Page 6 of 7 FirstFirst ... 4567 LastLast
Results 51 to 60 of 67

Thread: Ancestry DNA Update of Ethnicity Estimate

  1. #51
    Registered Users
    Posts
    21
    Sex
    Location
    USA, Middle Tennessee
    Ethnicity
    Northern European
    Nationality
    American

    Scotland England Wales Germany Sweden Ireland
    Quote Originally Posted by Saetro View Post
    That year I studied something other than statistics.
    I think it's great that Ancestry gave us this detail, but I need to get a friend to interpret it.
    The best most of us can do is black box stuff - measure outputs against inputs.

    And that is what this forum has been doing.
    So far, we have established that:
    Scottish has been separated from Irish - but there may be further tweaking to come.
    Scandinavian countries have had an attempt at refinement, but can still give some inaccurate assignments.
    Some Native American are looking a little better.

    Any more? (On the positive side, even if qualified.)
    I think Italy estimates could be better.. My wife's estimate appears to be more refined. Her mother was from Calabria Italy and father was part German, Scot and England. Since 2018 Ancestry has had Calabria correct but her new estimate adds Italy North, Cyprus, Greece Albania, Norway and Scotland. I haven't looked to see if her results are the same as the hack.

    Italy South, "Calabria" 42%
    England 26%
    Scotland 11%
    Germany 9%
    Norway 4%
    Greece Albania 3%
    Cyprus 3%
    Italy North 2%

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

     JMcB (09-18-2020),  msmarjoribanks (09-18-2020),  Robert1 (09-18-2020),  Saetro (09-19-2020)

  3. #52
    Registered Users
    Posts
    289
    Sex
    Location
    Wolverhampton, England
    Ethnicity
    English, Welsh
    Nationality
    British
    Y-DNA (P)
    R-DF98
    mtDNA (M)
    X2b4

    United Kingdom England Wales Devon Cornwall
    Quote Originally Posted by sktibo View Post
    Just to chain off of you here Robert, FWIW......

    It's impossible in my case that my Scotland percentage does not include my English ancestry, and in case it's relevant, I don't have any English ancestry from the parts of northern England which are highlighted by the Scotland region on Ancestry's map.
    If you don't mind me asking, where is your English great grandparents ancestry located?

  4. #53
    Silver Class Member
    Posts
    4,344
    Sex
    Location
    British Columbia
    Ethnicity
    Canadian Prairies Mix
    Nationality
    Canadian
    Y-DNA (P)
    R-Z198 (DF27)
    mtDNA (M)
    T2B-T152C

    Canada England Scotland Germany Poland France
    Quote Originally Posted by jadegreg View Post
    If you don't mind me asking, where is your English great grandparents ancestry located?
    Don't mind at all. Everyone loves to talk about themselves and I'm certainly no exception.

    One of my favorite parts of my genealogy is actually one of these English grandmothers of my father. She was born in Cardiff, and grew up speaking Welsh, and culturally, she was Welsh, but the interesting thing was, all of her ancestry was English. Her father moved to Wales, and so she was born there. Her father's ancestors were from Longbridge Deverill, Wiltshire, Tetbury, Gloucestershire, the Channel Islands, but only for one generation, and I believe it was Lincolnshire before that, and then I can't trace it with any confidence before Lincolnshire as I believe that was a location these ancestors may have migrated to for work from elsewhere in England. Some hints suggest Nottinghamshire, but, no hard paper records at this point. Her mother's ancestors were from Bristol, and before that, Winterbourne (which is now part of Bristol IIRC? a suburb?) Somerset, and Dorset. Aside from the Lincolnshire line, all from southern and south-western parts of England. The Lincolnshire line accounts for 25% of this great-grandmother's ancestry. I've always actively searched for Welsh DNA connections as this grandmother of my father did speak Welsh as a first language (According to my father, my aunts, and my cousins who are older than I am) however there are no Welsh DNA connections, my dad gets 1% Wales now and one of his sisters gets 2%, neither being enough to actually indicate real Welsh ancestry, and all of the surnames from this line are English.

    The other English grandmother of my father has a much more straightforward story, genealogically speaking, being born in Derbyshire, and all of her ancestors being born in Nottinghamshire, Derbyshire, and I believe a few in Staffordshire too - all the same general area I believe. My aunt, who is the primary genealogist in my family claims this is the best researched line in our entire family, having some records for this one that go back to the start of the 1600's. We have living relatives in this area today.

