Eupedia Forums
Site NavigationEupedia Top > Eupedia Forum & Japan Forum
Page 1 of 2 12 LastLast
Results 1 to 25 of 36

Thread: Results through K15

  1. #1
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    Results through K15

    Obtuvo estos resultados a través de Eurogenes K15. Nmontes. No hay oráculo.

    Population
    North_Sea 20.35 Pct
    Atlantic 31.76 Pct
    Baltic 4.75 Pct
    Eastern_Euro 2.31 Pct
    West_Med 23.82 Pct
    West_Asian 6.42 Pct
    East_Med 3.13 Pct
    Red_Sea 3.37 Pct
    South_Asian -
    Southeast_Asian -
    Siberian 0.99 Pct
    Amerindian -
    Oceanian 0.16 Pct
    Northeast_African 2.55 Pct
    Sub-Saharan 0.41 Pct
    Last edited by Carlos; 14-03-19 at 22:25.

  2. #2
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    Nmontes

    pen0-

    [1] "distancia% = 0.4171"

    X

    French_Basque, 44.6
    Orcadian, 25.8
    Sardinian, 18.4
    Ossetian, 7.4
    Somali, 3.6
    Evens, 0.2

    pen0.001-
    [1] "distance% = 0.5924"

    X

    Spanish_Cantabria, 86.4
    French_Basque, 7.6
    Francés, 1.8
    orcadian, 1
    Southwest_English, 1
    Afghan_Uzbeki, 0.2
    Chechen, 0.2
    Ethiopian_Anuak, 0.2
    Ethiopian_Oromo, 0.2
    Hadjar, 0.2
    Kanjar, 0.2
    North_Dutch, 0.2
    North_Osetia, 0.2
    San, 0.2
    Spanish_Castilla_La_Mancha, 0.2
    Sudanese, 0.2

    pen0.01
    1 0.5738 "

    X

    Spanish_Cantabria, 89,8
    French_Basque, 5,8
    Balkar, 0,6
    francés, 0,6
    checheno, 0,4
    Southwest_English, 0,4
    Afghan_Pashtun, 0,2
    Ethiopian_Ari_cultivator, 0,2
    Evenki, 0,2
    Hadza, 0,2
    irlandés, 0,2
    North_Ossetian, 0,2
    South_Dutch, 0,2
    Southeast_English, 0,2
    Spanish_Cataluna, 0,2
    tabassaran, 0,2
    West_Norwegian , 0.2
    West_Scottish, 0.2
    Last edited by Carlos; 08-03-19 at 17:30.

  3. #3
    Anthrogenica Mod Achievements:
    1 year registered1000 Experience Points

    Join Date
    12-02-18
    Posts
    40
    Points
    1,775
    Level
    11
    Points: 1,775, Level: 11
    Level completed: 75%, Points required for next Level: 75
    Overall activity: 9.0%


    Country: Portugal - Azores



    That's K13, not K15 :)

  4. #4
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    ^^
    The first post is now fixed. It's with K15

  5. #5
    Regular Member Achievements:
    1 year registered5000 Experience Points

    Join Date
    09-06-18
    Posts
    196
    Points
    7,008
    Level
    24
    Points: 7,008, Level: 24
    Level completed: 92%, Points required for next Level: 42
    Overall activity: 28.0%


    Ethnic group
    Catalan
    Country: Spain - Catalonia



    My Eurogenes EUtest V2 K15 Oracle results:


    Admix Results (sorted):

    # Population Percent
    1 Atlantic 27.8
    2 North_Sea 22.55
    3 West_Med 21.84
    4 East_Med 9.23
    5 Baltic 7.97
    6 Red_Sea 5.04
    7 West_Asian 3.79
    8 Amerindian 0.82
    9 Southeast_Asian 0.46
    10 Eastern_Euro 0.46
    11 Siberian 0.03
    12 Sub-Saharan 0.02

    Single Population Sharing:

    # Population (source) Distance
    1 Spanish_Castilla_Y_Leon 4.31
    2 Portuguese 4.88
    3 Spanish_Galicia 4.88
    4 Spanish_Cataluna 5.02
    5 Spanish_Extremadura 5.24
    6 Spanish_Murcia 5.86
    7 Spanish_Cantabria 7.01
    8 Spanish_Castilla_La_Mancha 7.61
    9 Spanish_Valencia 7.71
    10 Spanish_Andalucia 8.2
    11 Spanish_Aragon 8.52
    12 Southwest_French 8.83
    13 North_Italian 9.22
    14 French 9.68
    15 South_Dutch 14.71
    16 Tuscan 15.05
    17 West_German 16.84
    18 Southwest_English 17.66
    19 French_Basque 17.99
    20 Serbian 18.74

    These results are really odd, since my mother comes from Murcia-Andalusia (and all known ancestors from the same place) and my father, from Catalonia (all known ancestors also from Catalonia). Instead, the minimum distance is from Castilla y León, Portugal and Galicia, which are the places in Iberia farthest away from the family locations of my parents. I have no known relationship to these places. There was the Reconquista, but a paper I saw a few months ago showed the movement was north-south, so if any, I would expect aragonese or something like this.

    Any idea?

  6. #6
    Regular Member Achievements:
    1 year registered5000 Experience Points

    Join Date
    09-06-18
    Posts
    196
    Points
    7,008
    Level
    24
    Points: 7,008, Level: 24
    Level completed: 92%, Points required for next Level: 42
    Overall activity: 28.0%


    Ethnic group
    Catalan
    Country: Spain - Catalonia



    OK, a friend has told me, off-Forum, that Eurogenes is way off for Eastern Iberians like myself, since the Spanish samples in Eurogenes are biased towards Galicians, so every Catalan or Valencian gets a huge proportion of Galician, Portuguese, Castilian, Cantabric, Extremadura ... But this is fake.

    This is also consistent with my 23andme results, which show first Catalonia, then Andalusia, Murcia and Valencia, so I am purely "Mediterranean Iberian", according to 23andme, which is reasonable given my knowledge of my family.

    Of course, it is true that it seems 23andme focuses on the last 2 or 3 centuries, and Eurogenes targets older events. But even though, I do not see a massive movement of population of Western to Eastern Iberia a thousand years ago.

  7. #7
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    Using 1 population approximation:
    1 Spanish_Cantabria @ 6.280735
    2 Spanish_Castilla_La_Mancha @ 7.761991
    3 Southwest_French @ 8.481945
    4 Spanish_Castilla_Y_Leon @ 8.488644
    5 Spanish_Aragon @ 8.692999
    6 Spanish_Cataluna @ 9.260433
    7 Spanish_Extremadura @ 9.559198
    8 Spanish_Andalucia @ 9.937961
    9 Spanish_Valencia @ 9.954538
    10 Portuguese @ 10.074296
    11 Spanish_Murcia @ 10.178892
    12 Spanish_Galicia @ 10.916896
    13 French @ 14.676711
    14 French_Basque @ 15.663252
    15 North_Italian @ 16.065012
    16 South_Dutch @ 19.052065
    17 Southwest_English @ 21.704960
    18 Tuscan @ 22.904907
    19 West_German @ 23.140503
    20 Southeast_English @ 23.450478

    Using 2 populations approximation:
    1 50% Spanish_Cantabria +50% Spanish_Cantabria @ 6.280735


    Using 3 populations approximation:
    1 50% Spanish_Cantabria +25% Spanish_Cantabria +25% Spanish_Cantabria @ 6.280735


