The Unz Review - Mobile
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 Gene Expression BlogTeasers
Chinese, Koreans, and Japanese

Since there was some discussion about East Asian genetic structure below…I pulled about 20 South Koreans I have in my data. Merged them Han and Japanese from the HGDP. I then ran a PCA and plotted it, and also unsupervised ADMIXTURE, and plotted it.

The results are below.

Rplot42

Screenshot 2016-08-29 03.09.33

 
• Category: Science • Tags: Koreans 
Email This Page to Someone

 Remember My Information



=>
Commenters to Ignore...to FollowEndorsed Only
[]
  1. If I were someone who had written (or felt like I should have written) an article because my 23andme results seemed confusing, I would at least try and find out if they were wrong. Especially if my great-grandfather was adopted.

    By having your parents genotyped, you could see if all the confusing mixture came from just one side, or was a random, seemingly incorrect jumble on both sides.

    It could be that Koreans share a lot of allele frequency with close neighbors. But maybe they have very distinct haplotypes. (This is very true of African populations, but I don’t know much about Koreans).

    If the phasing, though, is based on only Chinese and Japanese samples, then it will segregate alleles into incorrect haplotypes. This could be a reason for the difficulties.

    I would be very interested in seeing the 23andme results of a Korean before and after adding their parents’ genomes for direct phasing.

    Read More
    ReplyAgree/Disagree/Etc.
    AgreeDisagreeLOLTroll
    These buttons register your public Agreement, Disagreement, Troll, or LOL with the selected comment. They are only available to recent, frequent commenters who have saved their Name+Email using the 'Remember My Information' checkbox, and may also only be used once per hour.
    Sharing Comment via Twitter
    http://www.unz.com/gnxp/chinese-koreans-and-japanese/#comment-1546467
    More... This Commenter Display All Comments
  2. I was really going to abide by my promise not to post here but just one more.

    This analysis very nicely illustrates what has been known to many; a known mixed population can superficially look “pure” by the (usually bad) choice of reference populations. In this case N_Han, a well known mixed population independently confirmed by many disciplines, looks “pure”.

    Japanese also harbor Jomon and Yayoi components, roughly speaking. But Jomon component appears as just Japanese because in this choice of population set it is all Japanese. Yayoi left descendants both in Korea and Japan but the migration was so massive that Japanese have more of them(>60 percent) than Koreans(as little as 20 percent, probably close to 30 percent) so they would be assigned as “Japanese” and make up nearly all of “Japanese component” among Koreans.

    Curiously this fits very well with Alexander Vovin’s thesis that Southern and Central Koreans originally spoke a Japonic language(s) preserved in place names etc. which have been the source of bizarre Koguryo-Japonic hypothesis recently advocated(actually has a long history in linguistic circles) by non-linguists such as Chris Beckwith and Jared Diamond. Vovin and Unger take the opposite view and according to them Korean is the intrusive language imposed on Japonic speaking original inhabitants.

    This also reminds us of an important caveat. A mixed population will be between the two parental populations in the PCA analysis but that a population is between two does not mean that the population is a mixture of the two.

    Malta boy generally regarded as “pure ANE” by many amateur bloggers also may have a very complicated genetic history in a similar fashion. Bifurcation into West and East Eurasians is a statistical concept and the crudest approximation. It is absurd to pigeonhole every ancient sample as East or West Eurasian because the actual history is far more complicated.

    Read More
    • Replies: @Razib Khan
    Bifurcation into West and East Eurasians is a statistical concept and the crudest approximation. It is absurd to pigeonhole every ancient sample as East or West Eurasian because the actual history is far more complicated.

    this is not true. there is some gene-flow between west and east eurasian. or, more precisely, west-north eurasian and east eurasian + amerindian. but it's not an exceedingly complex mixture. the ancient samples that are poor fits into this dichotomy come from a period close to the divergence of non-africans into separate populations.
    , @Twinkie

    Curiously this fits very well with Alexander Vovin’s thesis that Southern and Central Koreans originally spoke a Japonic language(s) preserved in place names etc. which have been the source of bizarre Koguryo-Japonic hypothesis recently advocated(actually has a long history in linguistic circles) by non-linguists such as Chris Beckwith and Jared Diamond. Vovin and Unger take the opposite view and according to them Korean is the intrusive language imposed on Japonic speaking original inhabitants.
     
    I am favorably disposed toward this view. It seems to me that given the extensive contacts between the Gaya Confederacy (or Mimana Command Post) and Yamato Japan (and later that between Baekche and Japan) were more than simply political or economic in nature (there were significant intermarriages between ruling houses), there must have been strong cultural and probably some residual ethnic similarities. I think the rice agriculture patterns were very similar between these two groups.

    It seems highly plausible to me that the people who formed Jin and later Samhan in central and southern Korean peninsula were either the same as or related to those who founded Yamato, and that both spoke variants of what some scholars called Japonic languages.

    Meanwhile the people who formed Goguryeo were of the mixed Buyeo-Gojoseon stock, and originated from what is today southern Manchuria. They were probably related to the Malgal/Mohe tribal peoples whom Goguryeo later subjected (Mohe cavalry frequently appears as Goguryeo auxiliaries in its wars). This stock of people were probably the ones who brought the "intrusive" so-called Koreanic languages (Buyeo, Goguryeo, and later Silla variants) to the south of the peninsula.

    The problem, from a historiographical perspective, is that this topic is highly charged politically in the region. Everytime some ancient polity is "claimed" by a modern nation-state as its own (e.g. Goguryeo as a northern "non-Han" Chinese as opposed to a Korean polity), others "flip out" and all heck breaks loose politically. Gaya/Mimana is another classic example - it was peopled by a population that probably shared a strong affinity with the Yamato, at a time when neither "Korea" nor "Japan" as such existed. But because of intense nationalism in the region, it's difficult for the local historians to take objective views on the matter, which leaves outsiders such as Gina Barnes* to strike a more reasonable and scientifically consistent note.

