Page 1 of 4 123 ... LastLast
Results 1 to 10 of 35

Thread: A local view on Eurogenes Northern Europe PCA

  1. #1
    Registered Users
    Posts
    493
    Sex
    Location
    Netherlands
    Ethnicity
    South-Dutch
    Nationality
    Dutch
    Y-DNA
    I2a2a1b2-CTS1977
    mtDNA
    H13a1a1

    Netherlands Belgium

    A local view on Eurogenes Northern Europe PCA

    Being a North-West European, I like Eurogenes Northern Europe PCA.
    However, especially for West Europeans, this PCA displays little structure. My first plot presents a scatter plot of the most Western part of the populations.
    Irish13.jpg
    The relative lack of structure is not surprising. The European populations have a history of millennia of profound mixing. As a consequence, the populations are heavily overlapping.
    However, another view is possible. A PCA orders the samples along long distance trends (the eigenvectors). If somewhere the structure of the populations is more local, this will remain invisible in a PCA.
    Fortunately there are modern algorithms which can zoom in on these local structures. I have used an algorithm (diffusion maps) which sees populations as a network of individuals who have neighbors who have neighbors etc. This way similarity and distance get a more local flavor.
    Now look again at the scatter plot of the same West European populations:
    Irish_diffusion24.jpg
    This plot displays a strong local structure among part of the Irish.
    The structure is compatable with the results of: "The Irish DNA Atlas: Revealing Fine-Scale Population Structure and History within Ireland" (https://www.nature.com/articles/s41598-017-17124-4)
    Quote: "The ‘Gaelic’ Irish clusters show the lowest ancestry proportions of German clusters, which in turn are thought to reflect Germanic/Saxon influence. Orkney shows the second-least 'Germanic' proportion, with English clusters showing the most. We also observe a low amount of Belgian-like ancestry within Ireland, compared to groups within Britain, further illustrating Ireland’s relative isolation from mainland Europe."

    Of course there is no free lunch. To discover local structure, a high sampling density is necessary. That is the case with the Irish (N=32) and maybe the reason that outside NW-Europe local structure is much less visible.
    Another disadvantage is that the algorithm uses non-Euclidean distances, which makes it impossible to use nMonte.
    Yet, an interesting plot.

  2. The Following 7 Users Say Thank You to Huijbregts For This Useful Post:

     ajc347 (05-17-2018), Dewsloth (05-16-2018), Garimund (05-16-2018), Jessie (05-16-2018), Kurd (05-18-2018), NORPopGen (05-17-2018), Vestri (05-16-2018)

  3. #2
    Registered Users
    Posts
    493
    Sex
    Location
    Netherlands
    Ethnicity
    South-Dutch
    Nationality
    Dutch
    Y-DNA
    I2a2a1b2-CTS1977
    mtDNA
    H13a1a1

    Netherlands Belgium
    Score of Jessie.
    In my previous post I demonstrated that with a special transformation I could get more discrimination between the samples of Eurogenes North European PCA.
    Or actually I demonstrated that some of the modern Irish samples appear to be admixed with Scottish or other populations, but that most of them cluster close together.
    Having seen that, I was curious about the score of Jessie, who is considered the most Irish of us all.
    Here it is, no comment.
    Jessie_diffusion24.jpg

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

     Ais (05-16-2018), Dewsloth (05-16-2018), Jessie (05-16-2018), sktibo (05-17-2018)

  5. #3
    Moderator
    Posts
    4,455
    Sex
    Location
    Normandy
    Ethnicity
    northwesterner
    Y-DNA
    U152>L2>Z367
    mtDNA
    H5a1

    Normandie Netherlands Friesland Finland
    Ger, out of curiosity, did you use the diffusion map package of R ?
    En North alom, de North venom
    En North fum naiz, en North manom

    (Roman de Rou, Wace, 1160-1170)

  6. #4
    Registered Users
    Posts
    1,163
    Sex
    Ethnicity
    Irish
    Nationality
    Irish
    Y-DNA
    M222 (S588)
    mtDNA
    J1c3f

    Ireland Australia Ireland County Tipperary
    Quote Originally Posted by Huijbregts View Post
    Score of Jessie.
    In my previous post I demonstrated that with a special transformation I could get more discrimination between the samples of Eurogenes North European PCA.
    Or actually I demonstrated that some of the modern Irish samples appear to be admixed with Scottish or other populations, but that most of them cluster close together.
    Having seen that, I was curious about the score of Jessie, who is considered the most Irish of us all.
    Here it is, no comment.
    Jessie_diffusion24.jpg
    I wonder if my west Irish ancestry gives me these sort of isolated positions? I'm sure it is a factor as a lot of genetic position on plots appears to parallel geography as well. Very interesting anyway.
    Last edited by Jessie; 05-16-2018 at 03:42 PM.

  7. The Following 2 Users Say Thank You to Jessie For This Useful Post:

     Huijbregts (05-16-2018), sktibo (05-17-2018)

  8. #5
    Registered Users
    Posts
    173
    Sex
    Ethnicity
    English
    Nationality
    English
    Y-DNA
    R1a - Z284

    England Norway Germany Poland
    Jessie's so Irish she's now got her own Ireland next to Ireland!
    Y-DNA R1a - Z284
    MtDNA - W1g

    Ancestry
    Mostly English with minor East German/Polish.
    *Eurogenes K15 - West_German + West_Norwegian + West_Norwegian + West_Norwegian @ 6.696671
    *MDLP K23B - Dane_ + Dutch_ + Dutch_ + Dutch_ @ 1.366235
    *PuntDNAL K12 Ancient - Vatya_SG_RISE479 + Nordic_LN_SG_RISE97 + BattleAxe_Sweden_SG_RISE94 + BattleAxe_Sweden_SG_RISE94 @ 1.023980

    Dad - Y-DNA R1a Z284 / MtDNA U5b2b1a1
    Mum - Y-DNA I1 / MtDNA W1g
    Granddad ( Mother's side ) - Y-DNA I1 / MtDNA H3y

  9. The Following User Says Thank You to Vestri For This Useful Post:

     Jessie (05-16-2018)

  10. #6
    Registered Users
    Posts
    493
    Sex
    Location
    Netherlands
    Ethnicity
    South-Dutch
    Nationality
    Dutch
    Y-DNA
    I2a2a1b2-CTS1977
    mtDNA
    H13a1a1

    Netherlands Belgium
    Quote Originally Posted by anglesqueville View Post
    Ger, out of curiosity, did you use the diffusion map package of R ?
    Yes. I estimated a model with 4 dimensions. The default value of eps.val may be somewhat small.

  11. The Following User Says Thank You to Huijbregts For This Useful Post:

     anglesqueville (05-16-2018)

  12. #7
    Registered Users
    Posts
    493
    Sex
    Location
    Netherlands
    Ethnicity
    South-Dutch
    Nationality
    Dutch
    Y-DNA
    I2a2a1b2-CTS1977
    mtDNA
    H13a1a1

    Netherlands Belgium
    Quote Originally Posted by Jessie View Post
    I wonder if my west Irish ancestry gives me these sort of isolated positions? I'm sure it is a factor as a lot of genetic position on plots appears to parallel geography as well. Very interesting anyway.
    Jessie
    The plot suggests that your sample has an exceptionally low percentage of Continental and even English admixture.
    The gap between you and the other Irish is a little odd. I am thinking about a statical artefact.
    Recall that there are only 23 Irish samples in the dataset. That is not nearly enough to cover all of the Irish genetical variance. Chances are that you do have close genetical relatives, but that they just were not among the 23 samples. This problem is even magnified if many of your folks have emigrated to America.

  13. The Following 2 Users Say Thank You to Huijbregts For This Useful Post:

     Jessie (05-17-2018), sktibo (05-18-2018)

  14. #8
    Registered Users
    Posts
    1,163
    Sex
    Ethnicity
    Irish
    Nationality
    Irish
    Y-DNA
    M222 (S588)
    mtDNA
    J1c3f

    Ireland Australia Ireland County Tipperary
    Quote Originally Posted by Huijbregts View Post
    Jessie
    The plot suggests that your sample has an exceptionally low percentage of Continental and even English admixture.
    The gap between you and the other Irish is a little odd. I am thinking about a statical artefact.
    Recall that there are only 23 Irish samples in the dataset. That is not nearly enough to cover all of the Irish genetical variance. Chances are that you do have close genetical relatives, but that they just were not among the 23 samples. This problem is even magnified if many of your folks have emigrated to America.
    I still have family in Ireland. Alot of my father's side did emigrate and there is no family left in the area he is from. My mother's side didn't tend to emigrate as much. I'm wondering if possibly these samples are from Dublin. That could be a factor as all my ancestry is from Munster and Connacht. I definitely don't appear to match the Irish samples very closely in the Northern Europe PCA judging from this and my results from Davidski.

  15. The Following 2 Users Say Thank You to Jessie For This Useful Post:

     Huijbregts (05-17-2018), sktibo (05-17-2018)

  16. #9
    Moderator
    Posts
    4,455
    Sex
    Location
    Normandy
    Ethnicity
    northwesterner
    Y-DNA
    U152>L2>Z367
    mtDNA
    H5a1

    Normandie Netherlands Friesland Finland
    I computed that night a diff map with R on a reduced sheet of Eurogenes northeuro, and played a little on Past3 with the matrix dmap$X. At first glimpse there seems to be a strong discrimination along an axis West-East. Interesting.
    En North alom, de North venom
    En North fum naiz, en North manom

    (Roman de Rou, Wace, 1160-1170)

  17. #10
    Registered Users
    Posts
    493
    Sex
    Location
    Netherlands
    Ethnicity
    South-Dutch
    Nationality
    Dutch
    Y-DNA
    I2a2a1b2-CTS1977
    mtDNA
    H13a1a1

    Netherlands Belgium
    Quote Originally Posted by anglesqueville View Post
    I computed that night a diff map with R on a reduced sheet of Eurogenes northeuro, and played a little on Past3 with the matrix dmap$X. At first glimpse there seems to be a strong discrimination along an axis West-East. Interesting.
    As far as I understand the mathematics, diffusion mapping can accurately model the diffusion between neighboring and overlapping populations. Knowing the diffusion process, it can also play the diffusion process in reverse, like a movie. Thus it ends up with a low dimensional deeper structure ( a 'manifold'). In this case the deeper structure might be the pulse migrations of the European past. In the Northern dataset the first dimension represent an East-West cline. But the most distinct structure is in dimensions 2 and 4. In this plane a compact cline is visible, which ends among the Irish. This may well have been (part of) the Germanic expansion.
    Here is a 3D plot of the dimension 1,2 and 4.
    3D.jpg
    Last edited by Huijbregts; 05-17-2018 at 11:43 AM.

Page 1 of 4 123 ... LastLast

Similar Threads

  1. Replies: 21
    Last Post: 03-08-2018, 01:12 AM
  2. Replies: 0
    Last Post: 11-13-2017, 02:16 PM
  3. The Genetic History of Northern Europe
    By Pribislav in forum Ancient (aDNA)
    Replies: 111
    Last Post: 03-06-2017, 03:41 PM
  4. Replies: 76
    Last Post: 10-20-2016, 07:54 AM
  5. E-V13 parent in Northern Europe
    By Bane in forum E1b-M215
    Replies: 13
    Last Post: 06-18-2016, 06:12 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
  •