Page 7 of 23 FirstFirst ... 5678917 ... LastLast
Results 61 to 70 of 230

Thread: A little G25-K7 based "calculator" for Europeans only.

  1. #61
    Moderator
    Posts
    5,400
    Sex
    Location
    Normandy
    Ethnicity
    northwesterner
    Y-DNA
    U152>L2>Z367
    mtDNA
    H5a1

    Normandie Netherlands Friesland Finland Orkney
    Quote Originally Posted by Capitalis View Post
    Bonjour anglesqueville, can I have your G25 scaled coordinates for my own interest in Northern French ancestry, please? Merci.
    Capitalis under a "French" label, I'm a complicated cocktail, and I didn't go back higher than 4 generations, who knows what I should find before? I'm actually not a good sample for you.
    En North alom, de North venom
    En North fum naiz, en North manom

    (Roman de Rou, Wace, 1160-1170)

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

     Capitalis (11-10-2018),  JMcB (11-09-2018)

  3. #62
    Moderator
    Posts
    5,400
    Sex
    Location
    Normandy
    Ethnicity
    northwesterner
    Y-DNA
    U152>L2>Z367
    mtDNA
    H5a1

    Normandie Netherlands Friesland Finland Orkney
    Quote Originally Posted by Tolan View Post
    And my unscaled result:
    [1] "distance%=11.9976"

    Tolan

    South-Euro,52
    Uralic,48

    Catastrophe!
    The spreadsheet you posted (#8) is scaled. If you used this sheet with your unscaled components it's not surprising you got weird outputs.
    En North alom, de North venom
    En North fum naiz, en North manom

    (Roman de Rou, Wace, 1160-1170)

  4. The Following 2 Users Say Thank You to anglesqueville For This Useful Post:

     Helgenes50 (11-09-2018),  Tolan (11-09-2018)

  5. #63
    Moderator
    Posts
    5,400
    Sex
    Location
    Normandy
    Ethnicity
    northwesterner
    Y-DNA
    U152>L2>Z367
    mtDNA
    H5a1

    Normandie Netherlands Friesland Finland Orkney
    Quote Originally Posted by Theconqueror View Post
    Distances on this calculator are huge. This is the problem with using averages. G25 has lots of nice granularity (you can be close to a number of samples within a population without being close to the average of that population) and I don't see how useful this type of calculator can be. This one in particular is very dichotomic north-south.
    You are of course allowed to replace the averages by the individual results, why not? That said the distances are not huge at all. I'm not a great fan of averages, but according to the goal of this (little, once more) thing, it was imo a reasonable choice. If you feel able to elaborate a more useful tool, well, I can't wait for it...
    Last edited by anglesqueville; 11-09-2018 at 04:46 PM.
    En North alom, de North venom
    En North fum naiz, en North manom

    (Roman de Rou, Wace, 1160-1170)

  6. The Following User Says Thank You to anglesqueville For This Useful Post:

     JMcB (11-09-2018)

  7. #64
    Gold Class Member
    Posts
    1,509
    Sex
    Location
    United States
    Ethnicity
    LebaGermish
    Y-DNA
    P312>DF19>DF88>S4268
    mtDNA
    J2a1a1e

    United States of America Lebanon Germany England Belgium Scotland
    Quote Originally Posted by Tolan View Post

    My result:
    [1] "distance%=1.3982"

    Tolan_scaled

    Central-Euro,50.2
    South-Atlantic,23.2
    Channel-North-Sea,15.2
    North-Atlantic,6.6
    South-Euro,4.8
    Quote Originally Posted by anglesqueville View Post
    [1] "distance%=0.7746"

    DewslothDad

    Central-Euro,35.4
    South_Atlantic,23.2
    Channel-North-Sea,21.6
    South_Euro,11.4
    North-Atlantic,8.4
    Huh, not that far off...
    R1b>M269>L23>L51>L11>P312>DF19>DF88>FGC11833 >S4281>S4268>Z17112 (S17075-)

    Y-cousin: 6DRIF-23 (DF19>>Z17112+, S17075+)

