Page 1 of 2 12 LastLast
Results 1 to 10 of 15

Thread: Quality of Raw Data of 23andme for G25 and Admix Calculators

  1. #1
    Registered Users
    Posts
    142
    Sex
    Ethnicity
    Venetian + CS Italian
    Y-DNA (P)
    I-Z58
    mtDNA (M)
    H7

    Republic of Venice Friuli Vatican

    Quality of Raw Data of 23andme for G25 and Admix Calculators

    Hello everyone, I know it may be a complicated question and maybe it would be better just to ask Davidski, but maybe some of you have a similar experience.

    I find 23andme ancestry estimates quite good, or, to be more precise, by far the best we can find around at this given time.

    However, I was a bit puzzled at the start with the results that the calculators were giving me on Gedmatch/Yourdnaportal etc. I took it as a game, of course, but I was wondering why would I plot differently from what I would have expected.

    One day I downloaded Admixture Studio and saw that the genotype ratio used by calculators for my data was around 20-30%. I've read online of people creating superkits and so on and I thought I could do something like that in the future.

    While I was looking for further tests to do for my haplogroup (which right now is I-Z58 but I want to dig deeper and see if there is any subclade I can join or if I am a basal I-Z58 etc) I've found out about the WGS from several companies, in particular YSEQ.

    In the meantime, I've decided to send my raw data to Davidski and got my coordinates for G25, where I plot a bit closer to what I expected compared to admixiture calculators (still just a bit south and a bit east to what I would've expected, but nothing dramatic).
    I've also read here and there that 23andme is not good for raw data because they make estimates based on previous calls which work quite well for their own ancestry estimates but not at all with anything else.

    I was wondering if any of you compared results from G25 and calculators obtained with 23andme raw data vs WGS and if there was an actual big difference. That could make a slight difference in choosing YSEQ vs FTDNA Big-Y 700 for Y-DNA since I'd love to see my aDNA results improving in their quality if that's a possibility.

    Thanks a lot and looking forward to hearing your opinion!
    Deep Ancestry: 63.2% AUT_LBK_N + 36.8% RUS_Vonyuchka_En @ 2.6339%
    Ancient: 89.0% ITA_Rome_Late_Antiquity + 11.0% HUN_Avar_Szolad @ 2.4649%

  2. #2
    Registered Users
    Posts
    18
    Sex

    I noticed difference between myheritage and g25. Each has its own estimates however, from company to company it varies a lot.

  3. #3
    Registered Users
    Posts
    367
    Sex
    Ethnicity
    Greek
    Nationality
    Australian
    Y-DNA (P)
    J2a
    mtDNA (M)
    J1

    Australia Greece Byzantine Empire
    The only thing I will say is that my friends who purchased G25 coords with the current versions of MyHeritage and 23andMe raw data had significantly larger distances to populations they belonged to. When comparing with the Vahaduo Difference tool, it showed they had a much higher amount of an ancient component (usually Farmer, but sometimes WHG) than normal for modern populations. I believe it is optimal to use as many SNPs as possible that are used in the production of your G25 coord for the most accurate results.

    Out of the commercial sites, it seems AncestryDNA might have the best SNP overlap with Davidski's SNP set but I am not entirely certain. My AncestryDNA coords seem to be much better than my 23andMe ones, as the fits improve and the shift towards Anatolia makes more sense than the overinflated European Farmer shift I had on my 23andMe coords.

    Here is a comparison of my 23andMe and AncestryDNA coords:

    I have lower distances to populations I am mostly belonging to with AncestryDNA data than 23andMe data:



    I match more strongly Anatolian populations with my 23andMe data than my AncestryDNA data, and less strongly to populations with higher European Farmer / WHG ancestry than myself.





    I can't say whether it is more accurate in truth, because I am not entirely sure of the genetic admixture some of my ancestors were expected to have. With high confidence I should roughly be 7/8 Dodecanese and 1/8 coming from Anatolia with an unknown genetic profile. The AncestryDNA produced coords suggest I am shifted towards Anatolia with either a Central Anatolian shift or Pontic or Armenian shift which I believe is probably likely given that my mother scores as much CHG as Cappadocian Greeks in G25 models.
    Code:
    23abc_AncestryDNA_scaled,0.110408,0.151314,-0.0290383,-0.0507112,0.0018465,-0.0156179,-0.00305514,-0.00138456,-0.00899905,0.00911181,0.00243583,-0.00149867,-0.00431116,0.00344057,-0.00773606,0.00106072,0.00195576,0.00152026,0.00251396,-0.00550264,-0.00786113,-0.00197844,0.0025882,0.00168699,0.000957998

  4. The Following 2 Users Say Thank You to 23abc For This Useful Post:

     Northern Adriatic (04-05-2021),  Rufus191 (04-06-2021)

  5. #4
    Registered Users
    Posts
    142
    Sex
    Ethnicity
    Venetian + CS Italian
    Y-DNA (P)
    I-Z58
    mtDNA (M)
    H7

    Republic of Venice Friuli Vatican
    Quote Originally Posted by 23abc View Post
    The only thing I will say is that my friends who purchased G25 coords with the current versions of MyHeritage and 23andMe raw data had significantly larger distances to populations they belonged to. When comparing with the Vahaduo Difference tool, it showed they had a much higher amount of an ancient component (usually Farmer, but sometimes WHG) than normal for modern populations. I believe it is optimal to use as many SNPs as possible that are used in the production of your G25 coord for the most accurate results.

    Out of the commercial sites, it seems AncestryDNA might have the best SNP overlap with Davidski's SNP set but I am not entirely certain. My AncestryDNA coords seem to be much better than my 23andMe ones, as the fits improve and the shift towards Anatolia makes more sense than the overinflated European Farmer shift I had on my 23andMe coords.

    Here is a comparison of my 23andMe and AncestryDNA coords:

    I have lower distances to populations I am mostly belonging to with AncestryDNA data than 23andMe data:



    I match more strongly Anatolian populations with my 23andMe data than my AncestryDNA data, and less strongly to populations with higher European Farmer / WHG ancestry than myself.





    I can't say whether it is more accurate in truth, because I am not entirely sure of the genetic admixture some of my ancestors were expected to have. With high confidence I should roughly be 7/8 Dodecanese and 1/8 coming from Anatolia with an unknown genetic profile. The AncestryDNA produced coords suggest I am shifted towards Anatolia with either a Central Anatolian shift or Pontic or Armenian shift which I believe is probably likely given that my mother scores as much CHG as Cappadocian Greeks in G25 models.
    Thanks a lot for sharing your experience! I do recall reading a similar comment on AncestryDNA having the best raw data among commercial websites and 23andme the worst. Interesting to see how with 23andme you were more "italian-like" and with Ancestry a bit closer to your paper trail. I guess that even in PCAs you may plot differently using the two sets of coordinates.

    I am not so keen on taking another commercial test because I think they work well as first introduction to DNA and I am quite satisfied with what 23andme has done. To bring it to the next level I guess, and hope, that WGS could do the job.
    Deep Ancestry: 63.2% AUT_LBK_N + 36.8% RUS_Vonyuchka_En @ 2.6339%
    Ancient: 89.0% ITA_Rome_Late_Antiquity + 11.0% HUN_Avar_Szolad @ 2.4649%

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

     23abc (04-05-2021)

  7. #5
    Registered Users
    Posts
    1,719
    Sex

    Here is a comparison chart for the SNP's tested and how they overlap between the companies:
    https://isogg.org/wiki/Autosomal_SNP_comparison_chart

    It always depends on your version, since most companies changed their set up. Also we don't really know for which raw data G25 being optimised, or do we?

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

     JMcB (04-05-2021),  Rufus191 (04-06-2021)

  9. #6
    Registered Users
    Posts
    142
    Sex
    Ethnicity
    Venetian + CS Italian
    Y-DNA (P)
    I-Z58
    mtDNA (M)
    H7

    Republic of Venice Friuli Vatican
    Quote Originally Posted by Riverman View Post
    Here is a comparison chart for the SNP's tested and how they overlap between the companies:
    https://isogg.org/wiki/Autosomal_SNP_comparison_chart

    It always depends on your version, since most companies changed their set up. Also we don't really know for which raw data G25 being optimised, or do we?
    Thanks for the first link. Impressive the change between v_3 and v_4 of 23andme (I've got tested recently with v_5). I assume a WGS like the WGS 400 of YSEQ to be more effective.

    And also, that's a good point. I will try to ask also Davidski opinion!
    Deep Ancestry: 63.2% AUT_LBK_N + 36.8% RUS_Vonyuchka_En @ 2.6339%
    Ancient: 89.0% ITA_Rome_Late_Antiquity + 11.0% HUN_Avar_Szolad @ 2.4649%

  10. #7
    Registered Users
    Posts
    340
    Sex
    Nationality
    Brazilian
    Y-DNA (P)
    R1b
    mtDNA (M)
    H1

    Brazil Azores Italy Ladinia
    Doing a WGS will not make any significant difference in comparison with results from the commercial companies. The reason for that is that you will have to convert your WGS file to a commercial company template in order for it to work on GEDmatch or Global 25. At the present moment there isn't a single autosomal test available to most people that can extract the full power of a WGS.

    I think if you have 23andme + Global 25 it's the best you could have by now. Possibly some components may change depending of the raw data due to the issues of estimating proportions of ancestry with components as old as Yamnaya, Barcin, European HG etc. Some of those components have overlapping ancestry, like both Barcin_N and Yamnaya having different types of HG ancestry (the reality is that most Europeans have predominantly West Eurasian Mesolithic ancestry), it certainly can interfere in the results and you must consider that modern European populations are very mixed with different sources of those components.

    Also, do not waste your money testing with AncestryDNA only for the raw data, their raw data is indeed better than MyHeritage and 23andme (I've tested in all of them), but the differences aren't going to be immense and you will have basically the same results you have now with slight differences or shifts, with AncestryDNA raw data results probably being slightly more accurate but not always. Moreover, despite being a relatively accurate company for autosomal results only behind 23andme, AncestryDNA isn't worth the price since they don't give haplogroups and you need to pay for pretty much anything on their website. A WGS is indeed a better choice but not for using on the tools we have now, but for tools that will certainly be developed in the future.

  11. The Following 2 Users Say Thank You to Caius Agrippa For This Useful Post:

     Mixed (04-05-2021),  Northern Adriatic (04-05-2021)

  12. #8
    Registered Users
    Posts
    1,891
    Sex
    Location
    Bay Area
    Nationality
    American
    Y-DNA (P)
    E-BY14160
    mtDNA (M)
    J1c
    mtDNA (P)
    I1a1

    Germany Italy Sweden Netherlands
    I have no scientific explanation but I know my 23 and Me raw data isn’t as “accurate” as the AncestryDNA, MyHeritage, FTDNA etc.. in terms of Gedmatch. I’ve seen G25 results that basically show the same thing is happening so I never bothered getting G25 coordinates with 23 and Me data.
    distance%=4.6465"
    Barcin_N,47.2
    Yamnaya_Samara,41.4
    WHG,10.6
    Ethiopia_4500BP,0.8

  13. The Following User Says Thank You to digital_noise For This Useful Post:

     Northern Adriatic (04-05-2021)

  14. #9
    Registered Users
    Posts
    340
    Sex
    Nationality
    Brazilian
    Y-DNA (P)
    R1b
    mtDNA (M)
    H1

    Brazil Azores Italy Ladinia
    Quote Originally Posted by digital_noise View Post
    I have no scientific explanation but I know my 23 and Me raw data isn’t as “accurate” as the AncestryDNA, MyHeritage, FTDNA etc.. in terms of Gedmatch. I’ve seen G25 results that basically show the same thing is happening so I never bothered getting G25 coordinates with 23 and Me data.
    MyHeritage new version is as bad as 23andme in my experience. The only company that does not input data is AncestryDNA. But in reality all of those raw datas are designed to work better on the algorithms of these companies, so you should be very cautious when getting results from any test with them,

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

     Nino90 (04-05-2021),  Riverman (04-05-2021)

  16. #10
    Registered Users
    Posts
    1,891
    Sex
    Location
    Bay Area
    Nationality
    American
    Y-DNA (P)
    E-BY14160
    mtDNA (M)
    J1c
    mtDNA (P)
    I1a1

    Germany Italy Sweden Netherlands
    Ah, I wasn’t aware that MH updated or changed their raw files. I think mine was from 2018-2019
    distance%=4.6465"
    Barcin_N,47.2
    Yamnaya_Samara,41.4
    WHG,10.6
    Ethiopia_4500BP,0.8

Page 1 of 2 12 LastLast

Similar Threads

  1. Raw DNA data and calculators
    By Prunj7 in forum Inquiries Corner
    Replies: 4
    Last Post: 11-03-2019, 10:28 PM
  2. Replies: 0
    Last Post: 03-28-2019, 09:20 AM
  3. Any other admix calculators to try?
    By gjenetiks in forum Autosomal (auDNA)
    Replies: 12
    Last Post: 09-23-2018, 07:32 PM
  4. Replies: 1
    Last Post: 05-01-2018, 10:07 PM
  5. Replies: 5
    Last Post: 04-20-2018, 07:42 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
  •