ArJDEoCgzRKedGR2ZWRoQ0VaWTc0dlV1cHh4ZUNJRUE#gid=7

    ReplyDelete
  11. @Anatolian Turkmen
    Just quit it mate.
    Don't you know Onur already? He is spreading this "there is no Turk in asia minor" argument in everywhere. He and his best friend lars are addicted to these anthropology blogs. Spreading "There is no Turk" argument must be greatest pleasure for them.

    ReplyDelete
  12. The factual errors I find in your posts are unexpected for a person who knows the Turkish history very well.

    Colonization is an organized event. It requires state sponsorship. With the practical collapse of the Great Seljuq Empire during the first half of the 12th century, Turkmens lost the state support for their colonization.

    Also, there is no evidence to support a centuries-long Turkmen migration wave to Anatolia or elsewhere. There may have been some small scale migrations after the first half of the 12th century, most notably during the mid-13th century in the aftermath of the Mongol invasion of the eastern Islamic-ruled lands. But the mid-13th century migration wave was a migration of refugees escaping from the Mongol hordes, as such, it was composed of small unorganized groups from various ethnic groups and regions and thus it was by no means a Turkic-only, let alone Turkmen-only, migration wave. Importantly, Anatolia was not the sole receiver of Muslim refugees from the east during that time, as all of the Islamic-ruled eastern Mediterranean world received Muslim refugees from the east. Egypt, for instance, was a safer haven for the refugees than Anatolia, as unlike Anatolia, it was not invaded by the Mongols.

    After the Battle of Manzikert (1071) Alp Arslan returned to Central Asia, not the Turkmen groups that had accrued to the Byzantine frontier. They advanced further into Anatolia and increased their depredations. But at first they were unorganized. But Suleiman ibn Qutalmish, who was the leader of a splinter group from the Seljuq family, very soon assumed the leadership of the Turkmen invaders in Anatolia and led them successfully against the Byzantine armies capturing most of Anatolia and established the Sultanate of Rum with its first capital in the western Anatolian city of Nicaea, no less, in 1077. No one can capture most of the Anatolian territories of the Byzantine Empire in such a short time and establish a long-lived state there without an appreciable number of troops. It is clear that by then there was some continuous migration of Turkmens to Anatolia. The Great Seljuq policy was to channel the depredatory Turkmen groups in their territories to their frontiers and beyond, and Anatolia was one of the most suitable locations to channel them (it was both outside the Great Seljuq territories and a Christian land). But with the practical collapse of the Great Seljuq Empire during the first half of the 12th century, there remained no state to channel the Turkmens to Anatolia and elsewhere, and the Turkmen colonization effectively ended.

    The source population of the Turkmen invaders/colonizers of Anatolia was in the middle parts of what is now Kazakhstan, around the Aral Sea and Syr Darya area. Their genetic make up remains to be conclusively discovered due to the lack of ancient DNA samples from them. On the other hand, the Seljuqs and their Turkmens were depicted with very salient Mongoloid features in statues (their statues are no different in the extent of Mongoloidness from those of Kok Turks) and miniatures. Recently, an 11th century Turkmen girl's skull found in a grave in central Anatolia was reconstructed:

    http://www.annieareyouok.com/2012/01/15/amansya-oluz-hoyukte-1000-yillik-turk-kizi/238359-anadoluturkler8/

    She pretty much resembles modern-day Kazakhs physically:

    http://upload.wikimedia.org/wikipedia/commons/d/d1/Kazakhs_people.jpg

    http://en.wikipedia.org/wiki/File:Kazakh_wedding_3.jpg

    ReplyDelete
    Replies
    1. Apparently the above link to the 11th century Turkmen girl reconstruction does not work anymore. So I am now giving another link to the same reconstruction:

      http://fotogaleri.ntvmsnbc.com/anadoluda-oncu-turklerin-ilk-izleri.html?position=8

      Delete
  13. BTW, Anatolian Turkmen, there is no change in my views on the Turkification on Anatolia. There is no need for change, at least for now.

    @Pecheneg

    As always, you are spreading falsehoods about me. I have a scientific and objective view on the Turkification of Anatolia, and Lars is not my friend, let alone best friend, nor do I share many views with him. Also, I never open this subject (=Turkification of Anatolia) myself (BTW, the number of blogs I visit is pretty limited and I visit no forum), but you or Anatolian Turkmen once in a few months open this subject again and again and do not even post on another subject to my knowledge. You are like dedicated propagandists of a certain religion or religious dogma rather than inquiring minds.

    ReplyDelete
  14. Onur phenotypes mean nothing. We need to see the genetics of the Amasya skeletons. Even then it is just one data point. So it means nothing.

    ReplyDelete
  15. Onur phenotypes mean nothing. We need to see the genetics of the Amasya skeletons. Even then it is just one data point. So it means nothing.

    Her phenotype points to significant level of Mongoloid ancestry in her. Though, of course, that doesn't tell its exact proportion. That it is just one data point is a more valid criticism. Certainly more Turkmen skulls from those times should be craniometrically analyzed (I am sure that team of scientists already analyzed many other Turkmen skulls from that excavated graveyard). If enough of them are analyzed and compared with present-day populations from various parts of the world, we can make pretty good predictions about the genetic make up of the invading/colonizing Turkmens in Anatolia just based on their phenotypes.

    ReplyDelete
  16. I think it would be an interesting exercise to make a host of inferences from this data under the assumption that it's correct, and, if not correct, at least evenly incorrect for all populations (and hence still showing a relative affect), and then test these inferences against findings of other measures.

    Having had time now to review the complete results, some things pop out at me:

    -- Iberian levels of the Amerindian component are essentially equal to those present in the Basques. Comparing D-statistics of Iberian sub-populations against Basques reveals that SSA admixture greatly reduces the D-statistic for any given population. Even though Valencia has slightly higher levels of the component than Basques, the Valencian D-statistic is about 40 less than the Basque, owing to just ~1% SSA admixture.

    -- There are no pure Amerindian Old World populations. Given that in many European populations there is an Amerindian component at K=4, but no Asian component, finding the time of fusion between Amerindian and Asian populations in Siberia can give us a minumum time for the introgression of the Amerindian component in Europe.

    -- Some heavily Y haplogroup N populations lack an Asian component at K=4 (e.g. Lithuanians), while others have this component (e.g. Finns). Possibly, introgression of N occurred in various waves?

    -- Some Y haplo R1b-present populations lack, or virtually lack, an Amerindian component (e.g. Armenians).

    -- Amerindians show a West Eurasian component at K=3. I've seen this on other calculators also (Harappa). However, your test of forced association revealed Amerindian associates fully with other East Eurasian components. Why would ADMIXTURE recognise this in one instance but not the other?

    -- Cross-referencing with other Dodecad calculators, or later K levels, can help us disentangle the pan-European Amerindian from the more recent N-related waves. Pre-N levels seem to be fairly even across Northern Europe (peaking in my calculations in both Lithuania and Ireland).

    ReplyDelete