    Ancestors: Francis Cooke (M223/I2a2a) b1583; Hester Mahieu (Cooke) (J1c2 mtDNA) b.1584; Richard Warren (E-M35) b1578; Elizabeth Walker (Warren) (H1j mtDNA) b1583;
    John Mead (I2a1/P37.2) b1634; Rev. Joseph Hull (I1, L1301+ L1302-) b1595; Benjamin Harrington (M223/I2a2a-Y5729) b1618; Joshua Griffith (L21>DF13) b1593;
    John Wing (U106) b1584; Hermann Wilhelm (DF19) b1635

  8. The Following User Says Thank You to Dewsloth For This Useful Post:

     anglesqueville (11-09-2018)

  9. #65
    Registered Users
    Posts
    2,618
    Sex
    Location
    Groningen
    Ethnicity
    Northwest European
    Nationality
    NL
    Y-DNA
    E-V22

    Netherlands
    Obvious result:


  10. The Following User Says Thank You to Finn For This Useful Post:

     lukaszM (11-09-2018)

  11. #66
    Gold Class Member
    Posts
    1,961
    Sex
    Location
    Brittany
    Ethnicity
    NW European
    Y-DNA
    I-L813 >Y36690

    Normandie Sweden France Bretagne
    Quote Originally Posted by anglesqueville View Post
    The spreadsheet you posted (#8) is scaled. If you used this sheet with your unscaled components it's not surprising you got weird outputs.
    Now I understand better, I used this sheet (post 8 )
    My results from David Wesolowski's Ancestry Detective Service:

    West British (Britonic?) 42.3%
    Continental Northern and Eastern European 36.6%
    Central French 21.1%

  12. #67
    Registered Users
    Posts
    1,206
    Sex
    Location
    Canada
    Ethnicity
    Norman/German
    Nationality
    Canadian
    Y-DNA
    R1B-DF99(FGC16982)

    Germany Imperial Normandie
    Quote Originally Posted by anglesqueville View Post
    You are of course allowed to replace the averages by the individual results, why not? That said the distances are not huge at all. I'm not a great fan of averages, but according to the goal of this (little, once more) thing, it was imo a reasonable choice. If you feel able to elaborate a more useful tool, well, I can't wait for it...
    Yes I am able as I have a PhD in epidemiology and a BSc and MSc in stats and economics but it is besides the point. It is an interesting artefact but only slightly useful, that's all.
    K8: French East/German South/Austrian 26%, French North East/Belgian/German West 25%, French North 25%, Irish/Scottish/Welsh 10%, French South/French Basque 9%, German East/Czech/Austrian 5%
    K16: German/Czech/Austrian 33%, French/Belgian 23%, Portuguese/Spanish 22%, Irish/Scottish/Welsh 13%
    K36: German Rheinland-Pfalz 35%, French North West 27%, Spain 27%, Ireland 11%
    G25: German 40%, French 28%, Spanish 23%, Irish 9%
    K16: German 50%, French 19%, French East 9%, Spanish 8%

  13. #68
    Registered Users
    Posts
    1,158
    Sex
    Location
    Viseu
    Ethnicity
    Romanised Paesuri
    Nationality
    Portuguese
    Y-DNA
    E-BY36857
    mtDNA
    H20

    Portugal 1143 Portugal 1485 Portugal Order of Christ
    Quote Originally Posted by Theconqueror View Post
    Yes I am able as I have a PhD in epidemiology and a BSc and MSc in stats and economics but it is besides the point. It is an interesting artefact but only slightly useful, that's all.
    Ok, I have a question then. Do you prefer scaled on unscaled coordinates?
    (I hope I don't start a small forum war over this)
    YDNA - E-Y31991>PF4428>BY36857. Domingos Rodrigues, b. circa 1680 Hidden Content , Viana do Castelo, Portugal
    mtDNA - H20. Maria Josefa de Almeida, b. circa 1750 Hidden Content , Porto, Portugal

    Global25 PCA West Eurasia dataset Hidden Content
    Hidden Content


    [1] "distance%=1.7726"

    Ruderico

    Celtiberian,77.6
    Roman_Imperial_proxy,13.6
    Guanche,8.8

  14. The Following User Says Thank You to Ruderico For This Useful Post:

     JMcB (11-09-2018)

  15. #69
    Registered Users
    Posts
    317
    Sex

    As a general rule for any PCA, the short of it is scaling preserves the real size of differentiation in the underlying data (and scaled differences correlate very well with population differentiation based on Fst frequency differences, more so the more accurate dimensions are captured).