    Certainly there must be some solid Anglo-Saxon heritage in there!
    Last edited by sktibo; 09-18-2020 at 12:16 AM.
    Paper trail ancestry to the best of my knowledge:
    English (possibly containing some Welsh ancestry) 31.25%, Scottish 17.96%, Scotch-Irish 12.5%, Eastern German 12.5%, Eastern European (Likely Polish possibly including Romanian) 12.5%, French 7.81%, Native American (Saulteaux and Assiniboine) 2.34%, and Colonial American, 3.125%, which cannot be traced with certainty. With certainty, there is Dutch (at least 1.36%) and some English.

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

     jadegreg (09-18-2020),  JMcB (09-18-2020),  msmarjoribanks (09-18-2020),  Robert1 (09-18-2020),  Saetro (09-19-2020)

  6. #54
    Registered Users
    Posts
    289
    Sex
    Location
    Wolverhampton, England
    Ethnicity
    English, Welsh
    Nationality
    British
    Y-DNA (P)
    R-DF98
    mtDNA (M)
    X2b4

    United Kingdom England Wales Devon Cornwall
    Thank you for sharing! Quite right, not even an inkling of Scottish/North England ancestry that far South on those lines. Anglo-Saxon and probably quite a bit of Danish Viking ancestry in the East Midlands lines.

    Looking at the White Paper again, and we do have an ethnicity estimate map for Scottish Ancestry. It appears that Southern England does pull anywhere between 0-25% Scotland Ancestry as a baseline. Maybe, you have a 'high' baseline of the Scottish Ancestry component, or more probably, it's an artefact of the algorithm.....

    Screenshot_2020-09-18 Ethnicity2020_white paper pdf.pngScreenshot_2020-09-18 Ethnicity2020_2white paper pdf.png
    Last edited by jadegreg; 09-18-2020 at 09:57 AM.

  7. The Following 5 Users Say Thank You to jadegreg For This Useful Post:

     JMcB (09-18-2020),  msmarjoribanks (09-18-2020),  Robert1 (09-18-2020),  Saetro (09-19-2020),  sktibo (09-18-2020)

  8. #55
    Registered Users
    Posts
    289
    Sex
    Location
    Wolverhampton, England
    Ethnicity
    English, Welsh
    Nationality
    British
    Y-DNA (P)
    R-DF98
    mtDNA (M)
    X2b4

    United Kingdom England Wales Devon Cornwall
    I think a lot of the issues we.re observing with missed ancestry really is a result of the ethnicity estimation method, beyond that of issues of recombination. This is how I've understood the basic principles, so please correct me, if I've got the wrong end of the stick

    One principal key to the model is that the genome is separated into 1001 discrete window for ethnicity estimation, and these windows vary in size anywhere from 3cM to 10cM, which to me seems rather large. Each window is assigned an ethnicity by comparing against haplotypes of the reference populations. Each window will receive 2 ethnicties, 1 for your mother and 1 for your father. Your overall Ethnicity estimates percentages are the sum and average of the results obtained across all these windows. Seems reasonable so far......

    Secondly, the result from a previous window, biases the following window in terms of selecting an ethnicity, such that if you got ENWE in the first window on a chromosome, then the following window is, potentially, more likely to be selected as EWNE from closely related populations ( i.e. Germanic Europe, Scotland, Wales), and ethnicity will only change if distinctly different from the ENWE reference haplotypes from the reference population. To me this is where you get missed/dropped ethnicities. Imagine the above scenario. In that second window 50% of the haplotypes resemble ENWE and 50% Germanic Europe. Because of the prior window, this second window will still get assigned to ENWE, at the loss of x cM Germanic Europe. Now let's consider a third adjacent window, Again it is comprised of 40% Germanic Europe, which represents a continuation of the prior haplotype, and 60% Scotland. This third window will then change to a Scotland ethnicity, again abandoning the prior Germanic Europe Ancestry. So In this scenario, we have lost large quantities of Germanic Europe, which has been shunted into ENWE and Scotland respectively. Admittedly these are extreme scenarios and may seldom occur, but the principles of operation hold (if I've understood correctly!). Here 23 & me, would probably assign the window to Broadly NW Europe.

    Thirdly, Ancestry argues that given the nature of recombination, that there's unlikely to be a recombination event for both maternal and paternal haplotypes in the same window, therefore ONLY one ethnicity is allowed to change within a window (though I would argue given the window size, and that many individuals, particularly of colonial American descent are heavily admixed, a population change is likely in both windows....). Here if paternal haplotype, are let's say, 39% Sweden and 61% Korea, and maternal haplotypes are 40% ENWE and 60% Germanic Europe, and the prior windows are SWEDEN and ENWE for paternal and maternal respectively, this would lead to another 60% of a window of Germanic Europe lost and further inflation of ENWE.