    Using 4 populations approximation:
    ++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++
    1 Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria @ 6.280735
    2 French_Basque + Spanish_Cantabria + Spanish_Galicia + Spanish_Galicia @ 6.395476
    3 Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria + Spanish_Castilla_La_Mancha @ 6.399209
    4 French_Basque + Spanish_Cantabria + Spanish_Cantabria + Spanish_Galicia @ 6.403708
    5 French_Basque + Spanish_Cantabria + Spanish_Castilla_Y_Leon + Spanish_Galicia @ 6.476579
    6 French_Basque + Portuguese + Spanish_Cantabria + Spanish_Galicia @ 6.479626
    7 Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria + Spanish_Castilla_Y_Leon @ 6.483661
    8 French_Basque + Spanish_Castilla_La_Mancha + Spanish_Galicia + Spanish_Galicia @ 6.517900
    9 French_Basque + Spanish_Cantabria + Spanish_Extremadura + Spanish_Galicia @ 6.533604
    10 French_Basque + Spanish_Cantabria + Spanish_Castilla_La_Mancha + Spanish_Galicia @ 6.567818
    11 French + French_Basque + Spanish_Cantabria + Spanish_Cantabria @ 6.567858
    12 French_Basque + Portuguese + Spanish_Cantabria + Spanish_Cantabria @ 6.642263
    13 French + French_Basque + Spanish_Cantabria + Spanish_Castilla_La_Mancha @ 6.642465
    14 French_Basque + Portuguese + Portuguese + Spanish_Cantabria @ 6.651925
    15 French + French_Basque + Spanish_Cantabria + Spanish_Extremadura @ 6.658022
    16 Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria + Spanish_Extremadura @ 6.664260
    17 French_Basque + Spanish_Cantabria + Spanish_Cataluna + Spanish_Galicia @ 6.667257
    18 French + French_Basque + Spanish_Andalucia + Spanish_Cantabria @ 6.669136
    19 Spanish_Aragon + Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria @ 6.670503
    20 Southwest_French + Spanish_Cantabria + Spanish_Cantabria + Spanish_Cantabria @ 6.678367

    Done.

    Elapsed time 1.7156 seconds.


    In MTA I have seen how the vettones get El Argar and vice versa, El Argar gets vascones and vicevers, Tartessos gets ilergetes and vice versa, e.t.c. so I have seen correspondences well before the reconquest and repopulation in which perhaps those correspondences between the different territories of Iberia were reinforced And all the ancient samples of Iberians and Basques are also obtained in considerable proportions.. I know a woman from Murcia who knows for sure that she descends from Cantabria, even has a manor house in Murcia, for the majority that memory is lost considering that in addition the majority of the old censuses were burned in the preambles of the war Spanish civilian


    I only ask that all this of genetics be left out and not influenced by modern regionalist nationalisms because what I believe could have happened with Ibero-Basque and each region of Spain have their own piece of nationalist Iberian. If you make the effort you can get it. ha

  8. #8
    Regular Member Achievements:
    1 year registered5000 Experience Points

    Join Date
    09-06-18
    Posts
    196
    Points
    7,008
    Level
    24
    Points: 7,008, Level: 24
    Level completed: 92%, Points required for next Level: 42
    Overall activity: 28.0%


    Ethnic group
    Catalan
    Country: Spain - Catalonia



    I do not understand what you mean, Carlos.

  9. #9
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    Quote Originally Posted by Farstar View Post
    I do not understand what you mean, Carlos.
    I see that you have understood perfectly. LOL

  10. #10
    Anthrogenica Mod Achievements:
    1 year registered1000 Experience Points

    Join Date
    12-02-18
    Posts
    40
    Points
    1,775
    Level
    11
    Points: 1,775, Level: 11
    Level completed: 75%, Points required for next Level: 75
    Overall activity: 9.0%


    Country: Portugal - Azores



    Quote Originally Posted by Farstar View Post
    OK, a friend has told me, off-Forum, that Eurogenes is way off for Eastern Iberians like myself, since the Spanish samples in Eurogenes are biased towards Galicians, so every Catalan or Valencian gets a huge proportion of Galician, Portuguese, Castilian, Cantabric, Extremadura ... But this is fake.

    This is also consistent with my 23andme results, which show first Catalonia, then Andalusia, Murcia and Valencia, so I am purely "Mediterranean Iberian", according to 23andme, which is reasonable given my knowledge of my family.

    Of course, it is true that it seems 23andme focuses on the last 2 or 3 centuries, and Eurogenes targets older events. But even though, I do not see a massive movement of population of Western to Eastern Iberia a thousand years ago.
    You (and your friend) are misunderstanding how both things work, 23andme looks at segments shared with their references and assigns you to them. A K15 just calculates different components from your raw data and calculates a distance with each vector (from your components). There's a lot of variation and overlapping in Iberia, you don't have to be closest to your regional reference in K15 or in a PCA like G25. Also, different references in K15 would produce different results, that's just simple statistics. I'm Portuguese, my closest reference in K15 is Aragón. Portugal is 11th.