    Professor Barnes, in my view, is the best authority in the West on the origins of ancient Korean and Japanese polity formations: https://www.soas.ac.uk/staff/staff30624.php

    Her books: https://www.amazon.com/s/ref=nb_sb_noss?url=search-alias%3Daps&field-keywords=gina+l+barnes
  3. @Hector
    I was really going to abide by my promise not to post here but just one more.

    This analysis very nicely illustrates what has been known to many; a known mixed population can superficially look "pure" by the (usually bad) choice of reference populations. In this case N_Han, a well known mixed population independently confirmed by many disciplines, looks "pure".

    Japanese also harbor Jomon and Yayoi components, roughly speaking. But Jomon component appears as just Japanese because in this choice of population set it is all Japanese. Yayoi left descendants both in Korea and Japan but the migration was so massive that Japanese have more of them(>60 percent) than Koreans(as little as 20 percent, probably close to 30 percent) so they would be assigned as "Japanese" and make up nearly all of "Japanese component" among Koreans.

    Curiously this fits very well with Alexander Vovin's thesis that Southern and Central Koreans originally spoke a Japonic language(s) preserved in place names etc. which have been the source of bizarre Koguryo-Japonic hypothesis recently advocated(actually has a long history in linguistic circles) by non-linguists such as Chris Beckwith and Jared Diamond. Vovin and Unger take the opposite view and according to them Korean is the intrusive language imposed on Japonic speaking original inhabitants.

    This also reminds us of an important caveat. A mixed population will be between the two parental populations in the PCA analysis but that a population is between two does not mean that the population is a mixture of the two.

    Malta boy generally regarded as "pure ANE" by many amateur bloggers also may have a very complicated genetic history in a similar fashion. Bifurcation into West and East Eurasians is a statistical concept and the crudest approximation. It is absurd to pigeonhole every ancient sample as East or West Eurasian because the actual history is far more complicated.

    Bifurcation into West and East Eurasians is a statistical concept and the crudest approximation. It is absurd to pigeonhole every ancient sample as East or West Eurasian because the actual history is far more complicated.

    this is not true. there is some gene-flow between west and east eurasian. or, more precisely, west-north eurasian and east eurasian + amerindian. but it’s not an exceedingly complex mixture. the ancient samples that are poor fits into this dichotomy come from a period close to the divergence of non-africans into separate populations.

    Read More
  4. The admixture graph in this post seems to fit the critique of Euny Hong and Hector and others that there are almost no 100% Korean ancestry individuals in the sample, despite the fact that inbred populations tend to form distinct ancestry components and the widespread conventional wisdom that Koreans are more inbred than either the Japanese or the Chinese.

    Read More
    • Replies: @Razib Khan
    the chinese are not a good comparison with japanese or koreans. there is some noticeable structure within chinese, including obviously relatively recently sinicized subcomponents in the 1000 genomes sample.

    i'm not sure koreans are more inbred than japanese.

    reading through the comments i'm starting to wonder if my friends were early adopters of 23andMe (they were), and were assigned using an older method whose results were frozen.

    that being said, if 23andMe's methods can correctly assign 'chinese' ancestry to people (and it does), then i don't see why it can't for korean, as they are more homogeneous.

    ultimately, the details are in the methods. the data though is probably sufficient.
  5. @ohwilleke
    The admixture graph in this post seems to fit the critique of Euny Hong and Hector and others that there are almost no 100% Korean ancestry individuals in the sample, despite the fact that inbred populations tend to form distinct ancestry components and the widespread conventional wisdom that Koreans are more inbred than either the Japanese or the Chinese.

    the chinese are not a good comparison with japanese or koreans. there is some noticeable structure within chinese, including obviously relatively recently sinicized subcomponents in the 1000 genomes sample.

    i’m not sure koreans are more inbred than japanese.

    reading through the comments i’m starting to wonder if my friends were early adopters of 23andMe (they were), and were assigned using an older method whose results were frozen.

    that being said, if 23andMe’s methods can correctly assign ‘chinese’ ancestry to people (and it does), then i don’t see why it can’t for korean, as they are more homogeneous.

    ultimately, the details are in the methods. the data though is probably sufficient.

    Read More
    • Replies: @Hector
    Hong is rather typical for a South Korean.
    But I could see 3 anomalies.

    1. She is the only Korean sample whose "Chinese" is greater than "Japanese" even though by a very slight margin.
    Most South Koreans have >2 ratio in favor of "Japanese".

    2. Only one with a non-zero SE Asian even though it is only 0.1 percent. All 40 Koreans(I checked up to about 15 and then got bored) I have access to have 0 there. But this could be due to the design of the new test.

    3. Only one with non-zero Oceanian. But this again could be due to the new test.

    About 70 percent of Koreans have <1 but non-zero "European" component but Hong had none. This again could be due to the design of the new test.

    I have seen a Chinese from Northwest who had a fairly substantial amount of "European", like 5 percent. I also saw a Chinese with a "Korean" component.But most Chinese had nearly 100 percent of themselves with occasionally substantial SE Asian.

    Chinese components among Koreans increase dramatically as you move from "conservative" to "speculative". This makes sense since Chinese and Koreans share prehistoric common ancestry but have been isolated from each other for a long time but the migration from Korea to Japan happened at the edge of historic times thus longer shared segments.

    P.S. The chromosome graph does not show actual shared segments. For instance a long Chinese segment actually is made up of many smaller segments that happened to be assigned to "Chinese". Apparently they pull up every Chinese segment into one of the homologous pair and the gap between those segments is painted as "Chinese". I think this is how a full blooded Korean gets an entire chromosome assigned to Chinese or Japanese.
  6. From the way Eurasian continent is shaped(relatively long East to West dimension, especially when you only consider habitable zones), no matter how complicated the population history is, statistics will generally give two components, East and West. That does not mean all populations, even roughly, diverged into two kinds in a cladistically coherent manner.

