Page 629 of 648 FirstFirst ... 129529579619627628629630631639 ... LastLast
Results 6,281 to 6,290 of 6475

Thread: Could Western Jews (Ash. and Seph.) descend from Aegeans and Levantine admixture?

  1. #6281
    Registered Users
    Posts
    241
    Ethnicity
    Jewish

    Quote Originally Posted by eolien View Post
    Some of you wonder what all these have to do with Jews. First of all, any person with scientific background knows that any results without controls doesn't mean anything. Controls can be negative and positive controls, in this case the variance about the individual samples show clear pattern about basic components of MENA populations. I think this is not by chance. Now using this panel, i will try to address the composition of jewish populations in G25 database.

    First, for the Western Jews, as you can immediately see the fits are not great because it is missing an important component from Europe. Second it is clear that Guanche and some Chen are preferred over Mozabite. This is clearly different from all the Egyptian and some palestinian samples. Third, Greek Central Anatolian has the biggest component whereas there is no affinity for Armenian/Greek Trabzon which IMO shows that Greek Central Anatolian is the proxy for the local anatolian populations. however this clean , clear cut but low fit picture collapses when we add different european populations

     

    Then why didn't you stop when the Lebanese samples began to show untenably low levels of Levantine ancestry?

  2. The Following 5 Users Say Thank You to StillWater For This Useful Post:

     Agamemnon (07-11-2019),  Erikl86 (07-11-2019),  hartaisarlag (07-11-2019),  Power77 (07-17-2019),  Targum (07-11-2019)

  3. #6282
    Gold Class Member
    Posts
    2,098
    Sex
    Ethnicity
    Ashkenazi Jew
    Y-DNA
    Q-YP3924
    mtDNA
    K1a1b1a

    Israel Israel Jerusalem
    Quote Originally Posted by Claudio View Post
    Erikl86.
    What’s your opinion on this conundrum above?
    I actually don't think that Romaniote and Italian Jews have absorbed much European ancestry that late, because in both places Christianity was the official religion by that time.
    Italian Jews might have managed to get additional North Italian admixture during Arian Lombard rule which permitted Chalcedonian Christian slaves for a while before adopting it themselves, but I can't see any Greeks in 9th century Byzantine Empire converting to Judaism.

    Ashkenazi Jews managed to get additional North European admixture most likely because of a rare coincidence of the Reformation which was an event which shakened the beliefs of Christians and the relative tolerance that Jews enjoyed in the Polish Lithuanian Commonwealth. So it's not like this was always happening.

    In fact, the possibility that Mizrahi Babylonian Jews settled and mixed with Western Jews during the Geonic period, makes it more possible that both Romaniote AND Italki Jews have actually became MORE Levantine/Near Eastern.
    Check out my Hidden Content
    My Y-DNA: Q-M242 -> Q-L232 -> Q-L275 -> Q-M378 -> Q-Y2016 -> Q-L245 -> Q-FGC1904 -> Q-Y2209 -> Q-Y2225 -> Q-Y2197 -> Q-Y2750 -> Q-YP1004 -> Q-YP3924;
    My mtDNA: K1a1b1a;

    My dad's mtDNA: K2a2a;

  4. The Following 12 Users Say Thank You to Erikl86 For This Useful Post:

     Agamemnon (07-11-2019),  Andrewid (07-11-2019),  Claudio (07-11-2019),  hartaisarlag (07-11-2019),  John Doe (07-14-2019),  jonahst (07-11-2019),  Power77 (07-17-2019),  Principe (07-11-2019),  Ruderico (07-11-2019),  Seabass (07-13-2019),  StillWater (07-11-2019),  Targum (07-11-2019)

  5. #6283
    Gold Class Member
    Posts
    2,098
    Sex
    Ethnicity
    Ashkenazi Jew
    Y-DNA
    Q-YP3924
    mtDNA
    K1a1b1a

    Israel Israel Jerusalem
    Quote Originally Posted by jonahst View Post
    Eolian, these models don't make any sense. You're using multiple populations with similar ancestries. This is why Ashkenazim, Iranian Jews, and many of the other individual samples get so little Samaritan. This isn't because they lack Levantine ancestry.

    Also, you can filter out outliers with the web app. I don't think using individual references from the same endogamous populations is useful when they vary so dramatically.
    You beat me to it. These models just make no sense, I don't know how to even begin tackling them.

