Page 1 of 7 123 ... LastLast
Results 1 to 10 of 68

Thread: SSA % based on gedmatch's population oracles

  1. #1
    Registered Users
    Posts
    235
    Sex

    SSA % based on gedmatch's population oracles

    So several months ago i was given 27% SSA by Geno 2.0 yet when it came to population oracles nearly half of gedmatch's calculators failed to detect any SSA pop. out of 20 samples.

    Knowingly that the population oracles by gedmatch is usually based on the ethnic results that you pass in to them; but it seems the algorithmic comparison between geno 2.0 and gedmatch is different; and they interpret the numbers differently, so i decided to do a study on my own to get to the bottom of things.

    Note: I tried my best to exclude any gedmatch calculators that would have been deemed too irrelevant for the study

    these are the calculators i based my study on
    SSA clusters out of 20:

    MDLP Project

    MDLP K23b: 0(SSA)
    MDLP world 22: 1(SSA)
    MDLP world: 3(SSA)

    Deocad

    Deocad world9: 3(SSA)
    Deocad k12b: 0(SSA)
    Deocad k7b: 2(SSA)
    Deocad v3: 0(SSA)

    Harrapaworld(sheds light on East African ethnicity)

    Harrapaworld: 0(SSA)

    PuntDNAL(African based)

    PuntDNAL k10 Ancient: 2(SSA)
    PuntDNAL k12 Modern: 3(SSA)
    PuntDNAL k15: 5(SSA)

    Eurogene

    Eurogene k13: 1(SSA)
    Eurogene v2k15: 0.5(SSA)
    Eurogene Jtest: 3(SSA)
    Eurogene ea test: 2(SSA)

    Gedrosia

    Neareast Neolithic K13: 0(SSA)
    Ancient Eurasia k6: 0(SSA)
    Gedrosia k3: 1(SSA)
    Gedrosia k11: 0(SSA)
    Eurasia k10 CHG: 0.5(SSA)
    Eurasia k9 ASI: 0(SSA)
    Gedrosia k15: 1(SSA)
    Eurasia k14 Neolithic: 0(SSA)

    Overall = 6.0%(SSA)



    SSA populations found in these samples were exclusively North East African and they usually range between 21.00-35.00; i also read somewhere that these populations lose their relevancy once they pass 16.00; so these SSA populations would have most likely been distant.

    Anyways i added all SSA numbers which amounted up to 28 and divided it by the total which was 460 and that gave me 6%.

    Obviously if i added or took away some calculators the percentage will change slightly but anyways based on these calculations i estimate my ancestry to be around 6% SSA.
    Last edited by Missouri1455; 01-13-2017 at 11:40 PM.

  2. The Following User Says Thank You to Missouri1455 For This Useful Post:

     Principe (01-13-2017)

  3. #2
    Registered Users
    Posts
    235
    Sex

    If i included more calculators; then the SSA % would most likely stabilize at 5% ill post them down
    Last edited by Missouri1455; 01-13-2017 at 11:18 PM.

  4. #3
    Registered Users
    Posts
    235
    Sex

    Ill add 4 calculators i did not include in the first post; and this SSA % will conclude my SSA % based on 27 population oracles; in total Gedmatch has 28 population oracles:

    Puntdnal K12 Ancient: 0(SSA)
    Gedorisa K12: 0(SSA)
    Eurasia K11 CHG-NAF: 0(SSA)
    Eurasia K14: 0(SSA)

    and with that the total % for my SSA stabilizes at 0.05 = 5%

    So based on 27 gedmatch oracles my SSA is 5% and not 27% that Geno 2.0 showed me; so with geno 2.0 i believe algorithms played a big role in inflating my SSA %

    Plus Geno 2.0 gave me 7% west African yet gedmatch population oracles failed to detect even one West African in the 27 oracle populations; so this proves that 7% west african was most likely "noise"

    I did not include "Puntdnal k8 african only" because it seemed solely focused on african populations so it will run out of accuracy if someone outside of africa used it and its not going to be able to reference me with an accurate population(Middle east) because it focuses solely on african populations.
    Last edited by Missouri1455; 01-13-2017 at 11:40 PM.

  5. #4
    Registered Users
    Posts
    95
    Sex
    Location
    None Ya
    Ethnicity
    Multi-ethnic Belizean
    Nationality
    Belizean/American
    Y-DNA (P)
    E1b1a1g1, Q1a3a1
    mtDNA (M)
    L2b2, L3e2b

    Belize CARICOM United States of America African Union Belize
    What's your GEDMatch ID?

  6. #5
    Gold Class Member
    Posts
    3,147
    Sex
    Location
    Pennsylvania
    Ethnicity
    West European
    Nationality
    USA
    Y-DNA (P)
    R1b-U152-Z36+FGC6511
    mtDNA (M)
    H11a2a

    United States of America Germany England France Scotland Ireland
    This seems odd. I started with Geno 2.0 tests for me, my wife, and my father in law a while ago now. All of our percentages are somewhat in the ballpark of what we get on Gedmatch calculators. You use the word above "Oracles". I would be more interested in seeing your list of admixture percentages from a test like Eurogenes K13 than an Oracle result.

  7. #6
    Registered Users
    Posts
    235
    Sex

    Quote Originally Posted by pgbk87 View Post
    What's your GEDMatch ID?
    Let me know what you want to know and ill post it here

  8. #7
    Registered Users
    Posts
    235
    Sex

    Quote Originally Posted by randwulf View Post
    This seems odd. I started with Geno 2.0 tests for me, my wife, and my father in law a while ago now. All of our percentages are somewhat in the ballpark of what we get on Gedmatch calculators. You use the word above "Oracles". I would be more interested in seeing your list of admixture percentages from a test like Eurogenes K13 than an Oracle result.
    The ethnic percentages resembles geno 2.0 results because the results are pretty much a "transfer" and won't mean much unless i redo the test with another dna company.

    Gedmatch population distance failed to detect any african clusters and if it detected any they range between 21.00-35.00 which is considered very distant from my actual result; and population distances start to lose relevancy around 16.00 and if any east african pop. was detected it would barley be more than one pop. unless the admix was "african" focused like puntdnal.


    Overall I scored 5% for my SSA which is equivalent to one SSA cluster out of 20.

    Oracles do mean a lot especially when it comes to detecting minority admixtures it clears out any doubts; for an ethnic makeup 27% is a pretty high number, there's no reason why it should go undetected multiple times unless it was highly inflated.

    Almost 50% of Gedmatch's calculators failed to detect any SSA in my oracles


    this is my Eurogene K13 results:

    1 East_Med 35.64
    2 Red_Sea 25.46
    3 Northeast_African 15.49
    4 Sub-Saharan 11.38

    5 West_Asian 4.13
    6 North_Atlantic 2.53
    7 West_Med 1.81
    8 South_Asian 1.62
    9 Siberian 1.3
    10 Oceanian 0.38
    11 East_Asian 0.26

    Even though I have SSA of approx. 27%
    The population distance below failed to detect any SSA pop.


    # Population (source) Distance
    1 Egyptian 13.91
    2 Bedouin 18.99
    3 Saudi 21.53
    4 Yemenite_Jewish 22.64
    5 Jordanian 23.12
    6 Tunisian 23.45
    7 Palestinian 23.79
    8 Algerian 23.79
    9 Mozabite_Berber 25.41
    10 Moroccan 25.56
    11 Syrian 27.66
    12 Libyan_Jewish 28.2
    13 Tunisian_Jewish 29.29
    14 Samaritan 29.76
    15 Lebanese_Muslim 30.65
    16 Lebanese_Christian 31.96
    17 Lebanese_Druze 32.44
    18 Algerian_Jewish 32.83
    19 Sephardic_Jewish 32.96
    20 Italian_Jewish 34.12

    As you can see i have 0 SSA again; and this have happend repeatedly with different calculators so this shows me that SSA should have not been that high.
    Last edited by Missouri1455; 01-14-2017 at 01:37 AM.

  9. #8
    Gold Class Member
    Posts
    3,147
    Sex
    Location
    Pennsylvania
    Ethnicity
    West European
    Nationality
    USA
    Y-DNA (P)
    R1b-U152-Z36+FGC6511
    mtDNA (M)
    H11a2a

    United States of America Germany England France Scotland Ireland
    I see. The single population Oracles don't work the way you are describing. They aren't going to find SSA populations when you are roughly 1/4 African. They are going to report the best partial African populations they can find. None of them are really good matches, as you can see, since the distances are high. The single search isn't going to report minority ancestry. The Mixed Mode Oracle results might. In your case, the four-population Oracles might actually work better than they do for most people. However, the Oracles really are a secondary process involving a search of a spreadsheet using your admixture results, which are the primary result of the calculator.

    I don't agree with your "transfer" statement. Your raw results at the tested SNPs are transferred. These are processed by the calculator independently. How many SNPs are reported at the bottom of your Eurogenes K13 test run? If there are only a couple of thousand, you could be getting skewed results.

  10. The Following 8 Users Say Thank You to randwulf For This Useful Post:

     Cinnamon orange (01-29-2017),  Deftextra (01-14-2017),  drobbah (01-14-2017),  kikkk (01-14-2017),  King (01-14-2017),  pgbk87 (01-14-2017),  Trixster (08-19-2017),  xKeleix (01-14-2017)

  11. #9
    Gold Class Member
    Posts
    1,572
    Sex
    Omitted
    Location
    Arabsiyo,Somaliland
    Nationality
    Canadian
    Y-DNA (P)
    E-V32>E-FT385910
    mtDNA (M)
    L0a1d
    Y-DNA (M)
    T-FGC92488

    Somaliland Adal Sultanate Yemen African Union Canada
    You seem to have some Horner and Bantu ancestry which is normal considering Yemen's history.I think the SSA component in EuroGenes K13 is based off W.Africans.You score almost double of what I get

    My Eurogenes results in comparison:

    Admix Results (sorted):

    # Population Percent
    1 Northeast_African 56.8
    2 Red_Sea 23.65
    3 East_Med 11
    4 Sub-Saharan 6.27
    5 West_Med 0.99
    6 Siberian 0.67
    7 Oceanian 0.62

    Single Population Sharing:

    # Population (source) Distance
    1 Somali 1.46
    2 Ethiopian_Oromo 4.78
    3 Ethiopian_Wolayta 5.24
    4 Ethiopian_Amhara 14.22
    5 Ethiopian_Tigray 16.3
    6 Ethiopian_Ari_cultivator 24.11
    7 Maasai 31.41
    8 Ethiopian_Gumuz 36.75
    9 Hadza 37.29
    10 Sandawe 40.22
    11 Ethiopian_Anuak 44.21
    12 Sudanese 47.74
    13 Moroccan 55.08
    14 Egyptian 55.4
    15 Algerian 56.67
    16 Mozabite_Berber 56.94
    17 Tunisian 58.26
    18 Bedouin 60.87
    19 Jordanian 62.51
    20 Saudi 64.44

  12. The Following 5 Users Say Thank You to drobbah For This Useful Post:

     Angoliga (01-19-2017),  Deftextra (01-14-2017),  Missouri1455 (01-14-2017),  pgbk87 (01-14-2017),  randwulf (01-14-2017)

  13. #10
    Registered Users
    Posts
    235
    Sex

    Quote Originally Posted by randwulf View Post
    I see. The single population Oracles don't work the way you are describing. They aren't going to find SSA populations when you are roughly 1/4 African. They are going to report the best partial African populations they can find. None of them are really good matches, as you can see, since the distances are high. The single search isn't going to report minority ancestry. The Mixed Mode Oracle results might. In your case, the four-population Oracles might actually work better than they do for most people. However, the Oracles really are a secondary process involving a search of a spreadsheet using your admixture results, which are the primary result of the calculator.

    I don't agree with your "transfer" statement. Your raw results at the tested SNPs are transferred. These are processed by the calculator independently. How many SNPs are reported at the bottom of your Eurogenes K13 test run? If there are only a couple of thousand, you could be getting skewed results.
    I see, thanks for the clarification; yes it is average

    In terms of SNPs; the more snps the accurate it is?

    Does ancestry/23ndme use more SNPs than geno 2.0?

    Do you have any idea on how the different companies do these algorithms?

    Thank you
    Last edited by Missouri1455; 01-14-2017 at 03:01 AM.

Page 1 of 7 123 ... LastLast

Similar Threads

  1. GEDMatch Oracles
    By miarosie in forum Autosomal (auDNA)
    Replies: 3
    Last Post: 03-03-2020, 12:37 AM
  2. GEDMatch Oracles
    By miarosie in forum Autosomal (auDNA)
    Replies: 1
    Last Post: 02-16-2020, 10:58 PM
  3. Guess this admixture kit by the GEDmatch oracles
    By Ahaddad in forum Autosomal (auDNA)
    Replies: 7
    Last Post: 08-05-2019, 01:10 AM
  4. Share some of your favouarate GEDmatch Oracles
    By SWAHILLI_PRINCE16 in forum Autosomal (auDNA)
    Replies: 3
    Last Post: 12-11-2017, 05:40 AM
  5. Using GEDmatch Oracles for matches between populations
    By Caratacus in forum Inquiries Corner
    Replies: 4
    Last Post: 02-03-2016, 10:15 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
  •