    Now, I've probably oversimplified, but these are the general principles as I've understood them. So here we can see how we may end up with some inflations and losses in ethnicity estimates; and as these ethnicities go further back into your ancestries, and are more heavily admixed with other dominant ancestries, the more likely they are to be lost and not recorded in your Ethnicity Estimate.

    Now, what is not clear, is whether on each run through your genome, different sized windows are used over the the same portions of your chromosome i.e. First run, the first portion of your chromosome receives a 10cm window, and then on second run gets a 3cM and a 7cM window covering the same positions. If these windows are spatially 'jittered', as in the prior example, then this will diminish these losses and/or gains.....but I'm not sure by how much.....

    EDIT:- And of course there's the phasing issues as well......
    Last edited by jadegreg; 09-19-2020 at 08:53 AM.

  9. The Following 5 Users Say Thank You to jadegreg For This Useful Post:

     Dts (09-18-2020),  JMcB (09-18-2020),  pmokeefe (10-01-2020),  Robert1 (09-18-2020),  Saetro (09-19-2020)

  10. #56
    Registered Users
    Posts
    148
    Sex
    Y-DNA (P)
    BY198
    mtDNA (M)
    T1a1

    Ireland Scotland Flanders Basque Georgia
    I am quite chuffed! As expected my Irish figure is in the 90% range. My paternal uncle has less and more Scots (mystery, the Scots seemed to be on my maternal side). Flags adjusted accordingly.

  11. The Following 4 Users Say Thank You to GogMagog For This Useful Post:

     Bog Body (09-18-2020),  jadegreg (09-18-2020),  JMcB (09-18-2020),  Robert1 (09-18-2020)

  12. #57
    Gold Class Member
    Posts
    13,126
    Sex
    Location
    Virginia, USA
    Ethnicity
    British and Irish
    Nationality
    USA
    Y-DNA (P)
    R1b-DF41>FGC36981
    mtDNA (M)
    U5a2c3a
    Y-DNA (M)
    R1b-M269
    mtDNA (P)
    K1a1a

    Wales Ireland Scotland France Bretagne England Switzerland
    Here's what I now have, including the ranges:

    Scotland 66% (50-69%)
    Wales 9% (0-23%)
    England & Northwest Europe 9% (0-14%)
    Norway 6% (0-10%)
    Germanic Europe 5% (0-17%)
    Ireland 3% (0-10%)
    Finland 2% (0-3%)

    So, whatever Ancestry's Scotland category is, it's a biggy for me, at least for now.

  13. The Following 7 Users Say Thank You to rms2 For This Useful Post:

     Dts (09-18-2020),  greerpalmer (09-21-2020),  jadegreg (09-18-2020),  JMcB (09-18-2020),  msmarjoribanks (09-18-2020),  Robert1 (09-18-2020),  Saetro (09-19-2020)

  14. #58
    Registered Users
    Posts
    1,896
    Ethnicity
    Northern Ireland
    Nationality
    Northern Irish
    Y-DNA (P)
    R1b-L193(BY2634)

    Northern Ireland Ireland Scotland
    My mother has just got her results! She has Welsh and some Scottish, so she must have some Planter ancestry in the 1700ís!

    I suspected that she had some because my Scottish percentage is very high for having just one US grandfather!

    Unfortunately she has hardly any Genetic Communities only Ulster and Donegal (one less than myself actually, no Central Donegal)

    I am surprised at the Welsh! I have no idea where that came from?

    8437B07E-7204-43A7-890C-F5FAA0993255.png
    Last edited by Nqp15hhu; 09-18-2020 at 02:01 PM.

  15. The Following 6 Users Say Thank You to Nqp15hhu For This Useful Post:

     Dts (09-18-2020),  jadegreg (09-18-2020),  Jessie (09-21-2020),  JMcB (09-18-2020),  rms2 (09-18-2020),  Robert1 (09-18-2020)

  16. #59
    Silver Class Member
    Posts
    4,344
    Sex
    Location
    British Columbia
    Ethnicity
    Canadian Prairies Mix
    Nationality
    Canadian
    Y-DNA (P)
    R-Z198 (DF27)
    mtDNA (M)
    T2B-T152C

    Canada England Scotland Germany Poland France
    Quote Originally Posted by jadegreg View Post
    Thank you for sharing! Quite right, not even an inkling of Scottish/North England ancestry that far South on those lines. Anglo-Saxon and probably quite a bit of Danish Viking ancestry in the East Midlands lines.