    In the way it is defined currently West Eurasians are a paraphyletic group and East and West Eurasians are not coherent cladistic groups. Lazaridis himself was ambivalent about how to define “West Eurasians”. He suggested that it is one possibility to consider Basal Eurasians as the true West Eurasians but decided otherwise(current form).

    Malta boy is West Eurasian not because he was clearly born as one but because of the way later generations intermingled and interbred. If some Basal Eurasian rich group survived and they were the researchers they would be strongly inclined to define themselves as West Eurasians and the rest as East Eurasians. Cladistically that makes a better sense.(not that cladistics should overrule other considerations but just saying as a matter of curiosity)

    I have no problem with researchers defining West and East Eurasians, whichever way they do as long as they don’t forget how it was defined in the first place. But I have a problem with journalists and amateur bloggers like Davidski who try to add a political dimension to all this (“Yeah we have been to America too”) justifying the conquest of Americas by Europeans.
    You may think I am hysterically sensitive but I have seen Davidski and Genetiker’s frequent outbursts “They hate us White folks” when they think researchers don’t give White people enough credit.(Genetiker’s outburst is quite often and easy to see. the first thing you see when you enter his blog on the right side. Davidski cleaned up from his wild days of being a Polish nationalist but I saw one outburst not too many months ago.)

    Read More
    • Replies: @Razib Khan
    i'm not going to engage on the bloggers issues. genitiker is a definite kook. i know davidski gets out of control sometimes too, though i've only seen it once (people bring this up all the time, did he delete all that stuff as there are never links?). though the fact that he has a free for all in his comments doesn't help.

    as for the distinction btwn west and east eurasians. west is definitely paraphyletic. but, it does seem that the back migration from east eurasia, broadly conceived, beyond india has been very modest. (some to europe too) i'm pretty sure that this has to due to with lots of population replacement in east asia that occurred recently (people have been hinting at me about this in the research community for a while, but wait until fu cleans up this topic).

    Malta boy is West Eurasian not because he was clearly born as one but because of the way later generations intermingled and interbred. If some Basal Eurasian rich group survived and they were the researchers they would be strongly inclined to define themselves as West Eurasians and the rest as East Eurasians. Cladistically that makes a better sense.(not that cladistics should overrule other considerations but just saying as a matter of curiosity)

    ANE shares more drift with WHG. though it's only 'west eurasian' in a technical sense. the divergence was pretty soon after the 'out of africa' diversification.

    i think the relatively deep divergence in some eurasian threads is because most of the 'intervening' ice age pops geographically may not have been leaving descendants. though there may be some lacunae in our ancient DNA to add some layers of complexity. the ancestry was 'reticulate' from an early stage.

    , @Shaikorth
    "He suggested that it is one possibility to consider Basal Eurasians as the true West Eurasians but decided otherwise(current form)."

    Since then genomes from Natufians and Iranian Neolithics have been studied, and if you take all WHG- and EHG-related ancestry out of those the result will be something that shouldn't resemble Paleolithic or modern West (or East) Europeans than they resemble each other. So with the benefit of hindsight probably a right decision.

  7. @Razib Khan
    the chinese are not a good comparison with japanese or koreans. there is some noticeable structure within chinese, including obviously relatively recently sinicized subcomponents in the 1000 genomes sample.

    i'm not sure koreans are more inbred than japanese.

    reading through the comments i'm starting to wonder if my friends were early adopters of 23andMe (they were), and were assigned using an older method whose results were frozen.

    that being said, if 23andMe's methods can correctly assign 'chinese' ancestry to people (and it does), then i don't see why it can't for korean, as they are more homogeneous.

    ultimately, the details are in the methods. the data though is probably sufficient.

    Hong is rather typical for a South Korean.
    But I could see 3 anomalies.

    1. She is the only Korean sample whose “Chinese” is greater than “Japanese” even though by a very slight margin.
    Most South Koreans have >2 ratio in favor of “Japanese”.

    2. Only one with a non-zero SE Asian even though it is only 0.1 percent. All 40 Koreans(I checked up to about 15 and then got bored) I have access to have 0 there. But this could be due to the design of the new test.

    3. Only one with non-zero Oceanian. But this again could be due to the new test.

    About 70 percent of Koreans have <1 but non-zero "European" component but Hong had none. This again could be due to the design of the new test.

    I have seen a Chinese from Northwest who had a fairly substantial amount of "European", like 5 percent. I also saw a Chinese with a "Korean" component.But most Chinese had nearly 100 percent of themselves with occasionally substantial SE Asian.

    Chinese components among Koreans increase dramatically as you move from "conservative" to "speculative". This makes sense since Chinese and Koreans share prehistoric common ancestry but have been isolated from each other for a long time but the migration from Korea to Japan happened at the edge of historic times thus longer shared segments.

    P.S. The chromosome graph does not show actual shared segments. For instance a long Chinese segment actually is made up of many smaller segments that happened to be assigned to "Chinese". Apparently they pull up every Chinese segment into one of the homologous pair and the gap between those segments is painted as "Chinese". I think this is how a full blooded Korean gets an entire chromosome assigned to Chinese or Japanese.

    Read More
  8. @Hector
    I was really going to abide by my promise not to post here but just one more.

    This analysis very nicely illustrates what has been known to many; a known mixed population can superficially look "pure" by the (usually bad) choice of reference populations. In this case N_Han, a well known mixed population independently confirmed by many disciplines, looks "pure".

    Japanese also harbor Jomon and Yayoi components, roughly speaking. But Jomon component appears as just Japanese because in this choice of population set it is all Japanese. Yayoi left descendants both in Korea and Japan but the migration was so massive that Japanese have more of them(>60 percent) than Koreans(as little as 20 percent, probably close to 30 percent) so they would be assigned as "Japanese" and make up nearly all of "Japanese component" among Koreans.