    Because the unscaled data treats all dimensions as exactly the same size, the distance from between two samples becomes a basically a function of their average separation across all the dimensions, and really how many dimensions they are separated in. Without any regard to if one dimension represents a very large genetic distance and another a very about of differentiation (e.g. an unscaled dataset of 60 dimension PCA will treat a Celts vs Germans dimension, if it pops out at PC 60, as of exactly the same size as African vs Eurasian at PC 1). It's very important to note that "unscaled" does not really imply that the set already captures real distance and "scaled" is reinforcing and compounding distance in a way that exaggerates real genetic distance; it's rather that unscaled data doesn't capture any distance that is expressed in the eigenvalues for each dimension.

    Functionally this doesn't change the rank order of which populations are closest so much... but it very much changes the populations are most distant at increasing real distance.

    The usual effect of this seems to be to me is that if you a sample is very central on a number of dimensions relating to, e.g. West Eurasian variation, and if it's not well proxied by other samples in the dataset in a way that minimizes distance, then it will have some greater tendency to pick up ancestry from outgroups that also fall closer to the centre on those dimensions. There's also the converse effect.

    I don't see any reason to use the scaled data and so never do, as never seems to have any upside to me, but in many instances the effect will be relatively minimal if a sample is well proxied by a number of close samples in your calc file. The most important thing is not to mix values from scaled and unscaled versions of the datasheet.

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

     anglesqueville (11-09-2018),  Capitalis (11-10-2018),  JMcB (11-09-2018),  poi (11-09-2018),  Ruderico (11-09-2018),  Ryukendo (01-28-2019)

  17. #70
    Registered Users
    Posts
    1,272
    Sex
    Location
    France
    Ethnicity
    Breton & Maineau
    Nationality
    French
    Y-DNA
    R-L21 (DF13*)
    mtDNA
    U4c1a

    France France Bretagne
    Quote Originally Posted by Tolan View Post

    My result:
    [1] "distance%=1.3982"

    Tolan_scaled

    Central-Euro,50.2
    South-Atlantic,23.2
    Channel-North-Sea,15.2
    North-Atlantic,6.6
    South-Euro,4.8
    OK: Here with unscaled result:

    [1] "distance%=0.8197"

    Tolan

    Central-Euro,49.8
    South_Atlantic,31.8
    North-Atlantic,12.8
    Channel-North-Sea,3.2
    South_Euro,2.4

    But the problem of nMonte: it is not necessarily the nearest solution that is the truest!
    It would be necessary to have the first 10 possibilities to have a better idea!
    Last edited by Tolan; 11-09-2018 at 06:45 PM.
    Y haplogroup: R1b: L21+ DF13+ (L1335- DF21- DF49- FGC11134- L513- Z251- Z253- CTS1751- CTS3386- DF41- FGC5496- L371- MC14- S1026- S1051- S16264- Z16500- Z16502- Z255-)
    For my autosomal analyses, see Hidden Content

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

     Dewsloth (11-09-2018),  Helgenes50 (11-09-2018),  JMcB (11-09-2018),  Ruderico (11-09-2018)

Page 7 of 23 FirstFirst ... 5678917 ... LastLast

Similar Threads

  1. Replies: 0
    Last Post: 06-24-2018, 08:50 AM
  2. Replies: 15
    Last Post: 01-30-2017, 09:51 PM
  3. "Indus Valley" and "Kalash" admixtures in Khvalynsk?
    By Tomenable in forum Ancient (aDNA)
    Replies: 12
    Last Post: 08-12-2016, 10:46 PM
  4. Replies: 43
    Last Post: 01-18-2015, 02:11 PM
  5. Among Europeans, who have "east asian"?
    By Tolan in forum Autosomal (auDNA)
    Replies: 2
    Last Post: 11-25-2013, 09:36 PM

Posting Permissions

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