How to add spoiler tags, edit posts, add images etc. How to - a user's guide to the new version of Boards
Mods please check the Moderators Group for an important update on Mod tools. If you do not have access to the group, please PM Niamh. Thanks!

DNA Analysis



  • pinkypinky wrote: »
    I've just seen this for the first time today.

    You need to understand that DNA shared can be a range of relationships and it's suggesting the most likely based on their experience.
    Actually telling it what the correct relationship is should help them improve their predictions.

    Understand that its not an exact science as regards the relationship degree ,rather that it (Ancestry suggestion) seems to be very optimistic as to the closeness of the relationship .
    None of the 30 or 40 known relatives were actually closer than the DNA suggested and only about 3 or 4 were correct .The rest were 1 to 2 steps further away than the algorithm suggested .
    Agree it should help to make the relationship predictor more accurate in time which is useful with many DNA matches especially those with no attached tree .

  • I have this new feature too and don't understand the point of it. The most valuable part of Ancestry's DNA for me has been the 'Shared Matches' feature. I have a large tree and can usually tell very quickly from shared matches in what line of my ancestry the match belongs if they are 4th cousin and closer. That also applies to quite a few more distant matches, especially if they have a decent tree. I can't think what this new feature brings to the table. Am I missing something?

  • VirginiaB wrote: »
    I have this new feature too and don't understand the point of it. The most valuable part of Ancestry's DNA for me has been the 'Shared Matches' feature. I have a large tree and can usually tell very quickly from shared matches in what line of my ancestry the match belongs if they are 4th cousin and closer. That also applies to quite a few more distant matches, especially if they have a decent tree. I can't think what this new feature brings to the table. Am I missing something?

    It might make the predicted/suggested relationship more accurate perhaps .if enough people fill in their definite matches then it should help somewhat I feel .

    I find Ancestry (and other sites) seem to indicate a relationship that is usually one or two steps closer than the reality .
    Then again that may be just me.Both my parents ancestors came from within 4 or 5 miles of where each was born (different counties )back as far as I can find records (c.1800) so that might skew things a little although would imagine most Irish people would have a similar experience .

    The shared match thing is very useful. Have created 25 (?) groups and put people into them according to shared matches.Most usually are only members of 1 group or 2 at most although a few I have in 3 or 4 groups .
    One group in particular of 200 plus matches has only one person who is also a member of a different group .The closest 30 or 40 to me in DNA terms in this group seem to have 20 plus shared matches with me .
    Problem is that I can still find no connection whatsoever .

    What would be useful is something like My Heritage has where you can see how close each shared match is to the person you share the match with and how much DNA they share with each other .
    Find Ancestry hints etc much better than other sites but still a lot of rubbish appears from time to time .

  • It has been a help to me by using triangulation; in effect it ‘tidies up’ connections that can be fitted into paternal or maternal lines which in themselves can lead to clues. To date I’ve had much more data than any of my matches, so it has not broken down any walls. BUT how exactly Ancestry can refine their relationship calculator to a more accurate level is not very clear to me. If for example. you take a match sharing 136 cM (2%) of DNA , that person can be any one of the following
    2nd cousin 1x removed
    Half second cousin
    First cousin 3 times removed
    Half first cousin twice removed
    There are further possibilities, but the above are the most likely, and the raw data of about 136 cMs always will open up these same relationship possibilities.
    Using ‘big data’ perhaps Ancestry could suggest potential relationships from a statistical model built on the number of returns by those who state exactly the relationships with 136cM individuals. However, it seems to me that the result would be the same because a large sample would show similar numbers of the relationships thus making the result meaningless. My guess is that the exercise could be more useful to develop and refine their ‘Hints’ tool.

  • I have a great-great grandmother, Elizabeth Towers, born c1819 in Eglish RC parish in Co Kings/Offaly. She married in 1841, same parish. On Ancestry DNA, I have quite a few matches with descendants of a Sarah Towers (birth name) who is given an approximate birth date of c1813, first official record is her marriage in 1838 in the nearby RC parish of Balliboy and Killoughy, Co Kings/Offaly .

    I am surprised at the number of matches I have to Sarah Towers' descendants who mostly seem to be in Ireland, many still in Offaly. Is it a stretch to conclude that Elizabeth and Sarah were sisters because of all these matches after this number of generations? Or could they just have been cousins?

  • Advertisement

  • I've lots of matches related to me at the level of 3x great-grandparents. How many centimorgans do you share? Have you looked at the Shared centimorgan tool for ideas?

    Genealogy Forum Mod

  • The amount of centimorgans varies of course. The most seems to be just under 40. As you know, Ancestry calls that 4th-6th cousins.

    I look at shared matches constantly. It's very valuable. Not sure if that's what you mean by the shared centimorgan tool?

  • Hmm, I don't see a way to edit my post. I should have said I have many shared matches with Sarah Towers' descendants among my known 1st and 2nd, cousins here in the US. They are all descendants of Elizabeth Towers, like me. The matches definitely descended from Sarah are further out--4th cousins and beyond.

