Announcement

Collapse
No announcement yet.

Native American % way off??

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Native American % way off??

    I'm so confused. I tested with ancestry and then uploaded my raw data to FamilyTreeDNA and my results are completely different, especially my native american %.

    Ancestry:

    Africa: 7%
    Native American: 45%
    Europe: 38%
    - Iberian Penisula 18%
    - Great Britain 14%
    - Irish 4%
    - European Jewish %1
    - Italy/Greece <1%
    West Asia 9%
    - Caucus 5%
    - Middle East 4%


    Family Tree DNA


    European %49
    - Southern Europe %35
    - Scandinavian %8
    - British Isles %5
    - Finland and North Siberian %1

    New World : %28

    East Asian: %10
    - North East Asia %10

    Central/South Asia: %7
    - Central Asia

    African: %5
    - West Africa

    Middle Eastern %1

    GEDMatch:


    North_Atlantic 18.25
    Baltic 5.34
    West_Med 13.64
    West_Asian 3.80
    East_Med 6.63
    Red_Sea 2.04
    South_Asian 0.47
    East_Asian -
    Siberian 2.64
    Amerindian 40.64
    Oceanian -
    Northeast_African 1.91
    Sub-Saharan 4.63


    Why the huge % change in Native American and European?? So confused. Any ideas?

  • #2
    I have been complaining about the misreporting of Native American in myOrigins at FTDNA for a long time. I see it happen with every single Latin American that has Native American ancestry. It even happened to the 13,000 year old 100% Clovis Anzick-1 infant when the DNA was uploaded to FTDNA http://www.fi.id.au/2014/11/clovis-a...-in-ftdna.html Compare that to his makeup in Gedmatch at http://www.fi.id.au/2014/10/new-clov...-gedmatch.html

    FTDNA myOrigins is the worst calculator when it comes to Native American DNA. Both Ancestry and 23andme do a better job.

    Comment


    • #3
      Welcome to our message board, AwesomePossum.

      As Armando said, we've seen these things happen many times. What happened in your case is

      46% as the total of 28% New World + 10% Northeast Asia + 7% Central Asia + 1% Finland in MyOrigins
      equals
      45% Native American in AncestryDNA

      This is because the Maya and Pima reference samples that had been included in an older calculator called Population Finder used here until 2014 are currently not included in MyOrigins because some of them were deemed impure (many of them are mixed to a limited extent with peoples from the Iberian peninsula).

      And in many GEDmatch calculators, the portions assigned to "Siberian" should be mentally added to your "Amerindian", so that 40.64% Amerindian + 2.64% Siberian = 43.28% real Native American for you.

      I see that AncestryDNA gave you a 1% European Jewish estimate. Out of curiosity, is your Middle Eastern portion categorized as North African, Asia Minor, or Eastern Middle East by MyOrigins? Some Sephardic DNA is classified under those categories, or under Southern European, by MyOrigins. Other Sephardic DNA is classified as "Ashkenazi Diaspora", but only tells the customer that when the person has at least 5% of that kind of DNA. I assume you have valid Jewish matches in GEDmatch.
      Last edited by khazaria; 9 May 2016, 10:36 PM.

      Comment


      • #4
        Thank you guys so much for your help in trying to understand more about how they calculate admixtures.

        Here's my results from MyOrigins (added it as an attachment below)
        They list middle eastern just under "Middle Eastern." The Native American % you explained makes a lot of sense. 43.28% All three also mix the European portions. Ancestry seems to lump a lot into Great Britain. The other two break that up a bit more. My Origins confuses me on the European side too.


        Originally posted by khazaria View Post
        Welcome to our message board, AwesomePossum.

        As Armando said, we've seen these things happen many times. What happened in your case is

        46% as the total of 28% New World + 10% Northeast Asia + 7% Central Asia + 1% Finland in MyOrigins
        equals
        45% Native American in AncestryDNA

        This is because the Maya and Pima reference samples that had been included in an older calculator called Population Finder used here until 2014 are currently not included in MyOrigins because some of them were deemed impure (many of them are mixed to a limited extent with peoples from the Iberian peninsula).