    Curiously this fits very well with Alexander Vovin's thesis that Southern and Central Koreans originally spoke a Japonic language(s) preserved in place names etc. which have been the source of bizarre Koguryo-Japonic hypothesis recently advocated(actually has a long history in linguistic circles) by non-linguists such as Chris Beckwith and Jared Diamond. Vovin and Unger take the opposite view and according to them Korean is the intrusive language imposed on Japonic speaking original inhabitants.

    This also reminds us of an important caveat. A mixed population will be between the two parental populations in the PCA analysis but that a population is between two does not mean that the population is a mixture of the two.

    Malta boy generally regarded as "pure ANE" by many amateur bloggers also may have a very complicated genetic history in a similar fashion. Bifurcation into West and East Eurasians is a statistical concept and the crudest approximation. It is absurd to pigeonhole every ancient sample as East or West Eurasian because the actual history is far more complicated.

    Curiously this fits very well with Alexander Vovin’s thesis that Southern and Central Koreans originally spoke a Japonic language(s) preserved in place names etc. which have been the source of bizarre Koguryo-Japonic hypothesis recently advocated(actually has a long history in linguistic circles) by non-linguists such as Chris Beckwith and Jared Diamond. Vovin and Unger take the opposite view and according to them Korean is the intrusive language imposed on Japonic speaking original inhabitants.

    I am favorably disposed toward this view. It seems to me that given the extensive contacts between the Gaya Confederacy (or Mimana Command Post) and Yamato Japan (and later that between Baekche and Japan) were more than simply political or economic in nature (there were significant intermarriages between ruling houses), there must have been strong cultural and probably some residual ethnic similarities. I think the rice agriculture patterns were very similar between these two groups.

    It seems highly plausible to me that the people who formed Jin and later Samhan in central and southern Korean peninsula were either the same as or related to those who founded Yamato, and that both spoke variants of what some scholars called Japonic languages.

    Meanwhile the people who formed Goguryeo were of the mixed Buyeo-Gojoseon stock, and originated from what is today southern Manchuria. They were probably related to the Malgal/Mohe tribal peoples whom Goguryeo later subjected (Mohe cavalry frequently appears as Goguryeo auxiliaries in its wars). This stock of people were probably the ones who brought the “intrusive” so-called Koreanic languages (Buyeo, Goguryeo, and later Silla variants) to the south of the peninsula.

    The problem, from a historiographical perspective, is that this topic is highly charged politically in the region. Everytime some ancient polity is “claimed” by a modern nation-state as its own (e.g. Goguryeo as a northern “non-Han” Chinese as opposed to a Korean polity), others “flip out” and all heck breaks loose politically. Gaya/Mimana is another classic example – it was peopled by a population that probably shared a strong affinity with the Yamato, at a time when neither “Korea” nor “Japan” as such existed. But because of intense nationalism in the region, it’s difficult for the local historians to take objective views on the matter, which leaves outsiders such as Gina Barnes* to strike a more reasonable and scientifically consistent note.

    Professor Barnes, in my view, is the best authority in the West on the origins of ancient Korean and Japanese polity formations: https://www.soas.ac.uk/staff/staff30624.php

    Her books: https://www.amazon.com/s/ref=nb_sb_noss?url=search-alias%3Daps&field-keywords=gina+l+barnes

    Read More
  9. Razib, is there any tendency for the Korean group to break off in PC3? The Korean population often seems to show in large scale East Asian PCA as an extension of the Chinese North-South cline with some separation from Japanese (e.g. in that PCA Shaikorth posted in the other thread, the Japanese were off cline an extension beyond China-Korea by being shifted towards Ryukyuans and Ainu).

    On the 23andme topic, they do offer some description – https://www.23andme.com/en-gb/ancestry_composition_guide/ . They discuss a tradeoff between “recall” – “Recall” corresponds to the question “Of the pieces of DNA that actually were from population A, how often did the system predict that they were from population A?” – and “precision” – “Precision” corresponds to the question “When the system predicts that a piece of DNA is from population A, how often is it actually from population A?”. They also give the rates for these for populations.

    Recall is somewhat lower for the Korean grouping as opposed to other East Asian groups (so presumably going by their definition some underassignment of Korean ancestry). However the high recall and precision rates for Chinese and Japanese would seem to mean they wouldn’t have experienced problems with Chinese and Japanese segments being falsely called (though perhaps % depends on the sample composition of their database).

    I would have thought that non-genuine Chinese / Japanese ancestry would have been assigned to the generic East Asian category, rather than being specifically called. For commentators with a history of looking at various Koreans results, where they tend to be lower than 100% Korean, is the remaining assignment usually to East Asian or specifically Chinese and Japanese?

    Read More
  10. Interesting historiographical perspective that gives the views from both sides

    https://en.wikipedia.org/wiki/Jizi

    There is a small sample of Korean yDNA data at FTDna. Koreans with surname Kim/Jin seem to be all yHg C and Kim’s are about 25% of the population. The southern provinces away from China seems to have more yHg C as well. That could have differentiated the Korean from the Chinese. The Kim’s could be descended from the ancient southern Korean Jin Dynasty. The Korean also have a small fraction of yHg D2 which is virtually absent in China.

    Korean and Hokkien words that are pronounced similarly https://www.youtube.com/watch?v=yB-c9fcRKe8

    Read More
    • Replies: @Anonymous
    "Korean and Hokkien words that are pronounced similarly https://www.youtube.com/watch?v=yB-c9fcRKe8"

    What's your point? You really seem to like these no-context interjections but I doubt your understanding of historical linguistics goes much beyond an amateur's infatuation with surface resemblance.

    Korean and Sinitic are genetically (in the linguistic sense) unrelated taxa. A large segment of the Korean lexicon was loaned from Chinese (though unweighted fractions are inflated by low-frequency specialist terminology in much the same way as for Latin forms in English). Sino-Korean frequently "froze" older phonological qualities that were better preserved to the present day in southern Chinese languages and topolects. It's plesiomorphy. Some of the deviance of northern Chinese languages is attributable to contact with (and acquisition by speakers of) non-Sinitic languages.