    Looking at the White Paper again, and we do have an ethnicity estimate map for Scottish Ancestry. It appears that Southern England does pull anywhere between 0-25% Scotland Ancestry as a baseline. Maybe, you have a 'high' baseline of the Scottish Ancestry component, or more probably, it's an artefact of the algorithm.....

    Screenshot_2020-09-18 Ethnicity2020_white paper pdf.pngScreenshot_2020-09-18 Ethnicity2020_2white paper pdf.png
    Going back and looking at the PCA and looking at how many Scottish references share the space on the PCA with the English i just don't think it's surprising. I think similarly to the previous ancestry update we'll see this change again, as there's been so much attention to how exaggerated the Scottish percentages are for NW Europeans that they can't possibly not be aware of it. I think a lot of folks are going to want to believe the current estimate hints at something real, like an increased inheritance from one's Scottish ancestors, but that clearly is not realistic with the current estimate.
    Last edited by sktibo; 09-18-2020 at 05:04 PM.
    Paper trail ancestry to the best of my knowledge:
    English (possibly containing some Welsh ancestry) 31.25%, Scottish 17.96%, Scotch-Irish 12.5%, Eastern German 12.5%, Eastern European (Likely Polish possibly including Romanian) 12.5%, French 7.81%, Native American (Saulteaux and Assiniboine) 2.34%, and Colonial American, 3.125%, which cannot be traced with certainty. With certainty, there is Dutch (at least 1.36%) and some English.

  17. The Following 6 Users Say Thank You to sktibo For This Useful Post:

     FionnSneachta (09-18-2020),  jadegreg (09-18-2020),  Jessie (09-21-2020),  JMcB (09-18-2020),  msmarjoribanks (09-19-2020),  Robert1 (09-18-2020)

  18. #60
    Registered Users
    Posts
    1,425
    Sex
    Location
    Pittsburgh, PA, USA
    Ethnicity
    Ger.-Brit.-Catalan-more
    Nationality
    (U.S.) American
    Y-DNA (P)
    R-YP619*
    mtDNA (M)
    H1bg

    United Kingdom Germany Bayern Catalonia France Ireland Switzerland
    Quote Originally Posted by jadegreg View Post
    I think a lot of the issues we.re observing with missed ancestry really is a result of the ethnicity estimation method, beyond that of issues of recombination. This is how I've understood the basic principles, so please correct me, if I've got the wrong end of the stick

    One principal key to the model is that the genome is separated into 1001 discrete window for ethnicity estimation, and these windows vary in size anywhere from 3cM to 10cM, which to me seems rather large. Each window is assigned an ethnicity by comparing against haplotypes of the reference populations. Each window will receive 2 ethnicties, 1 for your mother and 1 for your father. Your overall Ethnicity estimates percentages are the sum and average of the results obtained across all these windows. Seems reasonable so far......

    Secondly, the result from a previous window, biases the following window in terms of selecting an ethnicity, such that if you got ENWE in the first window on a chromosome, then the following window is, potentially, more likely to be selected as EWNE from closely related populations ( i.e. Germanic Europe, Scotland, Wales), and ethnicity will only change if distinctly different from the ENWE reference haplotypes from the reference population. To me this is where you get missed/dropped ethnicities. Imagine the above scenario. In that second window 50% of the haplotypes resemble ENWE and 50% Germanic Europe. Because of the prior window, this second window will still get assigned to ENWE, at the loss of x cM Germanic Europe. Now let's consider a third adjacent window, Again it is comprised of 40% Germanic Europe, which represents a continuation of the prior haplotype, and 60% Scotland. This third window will then change to a Scotland ethnicity, again abandoning the prior Germanic Europe Ancestry. So In this scenario, we have lost large quantities of Germanic Europe, which has been shunted into ENWE and Scotland respectively. Admittedly these are extreme scenarios and may seldom occur, but the principles of operation hold (if I've understood correctly!). Here 23 & me, would probably assign the window to Broadly NW Europe.