        And in many GEDmatch calculators, the portions assigned to "Siberian" should be mentally added to your "Amerindian", so that 40.64% Amerindian + 2.64% Siberian = 43.28% real Native American for you.

        I see that AncestryDNA gave you a 1% European Jewish estimate. Out of curiosity, is your Middle Eastern portion categorized as North African, Asia Minor, or Eastern Middle East by MyOrigins? Some Sephardic DNA is classified under those categories, or under Southern European, by MyOrigins. Other Sephardic DNA is classified as "Ashkenazi Diaspora", but only tells the customer that when the person has at least 5% of that kind of DNA. I assume you have valid Jewish matches in GEDmatch.
        Attached Files
        Last edited by AwesomePossum; 9 May 2016, 10:48 PM.

        Comment


        • #5
          Originally posted by Armando View Post
          I have been complaining about the misreporting of Native American in myOrigins at FTDNA for a long time. I see it happen with every single Latin American that has Native American ancestry. It even happened to the 13,000 year old 100% Clovis Anzick-1 infant when the DNA was uploaded to FTDNA http://www.fi.id.au/2014/11/clovis-a...-in-ftdna.html Compare that to his makeup in Gedmatch at http://www.fi.id.au/2014/10/new-clov...-gedmatch.html

          FTDNA myOrigins is the worst calculator when it comes to Native American DNA. Both Ancestry and 23andme do a better job.

          Yeah, I am going to test with 23andme to see what they give me. From what I hear they give you more details and more tooks to utilize.

          Comment


          • #6
            Originally posted by AwesomePossum View Post
            They list middle eastern just under "Middle Eastern.
            "Eastern Middle East" according to the screenshot. Thanks for that. Some European Jews do show small percentages of that in MyOrigins.

            Originally posted by AwesomePossum View Post
            Yeah, I am going to test with 23andme to see what they give me. From what I hear they give you more details and more tooks to utilize.
            Okay, post your future 23andme results for us in this thread later this year, if that's what you end up doing. They took away a few of their tools but their ethnicity estimates are often more accurate than the other sites.

            But to be honest, I don't think you really need 23andme because you're already in all the best databases for matching with relatives, and their ethnicity calculations are close enough to reality that you don't really have to doubt your ancestry. It might be better to spend the money on a different autosomal kit for a relative of yours such as if you have a living parent or grandparent or aunt/uncle or a fullblood sibling.

            Comment


            • #7
              Thanks so much for your help! I guess I'm curious to see how they estimate my native ancestry and European ancestry. I know these are estimates and it's hard to distinguished between certain areas but really do want to learn more. It shocked me to learn about the Great Britain percentage on ancestry and was wondering how that was determined. Also on GEDmatxh it shows that my distances are too far than most. When I check it just says that their might now be enough of my ancestry in the data base to give me off numbers. When I test the beta against 3 matches I get 50% Mayan, 25% Mozambite barber, ans 25% West English. I'm just trying to make sense of it all!

              I am planning on testing my both parents soon.

              Again thank for your help!!!



              Originally posted by khazaria View Post
              "Eastern Middle East" according to the screenshot. Thanks for that. Some European Jews do show small percentages of that in MyOrigins.



              Okay, post your future 23andme results for us in this thread later this year, if that's what you end up doing. They took away a few of their tools but their ethnicity estimates are often more accurate than the other sites.

              But to be honest, I don't think you really need 23andme because you're already in all the best databases for matching with relatives, and their ethnicity calculations are close enough to reality that you don't really have to doubt your ancestry. It might be better to spend the money on a different autosomal kit for a relative of yours such as if you have a living parent or grandparent or aunt/uncle or a fullblood sibling.

              Comment


              • #8
                Originally posted by AwesomePossum View Post
                I am planning on testing my both parents soon.
                Good, this is what I was recommending. After that happens, you'll be able to create a phased kit for yourself inside GEDmatch. That'll help you find which segments and ethnicities you inherited come from which parent.

                As for your British and Irish ethnic affinities, which might be ancient, many Spanish and Portuguese people score multiple percentage points of those in AncestryDNA. This thread shows some examples:
                http://www.forumbiodiversity.com/sho...nyone-have-any
                So those affinities usually come to Latin Americans like Mexicans through their Northern Spanish and Basque ancestors.

                Comment


                • #9
                  Originally posted by AwesomePossum View Post
                  Thanks so much for your help! I guess I'm curious to see how they estimate my native ancestry and European ancestry. I know these are estimates and it's hard to distinguished between certain areas but really do want to learn more. It shocked me to learn about the Great Britain percentage on ancestry and was wondering how that was determined. Also on GEDmatxh it shows that my distances are too far than most. When I check it just says that their might now be enough of my ancestry in the data base to give me off numbers. When I test the beta against 3 matches I get 50% Mayan, 25% Mozambite barber, ans 25% West English. I'm just trying to make sense of it all!

                  I am planning on testing my both parents soon.

                  Again thank for your help!!!
                  23andme also does a better job of identifying Iberian vs British Isles and also Iberian vs Italian than either Ancestry or FTDNA. You really should test yourself and both of your parents at 23andme because they use the parents to phase which creates a better estimate of your ethnic breakdown but FTDNA and Ancestry do not use the parents for phasing. 23andme also provides a breakdown of the ethnicity by chromosome and even shows the ethnic breakdown from each parent by chromosome when you test your parents. So 23andme is better both with the ethnicity calculator and the DNA tools for ethnicity.

                  Comment


                  • #10
                    Originally posted by AwesomePossum View Post
                    I'm so confused. I tested with ancestry and then uploaded my raw data to FamilyTreeDNA and my results are completely different, especially my native american %.

                    Ancestry:

                    Africa: 7%
                    Native American: 45%
                    Europe: 38%
                    - Iberian Penisula 18%
                    - Great Britain 14%
                    - Irish 4%
                    - European Jewish %1
                    - Italy/Greece <1%
                    West Asia 9%
                    - Caucus 5%
                    - Middle East 4%


                    Family Tree DNA


                    European %49
                    - Southern Europe %35
                    - Scandinavian %8
                    - British Isles %5
                    - Finland and North Siberian %1

                    New World : %28

                    East Asian: %10
                    - North East Asia %10

                    Central/South Asia: %7
                    - Central Asia

                    African: %5
                    - West Africa

                    Middle Eastern %1

                    GEDMatch:


                    North_Atlantic 18.25
                    Baltic 5.34
                    West_Med 13.64
                    West_Asian 3.80
                    East_Med 6.63
                    Red_Sea 2.04
                    South_Asian 0.47
                    East_Asian -
                    Siberian 2.64
                    Amerindian 40.64
                    Oceanian -
                    Northeast_African 1.91
                    Sub-Saharan 4.63


                    Why the huge % change in Native American and European?? So confused. Any ideas?
                    your results are quite similar to my grandma's

                    Comment


                    • #11
                      Originally posted by Armando View Post
                      I have been complaining about the misreporting of Native American in myOrigins at FTDNA for a long time. I see it happen with every single Latin American that has Native American ancestry. It even happened to the 13,000 year old 100% Clovis Anzick-1 infant when the DNA was uploaded to FTDNA http://www.fi.id.au/2014/11/clovis-a...-in-ftdna.html Compare that to his makeup in Gedmatch at http://www.fi.id.au/2014/10/new-clov...-gedmatch.html

                      FTDNA myOrigins is the worst calculator when it comes to Native American DNA. Both Ancestry and 23andme do a better job.
                      Hi Armando,

                      I'm aware of the fact that you, ever since, have been complaining about the accuracy of myOrigins with regard to Native American Ancestry estimates. When I first read your posts (long time ago) I decided to email FTDNA and ask them about this issue, since my own percentage of Native American ancestry had decreased according to myOrigins' results. The answer I got is as follows:
                      the myOrigins feature is based on the most current research that has been done on ancestral population markers, as where the old Population Finder was based on research done back around 2004. We have discovered that the reference populations that Population Finder used were not as "pure" Native American as originally thought. The reference populations actually contained a fair amount of European admixture. With myOrigins, we used a much more secluded Native American population for sampling to ensure the "purest" Native American DNA, which resutled in our Bering Expansion cluster. This has caused most people with Native American ancestry to have a much higher European percentage than the previously reported with the Population Finder.

                      Thus I decided to be a bit more careful on this issue. Subsequently I have read a certain number of scientific papers in which you can find assertions that would endorse FTDNA views on this subject.

                      We also present an evaluation of the effect of European gene flow into present day Native American populations on local ancestry inference in Latinos. This effect is important to assess as it is estimated that most Native American populations used as reference panels in local ancestry inference have been exposed to European gene flow. Through simulations we find that the presence of European segments in the Native American reference panels yields biased local ancestry estimates. (Fast and accurate inference of local ancestry in Latino populations, by Yael Baran et al. April 11, 2012)

                      Many methods have been developed to infer local ancestry in admixed individuals. Early methods [24–26] relied on ancestry informative markers within hidden Markov models to achieve high accuracy. With decreasing genotyping costs, newer methods [27–30] were designed to use the increasing amount of data from genome-wide genotyping arrays while accounting for linkage disequilibrium (LD) among variants. The currently established methods [31–33] model LD in the form of haplotypes to achieve superior accuracy over non-haplotype aware approaches. Recent work in parallel to ours [34] explored the use of conditional random forests in performing local ancestry analysis. Although extremely accurate for African Americans, these methods have not achieved the same level of high accuracy in Latino Americans, partially due to the lack of good proxies for the Native American component [35] and more recent divergence among ancestral populations. (Enhanced Methods for Local Ancestry Assignment in Sequenced Admixed Individuals, by Robert Brown and Bogdan Pasaniuc, April 17, 2014).

                      So, what do you make of it, paisano? (born in the Southern Cone).

                      Comment


                      • #12
                        Hola, paisano ¿Que tal?

                        Not all of the participants of the original reference populations are "pure" but some of them are "pure" as tested by the various academic studies by different teams of scientists. All of the academic studies are able to identify the European and African admixture of Native Americans in the reference populations and use the remaining DNA as a reference for Native American DNA. That process is called masking. That's what Rasmussen et al. 2014 did for the Clovis Anzick-1 infant as seen in the ADMIXTURE graph at http://www.nature.com/nature/journal...13025_SF3.html which includes Pima, Mayans, Karitiana, Surui and many other Native American and world populations. Population Finder used Pima, Mayans, Karitiana, and Surui plus Colombians for the Native American reference populations. The Rasmussen et al. ADMIXTURE graph also includes Mexicans and you can see that they have a lot of European blue and light blue but the Pima, Mayans, Karitiana, Surui and other Native American groups do not. That's because they knew how to remove the European component. As noted on the page with the ADMIXTURE graph "the Anzick-1 sample shares all the components present in different contemporary Native American populations" So if you look at the Clovis Anzick-1 infant in myOrigins at http://www.fi.id.au/2014/11/clovis-a...-in-ftdna.html and you see those components in your myOrigins at FTDNA then it is due to the Native American ancestry if the Native American, Northeast Asian, Northern Siberia, and Central Asian adds up to an amount close to what you had in Population Finder and also adds up amounts from AncestryDNA Native American and 23andme East Asian and Native American. AncestryDNA has additional Native American samples that the other companies do not have and they normally report a slightly higher amount of Native American in Latin Americans.

                        If you download the supplement PDF from Raghavan et al. at http://science.sciencemag.org/conten...ce.aab3884.DC1 and go to page 103 you can see an ADMIXTURE graph of many world populations. This time they did not remove the European and African admixture and you can see that there are Pima that do not have European or African DNA.

                        A recent study on Northwest Coast Native Americans used Mayans in the ADMIXTURE graph and identified one Mayan as not having European or African admixture. See page 3 of the PDF of http://biorxiv.org/content/early/2016/04/29/051078

                        In Rasmussen et al. 2015 they analyzed Kennewick man. They have an ADMIXTURE graph on him also at http://www.nature.com/nature/journal...e14625_F1.html They removed the European and African admixture of the Native Americans in that one also but that one also has Clovis Anzick-1 in the graph.

                        In the Moreno et al. 2014 study they used their own samples from Mexico and show the European and African ancestry in some Native Mexicans and a lot of variation in the Cosmopolitan Mexicans with the south of Mexico having the highest amount of Native American ancestry which is what is consistent with documented history. https://d2ufo47lrtsv5s.cloudfront.ne...0/F2.large.jpg

                        AncestryDNA and 23andme also assigns Native American much closer to what is found in the documented ancestry of a lot of Latin Americans whose genealogy and DNA results that I have seen. They especially don't have Northeast Asian ancestry in the past 500 years anywhere near the amount that they have in myOrigins and none have Siberian ancestry. None of them get Finnish at AncestryDNA or 23andme either which they do a very good job of finding based on what I have seen from the results of people from Finland.

                        Here is a summary of how we know myOrigins should be able to more accurately identify Native American DNA in Latin Americans.

                        1. The academic studies are all able to distinguish European and African DNA from Native American and not assign so much of it to other groups with the same NA populations that Population Finder used.

                        2. Clovis Anzick-1 shows to have significantly different ancestry in myOrigins than what multiple academic studies have found.

                        3. The other two big DNA companies provide results closer to documented ancestry of customers even though 23andme uses the same NA reference populations that Population Finder used and AncestryDNA uses those same NA reference populations and a few more samples. When people with 75% or more Native American ancestry test with the big 3 companies the one with the most misassigned DNA is myOrigins.

                        4. Even Eurogenes K13 in Gedmatch has Mayan at 86% and Pima at 89% so Davidski was able to see they weren't 100% Native American and levels above 3% Amerindian in Eurogenes K13 are very close to AncestryDNA and 23andme.

                        Comment


                        • #13
                          Well done.

                          Comment


                          • #14
                            @Armando - it seems to me you are comparing two different things though. MyOrigins is not trying to use blocks and filter to modern populations and within a few genealogical generations like say 23 (which, BTW, hardly always meets that goal anyway, just look at British, German, French results, for instance).

                            MyOrigins does tend to map out their populations poorly though. If you look at how Geno 2.0 NG maps out groups with the same or similar names, their maps seem to make more sense, often covering vastly larger regions of different shape.
                            Last edited by wombat; 19 May 2016, 09:43 PM.

                            Comment


                            • #15
                              Originally posted by wombat View Post
                              @Armando - it seems to me you are comparing two different things though. MyOrigins is not trying to use blocks and filter to modern populations and within a few genealogical generations like say 23 (which, BTW, hardly always meets that goal anyway,...
                              23andme's official statement for the Ancestry Composition results is the "results reflect where your ancestors lived before the widespread migrations of the past few hundred years", even though of course a lot of the DNA is even older than that, but that is a lot more than a few genealogical generations. I do think that FTDNA attempted to do something similar to 23andme but was unable to do that due to a poorer methodology. Even AncestryDNA is better than FTDNA and FTDNA has admitted that myOrigins needs an update based on statements at http://www.ancestorcentral.com/11th-...tic-genealogy/. Here is the methodology from FTDNA https://www.familytreedna.com/learn/...s-methodology/ which explains that Razib Khan used a similar methodology to the other companies and to the academic studies. You already know where to locate the methodology and source populations for 23andme.

                              Originally posted by wombat View Post
                              ...just look at British, German, French results, for instance).
                              There is a huge difference between trying to identify the DNA of British vs German vs French as opposed to trying to find the DNA of Native Americans vs Europeans vs Africans. The former are all from the same region of Europe with a lot of shared ancestry even in the past 4,000 years. The latter had a separation of at least 14,000 years prior to the past 500 years which caused a lot more drift leading to marked differences in the DNA.

                              Originally posted by wombat View Post
                              MyOrigins does tend to map out their populations poorly though. If you look at how Geno 2.0 NG maps out groups with the same or similar names, their maps seem to make more sense, often covering vastly larger regions of different shape.
                              I have seen people being given very poor results in Geno 2.0 NG also. It caused one person to wonder if her documented ancestry was correct but it's Geno 2.0 that was incorrect by a long shot. For a good baseline of how often Geno 2.0 NG has incorrect results and how far off they are I would need hundreds of examples and I am sure neither of us have those but I do have hundreds of samples of 23andme and myOrigins samples because so many people have tested with each company and shared their results and many of them also shared their documented ancestry.

                              Comment

                              Working...
                              X