  11. I think I located the source of this “Korean controversy” at 23andme.

    As I suspected, 23andme chip(a custom made illumina chip as I hear) only types for SNPs and determines which of the two homologous chromosomes it came from via a “phasing” process.
    As they admitted themselves, they don’t do recalibration every time a new sample arrives so it is most likely that they started out with far fewer reference samples than 76 Korean samples they currently have.

    This means in the phasing process, Korean sequences are rearranged to fit either Chinese or Japanese since their numbers overwhelm that of Koreans. This partially explains why even at 15 percent several whole chromosomes(from a Korean) are assigned as Japanese or Chinese.

    Also precision and recall considerations (whose link Matt provided) quite clearly favor larger sampled populations. Under the very likely assumption that Japanese reference samples massively out-number Koreans’(at least initially) and that most Koreans have less than 30 percent of “Japanese segments”, 98 percent precision rate for Japanese is very explainable.

    Comment Number 2 was going to be my final post here but I kind of got worked up and wrote 3 more including this one and 2 others Razib deleted.

    Read More
    • Replies: @Rick
    I believe I already suggested that phasing may be a major issue.

    And as I said before, this potential issue can be almost completely ruled out if there are people with a parent/child relationship who all have done 23andme and linked their accounts as parent and child.

    23andme could actually just let computer programs check and improve the phases of everyone's genomes by automatically phasing everyone against anyone assigned as 2nd cousin or closer based on the amount of sharing.

    They could use only their own info from that set, and it would probably be much better than any statistical haplotype dataset anywhere else.

    But they don't seem to be doing that at all. How can you tell? Because the ancestry reports get substantially better when you link your account to a parent or child.

    I have seen this many times. Even though all people involved had been with 23andme for years, only when they manually linked their accounts did the phasing change.

    And it's not a computing problem. They already spend all the computing to do family finder and show you the shared regions. All they have extra to do is phase shared regions off of each other, and then use those new statistics to phase all close relatives at all allowable regions. Using only close relatives, this is extremely accurate.

    If they limited it even to shared regions longer than 20cM, it would be extremely fast and make it all much more accurate.

  12. @Hector
    From the way Eurasian continent is shaped(relatively long East to West dimension, especially when you only consider habitable zones), no matter how complicated the population history is, statistics will generally give two components, East and West. That does not mean all populations, even roughly, diverged into two kinds in a cladistically coherent manner.

    In the way it is defined currently West Eurasians are a paraphyletic group and East and West Eurasians are not coherent cladistic groups. Lazaridis himself was ambivalent about how to define "West Eurasians". He suggested that it is one possibility to consider Basal Eurasians as the true West Eurasians but decided otherwise(current form).

    Malta boy is West Eurasian not because he was clearly born as one but because of the way later generations intermingled and interbred. If some Basal Eurasian rich group survived and they were the researchers they would be strongly inclined to define themselves as West Eurasians and the rest as East Eurasians. Cladistically that makes a better sense.(not that cladistics should overrule other considerations but just saying as a matter of curiosity)

    I have no problem with researchers defining West and East Eurasians, whichever way they do as long as they don't forget how it was defined in the first place. But I have a problem with journalists and amateur bloggers like Davidski who try to add a political dimension to all this ("Yeah we have been to America too") justifying the conquest of Americas by Europeans.
    You may think I am hysterically sensitive but I have seen Davidski and Genetiker's frequent outbursts "They hate us White folks" when they think researchers don't give White people enough credit.(Genetiker's outburst is quite often and easy to see. the first thing you see when you enter his blog on the right side. Davidski cleaned up from his wild days of being a Polish nationalist but I saw one outburst not too many months ago.)

    i’m not going to engage on the bloggers issues. genitiker is a definite kook. i know davidski gets out of control sometimes too, though i’ve only seen it once (people bring this up all the time, did he delete all that stuff as there are never links?). though the fact that he has a free for all in his comments doesn’t help.

    as for the distinction btwn west and east eurasians. west is definitely paraphyletic. but, it does seem that the back migration from east eurasia, broadly conceived, beyond india has been very modest. (some to europe too) i’m pretty sure that this has to due to with lots of population replacement in east asia that occurred recently (people have been hinting at me about this in the research community for a while, but wait until fu cleans up this topic).

    Malta boy is West Eurasian not because he was clearly born as one but because of the way later generations intermingled and interbred. If some Basal Eurasian rich group survived and they were the researchers they would be strongly inclined to define themselves as West Eurasians and the rest as East Eurasians. Cladistically that makes a better sense.(not that cladistics should overrule other considerations but just saying as a matter of curiosity)

    ANE shares more drift with WHG. though it’s only ‘west eurasian’ in a technical sense. the divergence was pretty soon after the ‘out of africa’ diversification.

    i think the relatively deep divergence in some eurasian threads is because most of the ‘intervening’ ice age pops geographically may not have been leaving descendants. though there may be some lacunae in our ancient DNA to add some layers of complexity. the ancestry was ‘reticulate’ from an early stage.

    Read More
  13. @Hector
    I think I located the source of this "Korean controversy" at 23andme.

    As I suspected, 23andme chip(a custom made illumina chip as I hear) only types for SNPs and determines which of the two homologous chromosomes it came from via a "phasing" process.
    As they admitted themselves, they don't do recalibration every time a new sample arrives so it is most likely that they started out with far fewer reference samples than 76 Korean samples they currently have.

    This means in the phasing process, Korean sequences are rearranged to fit either Chinese or Japanese since their numbers overwhelm that of Koreans. This partially explains why even at 15 percent several whole chromosomes(from a Korean) are assigned as Japanese or Chinese.

    Also precision and recall considerations (whose link Matt provided) quite clearly favor larger sampled populations. Under the very likely assumption that Japanese reference samples massively out-number Koreans'(at least initially) and that most Koreans have less than 30 percent of "Japanese segments", 98 percent precision rate for Japanese is very explainable.

    Comment Number 2 was going to be my final post here but I kind of got worked up and wrote 3 more including this one and 2 others Razib deleted.