    Code:
    4.68309727 Spanish_Aragon
    5.15799380 Spanish_Castilla_La_Mancha
    5.90200813 Spanish_Valencia
    6.07518724 Spanish_Andalucia
    7.09243964 Spanish_Cantabria
    7.10820652 Southwest_French
    7.87328394 Spanish_Castilla_Y_Leon
    8.02836845 Spanish_Murcia
    8.71643276 Spanish_Extremadura
    9.74420341 Spanish_Cataluna
    10.75186030 Portuguese
    12.70977183 France-South
    12.93304682 Spanish_Galicia
    13.72554188 French_Basque
    My list is kind of similar to Carlos'...where are you from?

  11. #11
    Regular Member Achievements:
    1 year registered5000 Experience Points

    Join Date
    09-06-18
    Posts
    196
    Points
    7,008
    Level
    24
    Points: 7,008, Level: 24
    Level completed: 92%, Points required for next Level: 42
    Overall activity: 28.0%


    Ethnic group
    Catalan
    Country: Spain - Catalonia



    I am Catalan. My father is Catalan, and my mother from south east Spain. 23andme gives me ancestry results consistent with this. Other calculators do not. This suggests that, at least for me, 23andme technology is superior to the others, right? Are there other people with reverse experience? Granted, 23andme and eurogenes target different time stamps.

  12. #12
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    Quote Originally Posted by Ruderico View Post
    You (and your friend) are misunderstanding how both things work, 23andme looks at segments shared with their references and assigns you to them. A K15 just calculates different components from your raw data and calculates a distance with each vector (from your components). There's a lot of variation and overlapping in Iberia, you don't have to be closest to your regional reference in K15 or in a PCA like G25. Also, different references in K15 would produce different results, that's just simple statistics. I'm Portuguese, my closest reference in K15 is Aragón. Portugal is 11th.

    Code:
    4.68309727 Spanish_Aragon
    5.15799380 Spanish_Castilla_La_Mancha
    5.90200813 Spanish_Valencia
    6.07518724 Spanish_Andalucia
    7.09243964 Spanish_Cantabria
    7.10820652 Southwest_French
    7.87328394 Spanish_Castilla_Y_Leon
    8.02836845 Spanish_Murcia
    8.71643276 Spanish_Extremadura
    9.74420341 Spanish_Cataluna
    10.75186030 Portuguese
    12.70977183 France-South
    12.93304682 Spanish_Galicia
    13.72554188 French_Basque
    My list is kind of similar to Carlos'...where are you from?
    I`m Andalusian
    I almost always get Cantabria as the closest modern population, but in these new results I get Aragón and Huelva; although the Spanish regions are really close.

    Last edited by Carlos; 12-01-20 at 12:34.

  13. #13
    Anthrogenica Mod Achievements:
    1 year registered1000 Experience Points

    Join Date
    12-02-18
    Posts
    40
    Points
    1,775
    Level
    11
    Points: 1,775, Level: 11
    Level completed: 75%, Points required for next Level: 75
    Overall activity: 9.0%


    Country: Portugal - Azores



    Quote Originally Posted by Farstar View Post
    I am Catalan. My father is Catalan, and my mother from south east Spain. 23andme gives me ancestry results consistent with this. Other calculators do not. This suggests that, at least for me, 23andme technology is superior to the others, right? Are there other people with reverse experience? Granted, 23andme and eurogenes target different time stamps.
    That's the issue, you're comparing different things. 23andme also assigns me Portugal because I match segments with Portuguese individuals, but my results are unlike most Portuguese (~86% Iberian, 0% Italian, 0% West Asian, 0% SSA, etc)

  14. #14
    Advisor Achievements:
    VeteranThree Friends50000 Experience PointsRecommendation Second Class
    Awards:
    Posting Award
    Angela's Avatar
    Join Date
    02-01-11
    Posts
    16,552
    Points
    359,115
    Level
    100
    Points: 359,115, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 99.6%


    Ethnic group
    Italian
    Country: USA - New York



    1 members found this post helpful.
    The purpose of the commercial tests, other than to sell your data for research, create drugs etc., is to tell people, most of whom may not know their ethnic identity, to which modern populations they are closest.