    I also have matches with known siblings of Elizabeth Towers who emigrated to England after the evictions of the 1840s and 1850s.

  • To edit a post, click on the three dots in the top right corner.

    Shared Centimorgan tool:

    It's based on crowdsourced data to give actual ranges for different relationships. It's much more accurate than the different databases.

    Is it possible that you have more than one link to Sarah Towers' descendants? That would strengthen the DNA share size.

    Do the known siblings of Elizabeth Towers' descendants match these possible Sarah people?

    Genealogy Forum Mod

  • Thanks very much for the info on how to edit a post and also for the shared centimorgan tool. I took a look and it will take some study for me to use.

    It is certainly possible that I might have more than one link to Sarah's descendants. I did think of that altho none of the names are recognizable to me tho that is far from conclusive. If Sarah Towers and Elizabeth Towers are sisters, they did have a lot of siblings according to an Eglish RC parish census done by the parish priest in 1840. He gave numbers and genders but only the name of the head of household, in this case Hugh Towers.

    And yes the known siblings of Elizabeth Towers' descendants, Thomas and Mary, are shared matches with Sarah Towers' descendants but they could still be cousins, not siblings, I assume. Thanks again.

  • Advertisement

  • My brother got a DNA test. We're from Co. Sligo, in fact I can trace my GG Grandparents to Co. Sligo. I'm not sure how accurate these things are but interesting all the same. I'll be shouting Mayo for Sam this weekend so.

  • What company is that test from?

    Genealogy Forum Mod

  • 23 and me

    A question for anyone who is into using these DNA companies. Is there ever a break down of your DNA going back further of the different groups of settlers like Norman, Viking, Celts, Beaker, Neolithic farmers, western hunter gathers and whatever else?

    My brothers one has this but not much detail

  • That's great detail - I used "my heritage" and they can't be anymore accurate than 40% Irish/English/Welsh, without even a breakdown between the countries, nevermind counties!!!

  • I am repeating myself again here.

    The ethnicity breakdowns are just for fun. The science is only accurate to continental level at present.

    Genealogy Forum Mod

  • They are being very broad with your paternal line, is the R1bM343 not broken down more?

  • I hope someone can help with this. I have a DNA match with many shared matches on Ancestry. I usually can figure out the relationship but this time I could not. We are a very large family, coming from Ireland to NY, and I have spent years putting it all together. I emailed the person managing the test. She said it was her husband's 94 year old mother, looking for the identity of her father. Her mother would never tell his name. They gave me access to her Ancestry page of matches and after spending some time on it, comparing with my matches and my tree, I believe I have the answer. He would have been my grandmother's first cousin, CK. The 94 year old woman has 974 cM shared with her match who is CK's grandson. Another of CK's grandsons shares 803 cMs with her. And so it goes. I am also related to all these people at a much lower number of cMs. CK had a brother but he had no known descendants. CK's sister has descendants who are matches but many less cMs than the grandsons.

    This woman has waited her whole life for an answer and I'd like to give it to her but not if I am off-base in my deduction. She also has children and grandchildren who would like to know. What do you think? Do I have enough info to say that CK is her father?

  • Hmm, tread very carefully, Virginia.

    CK is certainly a potential father but his brother could have been the father too.

    2 of CK's grandsons certainly match her at first cousin level.

    Use DNA Painter to look at the odds for the different cM matches.

    Genealogy Forum Mod

  • Thanks for that, pinkypinky. Would it be correct to say that CK's father is not identifiable with certainty from present DNA results?

  • With the caveat that I haven't done the work you have, I'd say not definitively. You've got a very good possibility.

    Genealogy Forum Mod

  • Advertisement

  • I will have to reply with that to the family. I feel so badly that I couldn't give this lady and her family a definite answer. But I can give her grandparents, great-grandparents, lots of cousins, how they are all connected and townland of origin so I hope that will be some consolation.

  • It sounds like you've done your best. Have you considered whether this potential man was in the right place at the right time to be her father as well?

    Genealogy Forum Mod

  • Yes, both CK and his brother were in the right place at the right time. I had the info in my tree already but I double-checked. NY and Brooklyn.

  • It was my brothers test so I'll have to find out if it's broken down more.

  • Looking at the latest update in ancestry,I'm curious as to how they are compiled as each site (ancestry,living dna and ftdna)gives a different result.

  • They release info on that. If you look at Debbie Kennett's twitter, she often posts about this kind of thing.

    Genealogy Forum Mod

  • Thanks for posting that which definitely makes things clearer.

  • It's broken down into R-S661. I don't really know what that means if anything

  • Advertisement

  • I have not gotten any 1st-4th cousin matches on Ancestry in at least six weeks and few in the last year or two. I have 268. My brother has 318 which seems like a large disparity. I find the matches feature very valuable in my research and am wondering about the lack of them for quite awhile and the disparity between my brother and me. The irony is that he doesn't even do any genealogy so never looks at them.

    Is anyone else experiencing a lack of matches and/or a disparity in number between yourself and a full sibling? Is this a normal statistical variation?

    Post edited by VirginiaB on