    I believe I already suggested that phasing may be a major issue.

    And as I said before, this potential issue can be almost completely ruled out if there are people with a parent/child relationship who all have done 23andme and linked their accounts as parent and child.

    23andme could actually just let computer programs check and improve the phases of everyone’s genomes by automatically phasing everyone against anyone assigned as 2nd cousin or closer based on the amount of sharing.

    They could use only their own info from that set, and it would probably be much better than any statistical haplotype dataset anywhere else.

    But they don’t seem to be doing that at all. How can you tell? Because the ancestry reports get substantially better when you link your account to a parent or child.

    I have seen this many times. Even though all people involved had been with 23andme for years, only when they manually linked their accounts did the phasing change.

    And it’s not a computing problem. They already spend all the computing to do family finder and show you the shared regions. All they have extra to do is phase shared regions off of each other, and then use those new statistics to phase all close relatives at all allowable regions. Using only close relatives, this is extremely accurate.

    If they limited it even to shared regions longer than 20cM, it would be extremely fast and make it all much more accurate.

    Read More
  14. Anonymous says:     Show CommentNext New Comment
    @dux.ie
    Interesting historiographical perspective that gives the views from both sides

    https://en.wikipedia.org/wiki/Jizi

    There is a small sample of Korean yDNA data at FTDna. Koreans with surname Kim/Jin seem to be all yHg C and Kim's are about 25% of the population. The southern provinces away from China seems to have more yHg C as well. That could have differentiated the Korean from the Chinese. The Kim's could be descended from the ancient southern Korean Jin Dynasty. The Korean also have a small fraction of yHg D2 which is virtually absent in China.

    Korean and Hokkien words that are pronounced similarly https://www.youtube.com/watch?v=yB-c9fcRKe8

    “Korean and Hokkien words that are pronounced similarly https://www.youtube.com/watch?v=yB-c9fcRKe8&#8243;

    What’s your point? You really seem to like these no-context interjections but I doubt your understanding of historical linguistics goes much beyond an amateur’s infatuation with surface resemblance.

    Korean and Sinitic are genetically (in the linguistic sense) unrelated taxa. A large segment of the Korean lexicon was loaned from Chinese (though unweighted fractions are inflated by low-frequency specialist terminology in much the same way as for Latin forms in English). Sino-Korean frequently “froze” older phonological qualities that were better preserved to the present day in southern Chinese languages and topolects. It’s plesiomorphy. Some of the deviance of northern Chinese languages is attributable to contact with (and acquisition by speakers of) non-Sinitic languages.

    Read More
    • Replies: @Twinkie

    Korean and Sinitic are genetically (in the linguistic sense) unrelated taxa. A large segment of the Korean lexicon was loaned from Chinese (though unweighted fractions are inflated by low-frequency specialist terminology in much the same way as for Latin forms in English). Sino-Korean frequently “froze” older phonological qualities that were better preserved to the present day in southern Chinese languages and topolects. It’s plesiomorphy.
     
    Yes! Only the ill-informed would draw any conclusions from "similar pronunciations."

    To the extent the language isolate that is Korean is genetically related to any language group, it is probably Tungusic, but even then it might be sprachbund rather than any significant genetic relationship.

    As for the comparison to Latin in English, yes, the structure is similar ("low-frequency specialist terminology"), but in terms of percentages, Chinese loan words are far more numerous in Korean than Latin is in English. You can have a functional spoken American English without Latin-derived words; modern Korean without Chinese loan words would be pretty unintelligible.
    , @dux.ie
    They are similar because they were influenced from the same source. Did I
    have any other points except that they are pronounced the same ??

    https://en.wikipedia.org/wiki/Tang_dynasty#Establishment

    """The Tang Emperors also had Xianbei maternal ancestry,[15][16]
    from Emperor Gaozu of Tang's Xianbei mother Duchess Dugu (and probably
    from other earlier maternal ancestors as well)."""

    The rime dictionary for pronunciation of Sui/Tang Chinese characters was
    compiled by someone with Hu/Xianbei descents, with heavy influence from
    the Hu/Xianbei regions,

    https://en.wikipedia.org/wiki/Qieyun

    It is not the first time that the Chinese absorbed and integrated outside
    sources and claimed as their own.
  15. @Hector
    From the way Eurasian continent is shaped(relatively long East to West dimension, especially when you only consider habitable zones), no matter how complicated the population history is, statistics will generally give two components, East and West. That does not mean all populations, even roughly, diverged into two kinds in a cladistically coherent manner.

    In the way it is defined currently West Eurasians are a paraphyletic group and East and West Eurasians are not coherent cladistic groups. Lazaridis himself was ambivalent about how to define "West Eurasians". He suggested that it is one possibility to consider Basal Eurasians as the true West Eurasians but decided otherwise(current form).

    Malta boy is West Eurasian not because he was clearly born as one but because of the way later generations intermingled and interbred. If some Basal Eurasian rich group survived and they were the researchers they would be strongly inclined to define themselves as West Eurasians and the rest as East Eurasians. Cladistically that makes a better sense.(not that cladistics should overrule other considerations but just saying as a matter of curiosity)

    I have no problem with researchers defining West and East Eurasians, whichever way they do as long as they don't forget how it was defined in the first place. But I have a problem with journalists and amateur bloggers like Davidski who try to add a political dimension to all this ("Yeah we have been to America too") justifying the conquest of Americas by Europeans.
    You may think I am hysterically sensitive but I have seen Davidski and Genetiker's frequent outbursts "They hate us White folks" when they think researchers don't give White people enough credit.(Genetiker's outburst is quite often and easy to see. the first thing you see when you enter his blog on the right side. Davidski cleaned up from his wild days of being a Polish nationalist but I saw one outburst not too many months ago.)

    “He suggested that it is one possibility to consider Basal Eurasians as the true West Eurasians but decided otherwise(current form).”