    Also why to use so many modern populations when we have several ancient ones.
    Check out my Hidden Content
    My Y-DNA: Q-M242 -> Q-L232 -> Q-L275 -> Q-M378 -> Q-Y2016 -> Q-L245 -> Q-FGC1904 -> Q-Y2209 -> Q-Y2225 -> Q-Y2197 -> Q-Y2750 -> Q-YP1004 -> Q-YP3924;
    My mtDNA: K1a1b1a;

    My dad's mtDNA: K2a2a;

  6. The Following 6 Users Say Thank You to Erikl86 For This Useful Post:

     Agamemnon (07-11-2019),  Andrewid (07-11-2019),  jonahst (07-11-2019),  Power77 (07-17-2019),  Ruderico (07-11-2019),  Targum (07-11-2019)

  7. #6284
    Registered Users
    Posts
    314
    Sex
    Location
    Europe
    Ethnicity
    Sephardi
    Nationality
    Turkish
    Y-DNA
    E-M84
    mtDNA
    H14b

    Quote Originally Posted by Erikl86 View Post
    You beat me to it. These models just make no sense, I don't know how to even begin tackling them.

    Also why to use so many modern populations when we have several ancient ones.
    Which populations have same ancestry? I don't understand the remark. Every population has some common ancestry since we all come from Africa. The difference is the time point and if nmonte prefers an assyrian sample over a samaritan, it has a reason for that choice. The web nMonte chooses individuals from the reference populations. Therefore when an armenian is chosen on top of assyrian, it means it is catching something that assyrians don't have such as real south caucasian.

    Quote Originally Posted by eolien View Post
    continuing:

    Replacing Abruzzo with Maltese does not change much in our small Levantine panel, for Western jews it often worsens the fit and changes other components without affecting Samaritan per se much.



    Removing Armenian and Greek Trabzon from reference populations and instead relying on Assyrian and Central Greek Anatolia for those Armenian/South Caucasian like components (as suggested above by Seabass), to my surprise does not change the fitness and cleans up the panel. See how poor the Ashkenazi score for Samaritan



     




    As you see above in this panel of populations especially the Ashkenazi score very low Samaritan but also have worse fit. Therefore as I reported below I added German and Polish to the panel. Surprisingly the fit greatly improved the Ashkenazi samples and increased their Samaritan and lowered their Abruzzo. Higher Polish component is also associated with the Eastern ashkenazi samples. It is more encouraging that Romaniotes don't have any German but more difficult to explain that some Mizrachi and North Africans do have.


     





    I think the real proof would be when we have Christian Palestinians samples from Galilee and West Bank, and run them in comparison to Lebanese samples and to Cypriotes. I expect a higher affinity of the former to Samaritans.

    I should also try Iranian and Iraqi Jews with some Persians to increase the fitness.

  8. The Following 2 Users Say Thank You to eolien For This Useful Post:

     Iseid0441 (07-12-2019),  Targum (07-11-2019)

  9. #6285
    Registered Users
    Posts
    241
    Ethnicity
    Jewish

    Quote Originally Posted by eolien View Post
    Which populations have same ancestry? I don't understand the remark. Every population has some common ancestry since we all come from Africa. The difference is the time point and if nmonte prefers an assyrian sample over a samaritan, it has a reason for that choice. The web nMonte chooses individuals from the reference populations. Therefore when an armenian is chosen on top of assyrian, it means it is catching something that assyrians don't have such as real south caucasian.
    nMonte merely chooses the combination of the reference populations you feed it to minimize distance. You can't therefore rule that Armenian is chosen due to containing a relevant element in your tested population that Assyrian is missing, as one isn't simply chosen over the other, but is chosen in the presence of all the other populations you've added to the reference panel - which is why you've been recommended to choose ancient references, as they should have less overlaps and hence not be as affected by the presence of other references.




    As you see above in this panel of populations especially the Ashkenazi score very low Samaritan but also have worse fit. Therefore as I reported below I added German and Polish to the panel. Surprisingly the fit greatly improved the Ashkenazi samples and increased their Samaritan and lowered their Abruzzo. Higher Polish component is also associated with the Eastern ashkenazi samples. It is more encouraging that Romaniotes don't have any German but more difficult to explain that some Mizrachi and North Africans do have.
    See the answer above. Your Iranian Jewish sample point likely needs an additional R1a-associated population component in the presence of your other reference populations, and hence it grabbed it from the Polish reference. What are you going to do now - add in an additional Iranian component and over fit further? Some of the Ashkenazi fits actually got worse. Also, notice that the highest Samaritan is assigned to the Ashkenazi with the highest North European. Do you think this is a coincidence? Intuitively , wouldn't we expect the opposite? Davidski says that the 7 Ashkenazi samples are fairly similar and probably the reason why he chose them. All what the Samaritan is doing for the Ashkenazi1e is balancing out the high North Euro component, and vice versa.
    Last edited by StillWater; 07-11-2019 at 08:03 PM.