    Thirdly, Ancestry argues that given the nature of recombination, that there's unlikely to be a recombination event for both maternal and paternal haplotypes in the same window, therefore ONLY one ethnicity is allowed to change within a window (though I would argue given the window size, and that many individuals, particularly of colonial American descent are heavily admixed, a population change is likely in both windows....). Here if paternal haplotype, are let's say, 39% Sweden and 61% Korea, and maternal haplotypes are 40% ENWE and 60% Germanic Europe, and the prior windows are SWEDEN and ENWE for paternal and maternal respectively, this would lead to another 60% of a window of Germanic Europe and lost and further inflation of ENWE.

    Now, I've probably oversimplified, but these are the general principles as I've understood them. So here we can see how we may end up with some inflations and losses in ethnicity estimates; and as these ethnicities go further back into your ancestries, and are more heavily admixed with other dominant ancestries, the more likely they are to be lost and not recorded in your Ethnicity Estimate.

    Now, what is not clear, is whether on each run through your genome, different sized windows are used over the the same portions of your chromosome i.e. First run, the first portion of your chromosome receives a 10cm window, and then on second run gets a 3cM and a 7cM window covering the same positions. If these windows are spatially 'jittered', as in the prior example, then this will diminish these losses and/or gains.....but I'm not sure by how much.....

    EDIT:- And of course there's the phasing issues as well......
    Just from looking at the genome comparisons (at 23andMe) between my five full siblings and me, and between my siblings and each other, I don't believe Ancestry is correct in believing that it's unlikely for a crossover to occur within a distance of 3-10 cM on both copies of a chromosome. Or rather, it may be unlikely, except that the "unlikely" still will have occurred for many of their customers. After all, they have several millions.

    This is yet another reason why we need a chromosome browser. And we need to have our "ethnicity estimate" presented to us in graphic form, as 23andMe does. This makes it possible to associate the purported "origin" of a given segment with the ancestor(s) responsible for the segment.

    As an example, let's consider a segment on distal end of the q arm of chromosome 15 that Ancestry Composition has identified as "Native American". I have a 3rd cousin who shares 59 cM with me in five segments, including a segment in the same region of chromosome 15 that is labeled as "Native American".

    My cousin's "Native American" is nearly twice as much as mine -- 4.0% to my 2.1% -- including a slightly longer segment in the same region of chromosome 15 as my own. (My cousin's segment extends beyond the region of our shared DNA.)

    Now, two things are worth noting here: (1) both my 3rd cousin and I share a known Native American ancestor who could potentially be the source of this segment; and (2) this cousin is also my 3rd cousin once removed. That's because his father is my 2nd cousin once removed, and his mother is also my 3rd cousin. Both of them share this Native American ancestor.

    I also have five full siblings that I can compare in Ancestry Composition. So I can see places where we share DNA and compare Ancestry Composition's calls for each of us in those regions. Since our father is also tested, I can tell which parent contributed what DNA. For the most part, our results are consistent with each other. However, it's also possible to see where an error of some sort seems to have occurred.

    (For example, I have a few instances in which Ancestry Composition called a segment as Southern European or Spanish for a sibling, yet they did not make the same call for me -- despite our having matching DNA on the relevant strand in the same region. I believe that sometimes the call on one copy of a parental chromosome is affected by whatever is on the opposite copy. It's also unfortunate that even when multiple offspring are linked to the same parent, 23andMe doesn't take all offspring into account when calculating the parent's results.)
    Last edited by geebee; 09-18-2020 at 05:29 PM.
    Besides British-German-Catalan, ancestry includes smaller amounts of French, Irish, Swiss, Choctaw & another NA tribe, possibly Catawba. Avatar picture is: my father, his father, & his father's father; baby is my eldest brother.

    GB

  19. The Following 4 Users Say Thank You to geebee For This Useful Post:

     jadegreg (09-18-2020),  Jessie (09-21-2020),  Robert1 (09-18-2020),  Saetro (09-19-2020)

Page 6 of 7 FirstFirst ... 4567 LastLast

Similar Threads

  1. Replies: 219
    Last Post: 10-21-2020, 08:37 AM
  2. Replies: 61
    Last Post: 07-26-2019, 01:23 PM
  3. MyHeritage DNA Ethnicity Estimate update 2019?
    By shazou in forum MyHeritage
    Replies: 4
    Last Post: 06-03-2019, 03:16 PM
  4. Is the Ethnicity Estimate getting an update? 2018
    By laulei in forum AncestryDNA
    Replies: 2681
    Last Post: 05-20-2019, 04:26 PM
  5. Ethnicity estimate for Kashubian ancestry
    By leonardo in forum AncestryDNA
    Replies: 3
    Last Post: 06-13-2018, 07:01 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
  •