    These "amateur" tools are for people who want to know "ancient" ancestry, the population genetics of various groups, which is fine, but they are created by amateurs themselves who can be biased, may have massaged data or strategically included or not included certain samples, may have an imperfect understanding of the groups, or the programs, which they did NOT create, as is the case with many of the academics, and who in almost all cases have far more confidence in their understanding than is warranted.

    Take it all with a truck load of salt, but the latter more salt than the former.


    Non si fa il proprio dovere perchè qualcuno ci dica grazie, lo si fa per principio, per se stessi, per la propria dignità. Oriana Fallaci

  15. #15
    Regular Member Achievements:
    1 year registered5000 Experience Points

    Join Date
    09-06-18
    Posts
    196
    Points
    7,008
    Level
    24
    Points: 7,008, Level: 24
    Level completed: 92%, Points required for next Level: 42
    Overall activity: 28.0%


    Ethnic group
    Catalan
    Country: Spain - Catalonia



    1 members found this post helpful.
    Quote Originally Posted by Angela View Post
    The purpose of the commercial tests, other than to sell your data for research, create drugs etc., is to tell people, most of whom may not know their ethnic identity, to which modern populations they are closest.

    These "amateur" tools are for people who want to know "ancient" ancestry, the population genetics of various groups, which is fine, but they are created by amateurs themselves who can be biased, may have massaged data or strategically included or not included certain samples, may have an imperfect understanding of the groups, or the programs, which they did NOT create, as is the case with many of the academics, and who in almost all cases have far more confidence in their understanding than is warranted.

    Take it all with a truck load of salt, but the latter more salt than the former.
    Yes, Angela, what you explain makes sense to me.

    I hope some day, some tool will appear to show the ancestry composition, expliciting timing. In the end, all "pure" ancestry is in fact a previous admixture, just older. But I guess that to do that, one needs not only PCA or distances, but some kind of diffusion models for the populations. I am sure the next years and decades will be quite interesting on this regard.

  16. #16
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    Quote Originally Posted by Farstar View Post
    I am Catalan. My father is Catalan, and my mother from south east Spain. 23andme gives me ancestry results consistent with this. Other calculators do not. This suggests that, at least for me, 23andme technology is superior to the others, right? Are there other people with reverse experience? Granted, 23andme and eurogenes target different time stamps.
    I did it in FTDNA of course. I know of people with some of their Spanish parents who in 23 do not get Iberian.

  17. #17
    Regular Member Achievements:
    Three Friends25000 Experience Points1 year registered
    Awards:
    Most Popular
    Duarte's Avatar
    Join Date
    08-01-19
    Posts
    919
    Points
    43,729
    Level
    64
    Points: 43,729, Level: 64
    Level completed: 60%, Points required for next Level: 521
    Overall activity: 66.0%

    Y-DNA haplogroup
    R1b-DF27-FGC35133

    Country: Brazil



    1 members found this post helpful.
    GedMatch - Eurogenes EUtest V2 K15

    Code:
    # Population (source) Distance
    1 Portuguese 6.49
    2 Spanish_Extremadura 7.07
    3 Spanish_Galicia 7.38
    4 Spanish_Castilla_Y_Leon 7.57
    5 Spanish_Murcia 7.58
    6 Spanish_Cantabria 7.68
    7 Spanish_Cataluna 8.03
    8 Spanish_Castilla_La_Mancha 8.78
    9 Spanish_Andalucia 8.8
    10 Spanish_Valencia 9.13
    11 Spanish_Aragon 9.86
    12 Southwest_French 10.37
    13 North_Italian 11.67
    14 French 13.72
    It's fun, but I agree with Angela:

    Quote Originally Posted by Angela View Post
    The purpose of the commercial tests, other than to sell your data for research, create drugs etc., is to tell people, most of whom may not know their ethnic identity, to which modern populations they are closest.

    These "amateur" tools are for people who want to know "ancient" ancestry, the population genetics of various groups, which is fine, but they are created by amateurs themselves who can be biased, may have massaged data or strategically included or not included certain samples, may have an imperfect understanding of the groups, or the programs, which they did NOT create, as is the case with many of the academics, and who in almost all cases have far more confidence in their understanding than is warranted.

    Take it all with a truck load of salt, but the latter more salt than the former.

  18. #18
    Anthrogenica Mod Achievements:
    1 year registered1000 Experience Points

    Join Date
    12-02-18
    Posts
    40
    Points
    1,775
    Level
    11
    Points: 1,775, Level: 11
    Level completed: 75%, Points required for next Level: 75
    Overall activity: 9.0%


    Country: Portugal - Azores



    You are potentially accusing them of being fraudulent, which is pretty serious (or would be, should any of this be official, but it isn't). The effects you mention can be problematic in supervised tests like K13/K15, etc, but in an unsupervised test like G25 that's not as much of an issue since the algorithm is left to create its structure. Here the biggest issue might be dimensionality, G25 might have been created with too many dimensions for the sample size used in its input and some minor issues occasionally pop up when using the data in certain ways (like nMonte3 models, eventhough it can me mitigated with a penalty), but even then the tool appears to be pretty good and relatively coherent with published data, at least for West Eurasia, I am not familiar with the rest of the world in this PCA data


    Edit: On Iberians not matching their regional reference, it's unimportant. A reference changes once you change the samples used to calculate it, when there's a large overlap within different regions, and noticeable variation within that region, "matching" a different region in this K15 or K13 shouldn't be surprising. Iberia had a lot of population movements in the last millennia, some areas were left more isolated than others. This variation can see seen in studies of modern Iberian population, for example this K7 ADMIXTURE graph https://i.postimg.cc/446W4B58/k7admix.png

  19. #19
    Regular Member Achievements:
    1 year registered5000 Experience Points

    Join Date
    09-06-18
    Posts
    196
    Points
    7,008
    Level
    24
    Points: 7,008, Level: 24
    Level completed: 92%, Points required for next Level: 42
    Overall activity: 28.0%


    Ethnic group
    Catalan
    Country: Spain - Catalonia



    1 members found this post helpful.
    Quote Originally Posted by Ruderico View Post
    You are potentially accusing them of being fraudulent, which is pretty serious (or would be, should any of this be official, but it isn't). The effects you mention can be problematic in supervised tests like K13/K15, etc, but in an unsupervised test like G25 that's not as much of an issue since the algorithm is left to create its structure. Here the biggest issue might be dimensionality, G25 might have been created with too many dimensions for the sample size used in its input and some minor issues occasionally pop up when using the data in certain ways (like nMonte3 models, eventhough it can me mitigated with a penalty), but even then the tool appears to be pretty good and relatively coherent with published data, at least for West Eurasia, I am not familiar with the rest of the world in this PCA data


    Edit: On Iberians not matching their regional reference, it's unimportant. A reference changes once you change the samples used to calculate it, when there's a large overlap within different regions, and noticeable variation within that region, "matching" a different region in this K15 or K13 shouldn't be surprising. Iberia had a lot of population movements in the last millennia, some areas were left more isolated than others. This variation can see seen in studies of modern Iberian population, for example this K7 ADMIXTURE graph https://i.postimg.cc/446W4B58/k7admix.png
    I am sure Angela knows how to defend herself, but it is obvious she is not accusing of fraud to anybody. Have you felt called in?

    The issue of supervised or non-supervised is not the relevant point, if samples have been "massaged" ex-ante. For sure, misclassification is a serious issue, but if some samples that should be in, are not there, or viceversa, the unsupervised algorithm may fail to find the right structure. In fact, unsupervised algorithms tend to be "worse" than supervised ones (of course, lots of caveats here), so I am surprised that an unsupervised algorithm is seen as an evolution of supervised ones.

    About Iberian regions: if there is so much overlap, maybe it would be interesting to have not only a measure of the distance, but also a measure of the error. For example, if all the first 5 estimated regions were within the error bands, one could not say that the first region is more important than the fifth. Now, most people believe that the first one is the most important one, and anything above the 2nd or 3rd is discounted. But the ranking could be purely due to noise, if the distance among regions is so overlapping.

  20. #20
    Anthrogenica Mod Achievements:
    1 year registered1000 Experience Points

    Join Date
    12-02-18
    Posts
    40
    Points
    1,775
    Level
    11
    Points: 1,775, Level: 11
    Level completed: 75%, Points required for next Level: 75
    Overall activity: 9.0%


    Country: Portugal - Azores



    Why should I be felt called in? That'd be ridiculous, I didn't even make any tool.


    I don't know if the samples have been "massaged" or not and neither does anyone here. If there's evidence for that, I'd very much like to learn it. Edit: I'm referring to the tools us amateurs-at-best use, namely Eurogenes and Dodecad
    We only know the samples used for the input in K13/15 (and probably G25, although I'm not certain whether ancient samples were used or not) were from HGDP. What we can see is that, as a tool used used to visualise data, G25 seems more than decent. My argument on supervised is that they are easier to mess up when it comes to what we do here, since these were done back in 2013 and our knowledge of population genetics has changed since then. I'm pretty sure that if people were to remake these tools probably new components would be used, for example the NEAfr and SSA are poor to represent north African ancestry in Europe.


    As for Iberian regions, its study is best left for scientific studies in population genomics which is why I mentioned it. If what you want is to know a broad degree of similarity you can make statistical analysis yourself, there's plenty of public data in G25. People have been doing that since it was made available. If not, because you don't trust the tool, just read what we currently have, or wait for new studies.
    Last edited by Ruderico; 14-01-20 at 16:31.

  21. #21
    Advisor Achievements:
    VeteranThree Friends50000 Experience PointsRecommendation Second Class
    Awards:
    Posting Award
    Angela's Avatar
    Join Date
    02-01-11
    Posts
    16,552
    Points
    359,115
    Level
    100
    Points: 359,115, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 99.6%


    Ethnic group
    Italian
    Country: USA - New York



    Quote Originally Posted by Farstar View Post
    I am sure Angela knows how to defend herself, but it is obvious she is not accusing of fraud to anybody. Have you felt called in?

    The issue of supervised or non-supervised is not the relevant point, if samples have been "massaged" ex-ante. For sure, misclassification is a serious issue, but if some samples that should be in, are not there, or viceversa, the unsupervised algorithm may fail to find the right structure. In fact, unsupervised algorithms tend to be "worse" than supervised ones (of course, lots of caveats here), so I am surprised that an unsupervised algorithm is seen as an evolution of supervised ones.

    About Iberian regions: if there is so much overlap, maybe it would be interesting to have not only a measure of the distance, but also a measure of the error. For example, if all the first 5 estimated regions were within the error bands, one could not say that the first region is more important than the fifth. Now, most people believe that the first one is the most important one, and anything above the 2nd or 3rd is discounted. But the ranking could be purely due to noise, if the distance among regions is so overlapping.
    Couldn't have put it better. :)

  22. #22
    Regular Member Achievements:
    Veteran25000 Experience Points
    Carlos's Avatar
    Join Date
    26-09-11
    Posts
    1,970
    Points
    30,161
    Level
    53
    Points: 30,161, Level: 53
    Level completed: 38%, Points required for next Level: 689
    Overall activity: 58.0%

    Y-DNA haplogroup
    E-V22>YF66572
    MtDNA haplogroup
    J1c5c1

    Ethnic group
    España
    Country: Spain



    If, for example, I had been a child robbed during the Franco regime and taken to another country and had all these tests, calculations, MTA e.t.c. where Andalusia usually leaves in the 9th place and Cantabria is almost always in the first place in these last results, Aragón comes out in the first place if more information would have been thinking in those regions as my possible origin or that of my parents and of course everything is a lot more complicated because considering that my native population is just over 200 years old and although the population was with neighbors of older nearby populations but who have lived the reconquest and repopulation I can understand that in my hometown it is as if they had mixed Spaniards from other regions reinforced perhaps because since ancient times there is correspondence between the Spanish regions as I have seen in the results of La Cueva de la paloma, El sotilo, El Argar, Cógotas, etc.


    But I know all that, but in the case of someone looking for their origins as in the supposed case of a child robbed in Spain the first regions in the results would not have given the answer in its closest origin because there are many other factors .

    -------------------------------------------
    -------------------------------------------

    Now that I am trying to trace the origin of my DNA, I already know that all this is autosomal and it is only one of my ancestors but somewhere it has to go out somehow.





    On Aragon, yes, in FTDNA it had a match with modern people that go from the Basque country to Aragon, but with Huelva it had not obtained results anywhere, some connection must have been perhaps an older time like Tartessos.




    I know that you can do with different algorithms remove or add populations but also to extract the gold you have to make a sieve.

    If someone has something better to expose it.

  23. #23
    Anthrogenica Mod Achievements:
    1 year registered1000 Experience Points

    Join Date
    12-02-18
    Posts
    40
    Points
    1,775
    Level
    11
    Points: 1,775, Level: 11
    Level completed: 75%, Points required for next Level: 75
    Overall activity: 9.0%


    Country: Portugal - Azores



    I think if a person doesn't know their origins their best bet would be to look at DNA matches following a test. That's what, for example, people with unknown AJ ancestry do, they will always have a bunch of AJ matches that will tell them that, even if in one of these "calculators" that AJ gets 'mistaken' for some generic East Mediterranean component

  24. #24
    Advisor Achievements:
    Three FriendsTagger First Class1 year registered50000 Experience Points
    Awards:
    Master Tagger
    Jovialis's Avatar
    Join Date
    04-05-17
    Location
    New York City
    Posts
    3,379
    Points
    147,194
    Level
    100
    Points: 147,194, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 99.3%

    Y-DNA haplogroup
    R1b1a1a2b1 (R-F1794)
    MtDNA haplogroup
    H6a1b

    Ethnic group
    Italian
    Country: United States



    2 members found this post helpful.
    Quote Originally Posted by Ruderico View Post
    Why should I be felt called in? That'd be ridiculous, I didn't even make any tool.


    I don't know if the samples have been "massaged" or not and neither does anyone here. If there's evidence for that, I'd very much like to learn it. Edit: I'm referring to the tools us amateurs-at-best use, namely Eurogenes and Dodecad
    We only know the samples used for the input in K13/15 (and probably G25, although I'm not certain whether ancient samples were used or not) were from HGDP. What we can see is that, as a tool used used to visualise data, G25 seems more than decent. My argument on supervised is that they are easier to mess up when it comes to what we do here, since these were done back in 2013 and our knowledge of population genetics has changed since then. I'm pretty sure that if people were to remake these tools probably new components would be used, for example the NEAfr and SSA are poor to represent north African ancestry in Europe.


    As for Iberian regions, its study is best left for scientific studies in population genomics which is why I mentioned it. If what you want is to know a broad degree of similarity you can make statistical analysis yourself, there's plenty of public data in G25. People have been doing that since it was made available. If not, because you don't trust the tool, just read what we currently have, or wait for new studies.
    I follow the rules where ever I go. Which is part of the reason why I have the acumen to be a moderator on a population genetics forum.

    Pertinent to the discussion:

    https://anthrogenica.com/showthread....l=1#post637864

    Oddly, you seem to fall on the side of Davidski, even when he insults forum users on Anthrogenica.
    There can be no covenants between men and lions

  25. #25
    Advisor Achievements:
    Three FriendsTagger First Class1 year registered50000 Experience Points
    Awards:
    Master Tagger
    Jovialis's Avatar
    Join Date
    04-05-17
    Location
    New York City
    Posts
    3,379
    Points
    147,194
    Level
    100
    Points: 147,194, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Overall activity: 99.3%

    Y-DNA haplogroup
    R1b1a1a2b1 (R-F1794)
    MtDNA haplogroup
    H6a1b

    Ethnic group
    Italian
    Country: United States



    Quote Originally Posted by Jovialis View Post
    I follow the rules where ever I go. Which is part of the reason why I have the acumen to be a moderator on a population genetics forum.
    Pertinent to the discussion:
    https://anthrogenica.com/showthread....l=1#post637864
    Oddly, you seem to fall on the side of Davidski, even when he insults forum users on Anthrogenica.
    Tell me RUDERICO, do you actually believe the Sardinians are east of Anatolian_N? I saw you upvoted Davidski's post. Why don't you elaborate on this for us, because apparently the PCAs are wrong.

Page 1 of 2 12 LastLast

Posting Permissions

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