    Since then genomes from Natufians and Iranian Neolithics have been studied, and if you take all WHG- and EHG-related ancestry out of those the result will be something that shouldn’t resemble Paleolithic or modern West (or East) Europeans than they resemble each other. So with the benefit of hindsight probably a right decision.

    Read More
    • Replies: @Hector
    Since then genomes from Natufians and Iranian Neolithics have been studied, and if you take all WHG- and EHG-related ancestry out of those the result will be something that shouldn’t resemble Paleolithic or modern West (or East) Europeans than they resemble each other. So with the benefit of hindsight probably a right decision

    Actually we don't know yet the precise position of Iranian Basal and Near Eastern Basal Eurasians.
    They actually may be paraphyletic as well; ie. one is closer to the main Eurasian branch than the other is.

    And what you are referring to is accidents of history, not phylogenetic affinity.
    If by accidents of history all WHG and ANE heavy populations were relegated to a remote island in Western Europe and the rest of Western Eurasia is dominated by Basal rich populations you would see a different classification.

    Your point probably includes the fact there are nearly Basal free ancient West Eurasian specimens while all, modern or ancient, harbor significant ANE-WHG related ancestries.
    But that too could be accidents of history and/or archaeological conditions related to climates and the ethnic origins of researchers. We don't have as many non-European ancient samples for various reasons.

    Classification is generally subject to modification occasionally when new (shocking) finds are made. However the current definition of East-West Eurasians is too dependent on future finds.
    We want something more stable than that.

    All modern West Eurasian populations harbor Basal elements at least in "appreciable amounts".
    Moreover it reaches nearly 50 percent in some populations. You cannot dismiss it as insignificant(and probably embarrassing) outliers.

    Davidski recently tried a new approach where he made East Eurasians or East Asians hybrids while preserving West Eurasian group as a cladistically coherent descent group. He failed and botched the attempt as far as I know. But I have known him from the days of DNA-forums and I knew why he was doing that.
  16. @Shaikorth
    "He suggested that it is one possibility to consider Basal Eurasians as the true West Eurasians but decided otherwise(current form)."

    Since then genomes from Natufians and Iranian Neolithics have been studied, and if you take all WHG- and EHG-related ancestry out of those the result will be something that shouldn't resemble Paleolithic or modern West (or East) Europeans than they resemble each other. So with the benefit of hindsight probably a right decision.

    Since then genomes from Natufians and Iranian Neolithics have been studied, and if you take all WHG- and EHG-related ancestry out of those the result will be something that shouldn’t resemble Paleolithic or modern West (or East) Europeans than they resemble each other. So with the benefit of hindsight probably a right decision

    Actually we don’t know yet the precise position of Iranian Basal and Near Eastern Basal Eurasians.
    They actually may be paraphyletic as well; ie. one is closer to the main Eurasian branch than the other is.

    And what you are referring to is accidents of history, not phylogenetic affinity.
    If by accidents of history all WHG and ANE heavy populations were relegated to a remote island in Western Europe and the rest of Western Eurasia is dominated by Basal rich populations you would see a different classification.

    Your point probably includes the fact there are nearly Basal free ancient West Eurasian specimens while all, modern or ancient, harbor significant ANE-WHG related ancestries.
    But that too could be accidents of history and/or archaeological conditions related to climates and the ethnic origins of researchers. We don’t have as many non-European ancient samples for various reasons.

    Classification is generally subject to modification occasionally when new (shocking) finds are made. However the current definition of East-West Eurasians is too dependent on future finds.
    We want something more stable than that.

    All modern West Eurasian populations harbor Basal elements at least in “appreciable amounts”.
    Moreover it reaches nearly 50 percent in some populations. You cannot dismiss it as insignificant(and probably embarrassing) outliers.

    Davidski recently tried a new approach where he made East Eurasians or East Asians hybrids while preserving West Eurasian group as a cladistically coherent descent group. He failed and botched the attempt as far as I know. But I have known him from the days of DNA-forums and I knew why he was doing that.

    Read More
    • Replies: @Matt_
    All modern West Eurasian populations harbor Basal elements at least in “appreciable amounts”.

    Moreover it reaches nearly 50 percent in some populations. You cannot dismiss it as insignificant(and probably embarrassing) outliers.

    For the most recent estimates, Lazaridis's latest methods don't detail recent people, but do give recent Iranians at 30% (http://www.nature.com/nature/journal/v536/n7617/images_article/nature19310-f2.jpg). They are probably at the high end for recent people. Generally the Neolithic and post-Neolithic populations of West Eurasia seem to mostly vary over the 30-20% interval, from Iran-Europe (except for some Neolithic Iranians, which seems odd). Those estimates are the best there are atm I guess. I don't think we necessarily have a good way to measure how much ancestry from this (still unproven) population(s) yet.

    Davidski recently tried a new approach where he made East Eurasians or East Asians hybrids while preserving West Eurasian group as a cladistically coherent descent group.

    Those models (with various admixture edges to a clade consisting of all ENA and Ust Ishim) were quite unusual. The Lazaridis paper has models where East Asians+Siberians vary in a 10-30+% interval in contribution from the ANE group though. Will have to see if those hold up over time.

  17. It’s easy enough to figure the affinities of Natufian sans WHG or Iran_N sans EHG, they’d be a lot more distant from every modern West Eurasian group. How many West Eurasian groups you can name that would be significantly closer to Natufian-minus-WHG than to Loschbour? Or even closer to Natufian than to Loschbour?

    Are you sure you’re not disputing Lazaridis’ classification of Basal based on what-ifs?
    It’s also an “accident of history” that modern West Eurasian doesn’t consist of Mota or Ust’-Ishim ancestry. It’s what it is.

    Read More
  18. @Anonymous
    "Korean and Hokkien words that are pronounced similarly https://www.youtube.com/watch?v=yB-c9fcRKe8"

    What's your point? You really seem to like these no-context interjections but I doubt your understanding of historical linguistics goes much beyond an amateur's infatuation with surface resemblance.

    Korean and Sinitic are genetically (in the linguistic sense) unrelated taxa. A large segment of the Korean lexicon was loaned from Chinese (though unweighted fractions are inflated by low-frequency specialist terminology in much the same way as for Latin forms in English). Sino-Korean frequently "froze" older phonological qualities that were better preserved to the present day in southern Chinese languages and topolects. It's plesiomorphy. Some of the deviance of northern Chinese languages is attributable to contact with (and acquisition by speakers of) non-Sinitic languages.

    Korean and Sinitic are genetically (in the linguistic sense) unrelated taxa. A large segment of the Korean lexicon was loaned from Chinese (though unweighted fractions are inflated by low-frequency specialist terminology in much the same way as for Latin forms in English). Sino-Korean frequently “froze” older phonological qualities that were better preserved to the present day in southern Chinese languages and topolects. It’s plesiomorphy.

    Yes! Only the ill-informed would draw any conclusions from “similar pronunciations.”

    To the extent the language isolate that is Korean is genetically related to any language group, it is probably Tungusic, but even then it might be sprachbund rather than any significant genetic relationship.

    As for the comparison to Latin in English, yes, the structure is similar (“low-frequency specialist terminology”), but in terms of percentages, Chinese loan words are far more numerous in Korean than Latin is in English. You can have a functional spoken American English without Latin-derived words; modern Korean without Chinese loan words would be pretty unintelligible.

    Read More
  19. Some years ago, I was drinking beer in a train compartment coming down from north Russia to Moscow. My companion was one of our translators, and he related how his father, a noted linguist, was persona-non-grata in Japan because he posited a theory that the Japanese and their language were derived from Korean.

    Read More
  20. @Anonymous
    "Korean and Hokkien words that are pronounced similarly https://www.youtube.com/watch?v=yB-c9fcRKe8"

    What's your point? You really seem to like these no-context interjections but I doubt your understanding of historical linguistics goes much beyond an amateur's infatuation with surface resemblance.

    Korean and Sinitic are genetically (in the linguistic sense) unrelated taxa. A large segment of the Korean lexicon was loaned from Chinese (though unweighted fractions are inflated by low-frequency specialist terminology in much the same way as for Latin forms in English). Sino-Korean frequently "froze" older phonological qualities that were better preserved to the present day in southern Chinese languages and topolects. It's plesiomorphy. Some of the deviance of northern Chinese languages is attributable to contact with (and acquisition by speakers of) non-Sinitic languages.

    They are similar because they were influenced from the same source. Did I
    have any other points except that they are pronounced the same ??

    https://en.wikipedia.org/wiki/Tang_dynasty#Establishment

    “””The Tang Emperors also had Xianbei maternal ancestry,[15][16]
    from Emperor Gaozu of Tang’s Xianbei mother Duchess Dugu (and probably
    from other earlier maternal ancestors as well).”””

    The rime dictionary for pronunciation of Sui/Tang Chinese characters was
    compiled by someone with Hu/Xianbei descents, with heavy influence from
    the Hu/Xianbei regions,

    https://en.wikipedia.org/wiki/Qieyun

    It is not the first time that the Chinese absorbed and integrated outside
    sources and claimed as their own.

    Read More
  21. @Hector
    Since then genomes from Natufians and Iranian Neolithics have been studied, and if you take all WHG- and EHG-related ancestry out of those the result will be something that shouldn’t resemble Paleolithic or modern West (or East) Europeans than they resemble each other. So with the benefit of hindsight probably a right decision

    Actually we don't know yet the precise position of Iranian Basal and Near Eastern Basal Eurasians.
    They actually may be paraphyletic as well; ie. one is closer to the main Eurasian branch than the other is.

    And what you are referring to is accidents of history, not phylogenetic affinity.
    If by accidents of history all WHG and ANE heavy populations were relegated to a remote island in Western Europe and the rest of Western Eurasia is dominated by Basal rich populations you would see a different classification.

    Your point probably includes the fact there are nearly Basal free ancient West Eurasian specimens while all, modern or ancient, harbor significant ANE-WHG related ancestries.
    But that too could be accidents of history and/or archaeological conditions related to climates and the ethnic origins of researchers. We don't have as many non-European ancient samples for various reasons.

    Classification is generally subject to modification occasionally when new (shocking) finds are made. However the current definition of East-West Eurasians is too dependent on future finds.
    We want something more stable than that.

    All modern West Eurasian populations harbor Basal elements at least in "appreciable amounts".
    Moreover it reaches nearly 50 percent in some populations. You cannot dismiss it as insignificant(and probably embarrassing) outliers.

    Davidski recently tried a new approach where he made East Eurasians or East Asians hybrids while preserving West Eurasian group as a cladistically coherent descent group. He failed and botched the attempt as far as I know. But I have known him from the days of DNA-forums and I knew why he was doing that.

    All modern West Eurasian populations harbor Basal elements at least in “appreciable amounts”.

    Moreover it reaches nearly 50 percent in some populations. You cannot dismiss it as insignificant(and probably embarrassing) outliers.

    For the most recent estimates, Lazaridis’s latest methods don’t detail recent people, but do give recent Iranians at 30% (http://www.nature.com/nature/journal/v536/n7617/images_article/nature19310-f2.jpg). They are probably at the high end for recent people. Generally the Neolithic and post-Neolithic populations of West Eurasia seem to mostly vary over the 30-20% interval, from Iran-Europe (except for some Neolithic Iranians, which seems odd). Those estimates are the best there are atm I guess. I don’t think we necessarily have a good way to measure how much ancestry from this (still unproven) population(s) yet.

    Davidski recently tried a new approach where he made East Eurasians or East Asians hybrids while preserving West Eurasian group as a cladistically coherent descent group.

    Those models (with various admixture edges to a clade consisting of all ENA and Ust Ishim) were quite unusual. The Lazaridis paper has models where East Asians+Siberians vary in a 10-30+% interval in contribution from the ANE group though. Will have to see if those hold up over time.

    Read More

Comments are closed.