  10. The Following 2 Users Say Thank You to StillWater For This Useful Post:

     Erikl86 (07-11-2019),  jonahst (07-11-2019)

  11. #6286
    Registered Users
    Posts
    736
    Sex
    Ethnicity
    Ashkenazi Jew
    Nationality
    American
    Y-DNA
    Q-Y2750
    mtDNA
    H47

    United States of America Israel
    Quote Originally Posted by eolien View Post
    Which populations have same ancestry? I don't understand the remark. Every population has some common ancestry since we all come from Africa. The difference is the time point and if nmonte prefers an assyrian sample over a samaritan, it has a reason for that choice. The web nMonte chooses individuals from the reference populations. Therefore when an armenian is chosen on top of assyrian, it means it is catching something that assyrians don't have such as real south caucasian.





    As you see above in this panel of populations especially the Ashkenazi score very low Samaritan but also have worse fit. Therefore as I reported below I added German and Polish to the panel. Surprisingly the fit greatly improved the Ashkenazi samples and increased their Samaritan and lowered their Abruzzo. Higher Polish component is also associated with the Eastern ashkenazi samples. It is more encouraging that Romaniotes don't have any German but more difficult to explain that some Mizrachi and North Africans do have.


     





    I think the real proof would be when we have Christian Palestinians samples from Galilee and West Bank, and run them in comparison to Lebanese samples and to Cypriotes. I expect a higher affinity of the former to Samaritans.

    I should also try Iranian and Iraqi Jews with some Persians to increase the fitness.
    Assyrians and BedouinA both have substantial Levantine, BedouinB probably overlaps with Levantines, and Samaritans underwent an extreme bottleneck (Lebanese Christians provide the best modern Levantine proxy for Western Jews). Why include Mozabite, Tunisian Berber, and Guanche? And Ashkenazim are not 25% German, and this inflation is clear since almost every group has individuals with substantial "German" ancestry in these models.

    The extreme variation in proportions for individuals from the same endogamous population should be a red flag.
    Last edited by jonahst; 07-11-2019 at 10:56 PM.

  12. The Following 4 Users Say Thank You to jonahst For This Useful Post:

     Claudio (07-11-2019),  Erikl86 (07-11-2019),  Ruderico (07-12-2019),  StillWater (07-11-2019)

  13. #6287
    Registered Users
    Posts
    314
    Sex
    Location
    Europe
    Ethnicity
    Sephardi
    Nationality
    Turkish
    Y-DNA
    E-M84
    mtDNA
    H14b

    Quote Originally Posted by StillWater View Post
    nMonte merely chooses the combination of the reference populations you feed it to minimize distance. You can't therefore rule that Armenian is chosen due to containing a relevant element in your tested population that Assyrian is missing, as one isn't simply chosen over the other, but is chosen in the presence of all the other populations you've added to the reference panel - which is why you've been recommended to choose ancient references, as they should have less overlaps and hence not be as affected by the presence of other references.






    See the answer above. Your Iranian Jewish sample point likely needs an additional R1a-associated population component in the presence of your other reference populations, and hence it grabbed it from the Polish reference. What are you going to do now - add in an additional Iranian component and over fit further? Some of the Ashkenazi fits actually got worse. Also, notice that the highest Samaritan is assigned to the Ashkenazi with the highest North European. Do you think this is a coincidence? Intuitively , wouldn't we expect the opposite? Davidski says that the 7 Ashkenazi samples are fairly similar and probably the reason why he chose them. All what the Samaritan is doing for the Ashkenazi1e is balancing out the high North Euro component, and vice versa.
    All the critics and suggestions are welcome. Firstly all the Ashkenazi samples get much better fit by adding German and Polish (i checked it again please tell me which one gets worse).

    Second, G25 works on 25 PCAs, so I hope at least few dimensions play a role in our samples, otherwise it would be a simple 2D situation. I also don't know if nMonte is a simple statistical fitness calculation, but I take your word for it. But I think it is more sophisticated than that. For example, although many of Romaniote, Ashkenazi, Italian and sephardic samples overlap in their positions in PCA1/2 map, according the test above Romaniote samples don't have any German or Polish and plenty of Greek Central Anatolia. I think this is not a coincidence or artifact.

    Finally, using ancient samples are easier said than done. Mixing neolithic with bronze age and middle age samples is not the best practice. But in principle it will not change the main pattern.

  14. The Following 3 Users Say Thank You to eolien For This Useful Post:

     Claudio (07-11-2019),  Ruderico (07-12-2019),  StillWater (07-11-2019)

  15. #6288
    Registered Users
    Posts
    241
    Ethnicity
    Jewish

    Quote Originally Posted by eolien View Post
    All the critics and suggestions are welcome. Firstly all the Ashkenazi samples get much better fit by adding German and Polish (i checked it again please tell me which one gets worse).
    There you're right. I must've been looking at different rows. However, is the improvement in fit sufficient given that you added 2 additional references? I suggest you use something like the AIC or BIC to keep models in check.

