Announcement

Collapse
No announcement yet.

New My Origins Results

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

  • Banquo
    replied
    My Origins needs to be fixed

    Hopefully FTDNA can see this function needs to be fixed, as the current version is wrong. My mother and uncle are in the database as well, and these percentages make no sense in our small cluster. I note the recurring theme of thousands of testers now getting an "Iberian" result, where none existed before.

    I possess the Scots Modal YDNA and come from a well documented line stretching several centuries. Even assuming NPE in a the non-YDNA lines of my autosomal heritage, it would still not compute to my British Isles background being truncated by more than half, my 1/3 Scandinavian heritage being all but wiped out and a heretofore non-existent Iberian heritage rising up to make 13% of my background.

    Leave a comment:


  • dave21286
    replied
    Did My Origins have another update recently? I, like just about everyone, got new results a few months ago. I decided to check the other day, and the results changed again! In two months, I went from 15% Scandinavian to basically zero. My Western Europe went from a small amount to about 20%. I'm now showing zero British Isles when it was at about 20%(Ancestry has me at 25%). Just about the only thing that all of the testing companies get right is my Italian classification. It seems to me that one of the problems for someone like me who is quite admixed is that we get an averaging effect. This is especially true for me on the GEDMATCH calculators, where my northern, central, eastern and southern European lineage comes out as Hungarian, Romanian, Bulgarian, etc. The new My Origins is also showing me as 17% Middle Eastern (Asia Minor), which I understand is quite common for someone of Italian descent (mother's paper trail is 100% Italian, with strong Norman features). We have no record of any Middle Eastern heritage, so if it's there, it must be quite old. My gut tells me that Ancestry's classifications are probably the most accurate since they are working with larger sample sizes.

    Leave a comment:


  • willy
    replied
    OK so I am not alone to find that new MyOrigins results are wrong I will say very wrong surely the people who made the algorithm is completely incompetent or the reference population is confused ? or the twice ?

    Originally posted by PaulHosse View Post
    In my case, the new MyOrigins results are wrong. Gone is my Greek ancestry (my paternal line is originally from Greece; settled in England in 1550) and my Ashkenazi ancestry (my maternal line was Belarus/Ukrainian Ashkenazi and Sephardic---oh and the Belarus ancestry is missing too). This is a bad as 23 & Me which has me as Native American and Yakut! MyOrigns does, however, have me as 7% Eastern European. I think that's part of their mistake.

    Leave a comment:


  • PaulHosse
    replied
    Disappointed Big Time!

    In my case, the new MyOrigins results are wrong. Gone is my Greek ancestry (my paternal line is originally from Greece; settled in England in 1550) and my Ashkenazi ancestry (my maternal line was Belarus/Ukrainian Ashkenazi and Sephardic---oh and the Belarus ancestry is missing too). This is a bad as 23 & Me which has me as Native American and Yakut! MyOrigns does, however, have me as 7% Eastern European. I think that's part of their mistake.

    Leave a comment:


  • John McCoy
    replied
    Human populations today tend to be diverse rather than homogeneous. There is good reason to believe that much of Europe has a history of diversity rather than homogeneity as well. The statistical derivation of a "cluster" or "reference group" that is relatively genetically homogeneous is therefore, to some extent, an artifact of the methodology.

    Cluster methodology, for example, automatically places the samples in relatively homogeneous groups, regardless of where they came from, and regardless of how much diversity was originally present in, say, "Scandinavia". It's up to the analyst to discern what degree of clustering is appropriate for the application and what the clusters finally selected actually represent. We can look at the methodology in many different ways, but the "reference groups" will always be proxies or substitutes for the real long-dead ancestral populations that we cannot sample.

    Leave a comment:


  • KATM
    replied
    Not sure if it will help, but I made a couple of posts in another thread recently regarding references populations, based upon some information I'd saved for 23andMe and FTDNA: http://forums.familytreedna.com/showthread.php?p=439950

    One has to do with the overlap of sources for reference populations (FTDNA, 23andMe, and Ancestry.com), and the other mentions quantities of people tested for some of the populations. The information in my posts may or may not be relevant, as I'm not sure what changes have been made since that information was originally posted elsewhere. Certainly we don't have information on myOrigins 2.0 populations used.

    Leave a comment:


  • georgian1950
    replied
    Originally posted by John McCoy View Post
    How homogeneous are the reference groups? This seems to be an extremely important question, for which we don't have a satisfactory answer. (This is a separate question, logically, from the problem of how "representative" the reference groups might be, another important issue.)
    For a group to be homogeneous in this day and age, it would have to be inland and extremely isolated. In my opinion only a very small fraction of global population is homogeneous.

    Jack Wyatt

    Leave a comment:


  • John McCoy
    replied
    How homogeneous are the reference groups? This seems to be an extremely important question, for which we don't have a satisfactory answer. (This is a separate question, logically, from the problem of how "representative" the reference groups might be, another important issue.)

    At least some of the methodology that has been applied to the "admixture problem" uses multivariate statistical methods such as cluster analysis, principal components analysis, etc. to arrive at "reference groups" that are defined not by their actual genealogy (their paper trail ancestry), but by their genetic similarity (and clearly, this is not the same thing as measuring an actual historical population!). Whether all of the reference groups now in use have been derived in this way, I don't know. Are these groups internally similar enough to be useful? Are they sufficiently distinct from one another? Are they large enough? How robust is the methodology? What computational shortcuts have been used, and what are the risks or biases that they introduce? There is reason to think that the full methodology should have some predictive value, but I don't think we have enough information to know how much.

    There is probably not even a precise definition of what is meant by "ethnic origin" as it applies to the present context -- another impediment to effective communication!

    Leave a comment:


  • georgian1950
    replied
    Originally posted by John McCoy View Post
    Second, ethnic origins results are produced by comparing your autosomal DNA (SNP's) with those of "reference groups" consisting of DNA samples from modern, probably still living people who have been found by statistical methods to be similar to each other and distinct from other groups. These "reference groups" are assumed to be (and logically should be) the descendants of relatively homogeneous and freely interbreeding populations that lived in the indefinite past, far enough back that the scrambling effect of recombination has blended everything together (the "melting pot" effect). The statements about your ethnic origins going back hundreds or thousands of years are apparently just a way of saying that some very long time must have elapsed since the "founding" of an "ethnic" population in order to produce a relatively homogeneous set of modern descendants. Note that the detection of a set of modern people who are similar to each other and distinct from other groups does not imply that their ancestors necessarily lived where their descendants live today -- we have to depend on history for that, and the history of "ethnic" migrations may not always be correct in detail.
    The problem with reference groups is that they all have been contaminated within the past 300 years. Until genetic genealogy figures this out and accounts for it, ethnicity estimates are just an entertainment.

    Jack Wyatt

    Leave a comment:


  • John McCoy
    replied
    These are good questions. Some partial answers are possible.

    First, the question of whether your own DNA actually reflects your ethnic ancestry: Because of recombination and the random assortment of chromosomes at each generation, it is almost certain that the percentages of DNA you have inherited from each of your 16 great-great-grandparents (for example) is not exactly 6.25%. Rather, some ancestors are under-represented and others over-represented in your DNA, and as a consequence, some of your ancestors a couple centuries back are likely not represented in your DNA at all. (This fact is related to the observation that perhaps as many as half of your fourth cousins will not be detectable as autosomal DNA matches. There are some interesting discussions on these points in the ISOGG article on "Autosomal Statistics", especially in some of the links.)

    Second, ethnic origins results are produced by comparing your autosomal DNA (SNP's) with those of "reference groups" consisting of DNA samples from modern, probably still living people who have been found by statistical methods to be similar to each other and distinct from other groups. These "reference groups" are assumed to be (and logically should be) the descendants of relatively homogeneous and freely interbreeding populations that lived in the indefinite past, far enough back that the scrambling effect of recombination has blended everything together (the "melting pot" effect). The statements about your ethnic origins going back hundreds or thousands of years are apparently just a way of saying that some very long time must have elapsed since the "founding" of an "ethnic" population in order to produce a relatively homogeneous set of modern descendants. Note that the detection of a set of modern people who are similar to each other and distinct from other groups does not imply that their ancestors necessarily lived where their descendants live today -- we have to depend on history for that, and the history of "ethnic" migrations may not always be correct in detail.

    Another aspect of the "reference groups" is that there is no logical necessity that the methodology, relying as it does on some limited number of modern samples, will actually capture all of the genetic variability that was present in real historical populations. If he "reference group" happens to be incomplete in this sense, we might find that some real "Scandinavian" ancestry, for example, is not detected. Does this ever happen? Perhaps, but how can we be sure?

    The question of how accurate "ethnic origins" results can be, and to how measure their accuracy, does not seem to have a satisfying answer. It is difficult to come up with an external reference point by which the accuracy of the results can be judged. One case that can be tested, and which has received some attention in this forum, is how often the results for a child are inconsistent with the results for the child's parents. These situations should give us an idea of how much variation is inherent in the algorithms, or, put another way, how much of the results might be considered statistical noise.

    Leave a comment:


  • JoHolt
    replied
    ethnicity changed, now boring but more realistic

    I think the whole ethnicity thing is entertaining, but I do not take it seriously. I have never understood the time frame that we are dealing with. Anyway, when I first got my ancestry results I was told I was 38% Scandinavian (range 18-59%). This was interesting but with the thousands I now have in my tree there is not one Scandinavian. Now I accept the fact that Viking invaders probably influenced my family.... but I found it hard to accept that they had this much of an impact. Then I uploaded my dna to ftdna and they told me I was 32% Scandinavian.... so well I thought... boy those Vikings really got around. Well then ftdna changed the database and I am now 8% Scandinavian. Honestly this makes much more sense. I wish I could recall the other earlier ftdna results but now I am a whopping 72% Western European... which lines up much more closely with my tree data... but how boring is that. Interestingly years ago my mitochondial results for my mother of more Belgian/Dutch ancestry found her haplogroup to be U6a1, commonly found in morocco, so the small Iberian ethnicity that appears also makes sense. But to be honest I really don't know what this data really means.... does it mean my ancestors a few hundred years ago are from these areas or 1000 years ago or 10000 years ago? Also, if your ancestors only pass on some dna... then you could lose some ethnic markers over time... does that mean you are not of this background or simply that your dna does not capture that ethnicity anymore... and if so what does any of this even mean?

    Leave a comment:


  • falkjohan
    replied
    Originally posted by falkjohan View Post
    Before I was 93 % Scandinavian and 3 % British... now I am 58 % Scandinavian and 38 % British. I can say this new logaritm is completely wrong! I'v done my family tree by church records and have 8 generations complete line back - all swedes, no British what so ever. So the thought that I should be 38 % British is just wrong! You had it right in the first place. This is not credible.
    As FTDNA results has gone bananas with Scandinavian origin in the new algoritm I did another DNA test - the Genographic Project - as a comparison. Right now they are much more accurate scoring me 92 % Scandinavian and 7 % eastern european. This is closer to the truth, though the FTDNA was more correct in the first place!
    Last edited by falkjohan; 15 May 2017, 04:37 PM.

    Leave a comment:


  • willy
    replied
    Originally posted by vinnie View Post
    Willy, may I ask what your ethnicity is, or what ethnic populations you expected for myOrigins?
    English , west en central Europe (France Germany Swiss) , Jewish according that I know about my family , MyOrigin was OK before that very bad new version

    Leave a comment:


  • vinnie
    replied
    Originally posted by willy View Post
    Hello , yes I do agree the new MyOrigin is completely wrong I have no Greek or Iberian before and now FTDNA found that ? So this is not good ! Could FTDNA made a better stuff
    Willy, may I ask what your ethnicity is, or what ethnic populations you expected for myOrigins?

    Leave a comment:


  • willy
    replied
    Hello , yes I do agree the new MyOrigin is completely wrong I have no Greek or Iberian before and now FTDNA found that ? So this is not good ! Could FTDNA made a better stuff

    Originally posted by DaveInGreece View Post
    Yes, I'm very aware that Vikings settled in Britain and Ireland and that British and Irish people ended up in Scandinavia (sometimes willingly, sometimes not). People with my ancestry expect to be assigned some "Scandinavian" - as I said in my previous posts (you bothered to read them, right?) - but only a few per cent at most. I've now got 22% "Scandinavian". Nearly a quarter. My ancestry isn't even from any of the areas of Britain and Ireland with a known history of interaction with the Vikings. My brother and sister only get a couple of per cent "Scandinavian". I used to get just a couple of per cent. Everywhere else I get just a couple of per cent. Everywhere except My Origins v2 assigns me a small amount of Scandinavian which is in line with the normal profile of someone with British and Irish ancestry. MyOrigins v2 now assigns me almost a quarter "Scandinavian" whereas many people with Swedish or Norwegian grandparents are now being assigned zero Scandinavian (they got the correct results on v1, and get the correct results everywhere else).



    You read my comments, right? You know, the comments (which you quoted) in which I mentioned Angles, Saxons, Normans and Vikings? The comments in which I mentioned that it's normal for people of British/Irish ancestry to have "quite a lot of 'West and Central Europe' (Anglo, Saxon and Norman ancestry) and a small amount of 'Scandinavian' (Viking ancestry)." Do I sound like someone who doesn't know about migrations? How the heck can you read the words Angle, Saxon, Norman and Viking in one sentence and then tell me that I need to learn about migrations?

    But since you're in a teaching mood, explain this to me:

    My 22% "Scandinavian" is minor compared to the 29% "Iberian" I now get (here and only here). What migration explains how a person with British/Irish ancestry gets assigned almost a third "Iberian"? At which point in history did "changing borders" make English people genetically "Spanish"?

    And after you rewrite history to explain that to me, try explaining the genetics. How can I have 29% "Iberian" when my brother and sister (both totally confirmed as being full siblings) don't have a trace of anything from southern Europe? I can clarify that I've used visual phasing and chromosome mapping so I know there's nothing at all weird about me inheriting more than them from one grandparent. Cousins on all sides of the family have been tested, and they show no southern European of any sort. I can send you my chromosome map if you want to work out which 29% of me matches neither of my siblings and none of my cousins (I should warn you that we're up to around 40-50 testers now, so it's not going to be quick). And don't forget to look for that 22% Scandinavian while you're at it.

    Leave a comment:

Working...
X