    Second, G25 works on 25 PCAs, so I hope at least few dimensions play a role in our samples, otherwise it would be a simple 2D situation. I also don't know if nMonte is a simple statistical fitness calculation, but I take your word for it. But I think it is more sophisticated than that. For example, although many of Romaniote, Ashkenazi, Italian and sephardic samples overlap in their positions in PCA1/2 map, according the test above Romaniote samples don't have any German or Polish and plenty of Greek Central Anatolia. I think this is not a coincidence or artifact.

    Finally, using ancient samples are easier said than done. Mixing neolithic with bronze age and middle age samples is not the best practice. But in principle it will not change the main pattern.
    You're counting the hits and ignoring the misses. Working in a 25 dimensional space doesn't mean that 5 dimensions aren't likely to result in an overfit, especially ones drawn from modern populations. Given that the model Davidski sends you with your coordinates includes populations like Kura-Araxes etc, the 25 coordinates are likely based on populations at least as old, and probably older.
    Last edited by StillWater; 07-11-2019 at 09:29 PM.

  16. The Following 2 Users Say Thank You to StillWater For This Useful Post:

     Claudio (07-11-2019),  jonahst (07-11-2019)

  17. #6289
    Registered Users
    Posts
    314
    Sex
    Location
    Europe
    Ethnicity
    Sephardi
    Nationality
    Turkish
    Y-DNA
    E-M84
    mtDNA
    H14b

    Quote Originally Posted by jonahst View Post
    Assyrians and BedouinA both had substantial Levantine, BedouinB probably overlaps with Levantines, and Samaritans underwent an extreme bottleneck (Lebanese Christians provide the best modern Levantine proxy for Western Jews). Why include Mozabite, Tunisian Berber, and Guanche? And Ashkenazim are not 25% German, and this inflation is clear since almost every group has individuals with substantial "German" ancestry in these models.

    The extreme variation in proportions for individuals from the same endogamous population should be a red flag.
    First of all, Ashkenazim are not 25% German, in my test, they got a range from 5% to 30%. But it is not impossible because we don't know the history of these samples. German here does not mean German per se because if you add Tuscan instead of Abruzzo it will decrease the German component. But i also think that the issue of bottleneck is overrated. Because we are talking about individual samples, nMonte is selecting at least in the web version (I don't know if this always the case) individuals from the indicated population. So we don't need to know the ethnic identity of the groups, we throw around 20-30 samples and nMonte chooses the best among them. It is not important if the chosen sample is called Druze or Christian. Perhaps Samaritans have less variance among themselves, but Druze were also quite endogamous and did not have a huge population size. Of course all these populations have Levantine origins whatever it means for you.

  18. The Following 2 Users Say Thank You to eolien For This Useful Post:

     Claudio (07-11-2019),  Targum (07-11-2019)

  19. #6290
    Registered Users
    Posts
    241
    Ethnicity
    Jewish

    Quote Originally Posted by eolien View Post
    First of all, Ashkenazim are not 25% German, in my test, they got a range from 5% to 30%. But it is not impossible because we don't know the history of these samples. German here does not mean German per se because if you add Tuscan instead of Abruzzo it will decrease the German component. But i also think that the issue of bottleneck is overrated. Because we are talking about individual samples, nMonte is selecting at least in the web version (I don't know if this always the case) individuals from the indicated population. So we don't need to know the ethnic identity of the groups, we throw around 20-30 samples and nMonte chooses the best among them. It is not important if the chosen sample is called Druze or Christian. Perhaps Samaritans have less variance among themselves, but Druze were also quite endogamous and did not have a huge population size. Of course all these populations have Levantine origins whatever it means for you.
    I don't quite understand what you're saying, but 25% German ancestry is near impossible, albeit in a rare case of multi-generational skewed inheritance, whereby both the Levantine and German segments are inherited disproportionally, and such that across the standard tests that Davidski looked at, this sample ended up plotting like the others. Also, do you think that :

    nMonte is selecting at least in the web version (I don't know if this always the case) individuals from the indicated population.
    because of what you see in the code output after?

  20. The Following 2 Users Say Thank You to StillWater For This Useful Post:

     hartaisarlag (07-11-2019),  jonahst (07-11-2019)

Page 629 of 648 FirstFirst ... 129529579619627628629630631639 ... LastLast

Similar Threads

  1. Replies: 19
    Last Post: 12-07-2018, 12:52 AM
  2. Present-day Lebanese descend from Biblical Canaanites
    By MikeWhalen in forum Ancient (aDNA)
    Replies: 0
    Last Post: 07-27-2017, 07:49 PM
  3. Early farmers from across Europe were direct descendants of Aegeans
    By rock hunter in forum Archaeology (Prehistory)
    Replies: 0
    Last Post: 06-21-2016, 10:43 PM
  4. Replies: 0
    Last Post: 06-14-2016, 01:02 PM
  5. Replies: 0
    Last Post: 05-26-2015, 05:29 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •