I have been mainly working with my Dad’s Norwegian DNA at 23andme and at Family Tree DNA. Often he will have a match at one company and there will be a match to someone else on the same segment location at the other company. So how to tell if they match each other? Since one could have the DNA segment that Dad got from his mother and the other could have the segment Dad got from his father, the only way to be sure it is the same segment is if they also match each other on that segment. This is what is known as triangulation.
If they have both uploaded to GEDmatch, I can compare their two kits there and see if they match on that segment. Often however one or the other has not uploaded or the GEDmatch site is down. So I needed another way to figure this out.
It occurred to me that I could check if the new match also matches me there, since my results are at both web sites as well.
Obviously when they both match me on that DNA segment, I know they match each other. If one matches me and the other does not, then I know they are not a match. But what if they both do not match me? Then I must have inherited that segment from one of Dad’s parents and they are matching the DNA piece from his other parent.
In the case above, Dad has a 23cM match at Family Tree DNA with an adoptee at the same spot where he has many smaller matches over at 23andme. So do those folk match DM?
So to figure this out I log into my account over at Family Tree DNA and select in common with (ICW) my father on the Family Finder matches page by clicking just under his profile so that more choices appear including the ICW option:
Next I put the surname of the match I am interested into the name box by clicking on Name: in the gray bar so that the blue box appears for me to fill in. It was a very common surname, so there would have been too many pages of names to look through if I just did the surname box without the ICW selection.
And the result is that DM is not in my family tree DNA match list.
Over at 23andme there are many people who match Dad at places on this spot in chromosome 3. Only two of them match me, so those two are not a match to DM, but the others are. The ones that do not match me match my brother and first cousin so we know they are real matches. Here is what that section of my spreadsheet of my matches to Dad looks like. As you can see, I use colors to show the triangulations when people match each other (the bottom group are all one family):
Thus when using my match results to triangulate between the companies, if the groups or individuals I am trying to compare do not match me at either company then they are a match to each other as long as they are a true match to Dad. I consider the match good if the segment is 10cM or greater. Smaller than that it is not for sure, but often matches to other cousins or relatives will confirm that the match is a real one, so IBD rather than IBS (see the article at http://dna-footprints.com/203/the-abcs-of-dna-ibd-vs-ibs/ for good explanations of IBS and IBD)
This same technique can be used to triangulate between matches at Family Tree DNA where you cannot compare your matches directly to each other. I also have written an automated tool for working with match data from family tree DNA called Kworks for the DNAadoption site; it will make a spreadsheet of matches with an ICW grid filled in, but that is a blog post for another day.
A reader has cleverly pointed out that I cannot be absolutely sure that if person A matches me and Dad on a specific segment at 23andme and person B matches us at the same location over at family tree DNA that they match each other. It could be that one of them matches me on the segment I got from my mother and matches Dad on the segment he did not pass to me.
True but since my mother and father are from very different population groups, this is a very unlikely scenario. The other match would need similar ancestral populations to both my parents plus my parents would need to match on that spot. A very long shot but not impossible.
Those of you who have tested both your parents can rule out this possibility if your parents do not match each other at all. Another way to rule this out is to run the utility at GEDmatch that checks if your parents are related. Mine were not so I am confident in this approach for my data.
Neither of my parents have tested. Even worse, my parents are from the same small-town population group, so things can get very murky. However, I’ve tested myself and four of my siblings. This has allowed me to neatly sort overlapping matches into two distinct groups, one for each of our parents, even when the matches come in from three different companies. (This works perfectly even where we don’t have any “Rosetta Stone matches” who have tested at more than one company)
Example:
We share a large group of matches from Chromosome 19 (8000000-15000000, each segment about 15 cM in length).
About half of the matches match siblings A, B, C and E while the other half match only A, B and E.
Ultimately, I found that the first group clusters geographically in NC and triangulates (four times over) to a Mullis line from my mother’s side. The second group clusters geographically in TX and LA.
There are numerous examples like this across my genome.
With color coding this method is so easy and powerful, directly checking to see which of these matches match each other seems almost superfluous. Almost.
Of course, this method conceivably could be complicated by segments that fall in locations that are full-identical regions for all five of us, but so far I haven’t found any segments that exemplify that unlikely pattern.
well done jason, four siblings must be a big help! I find cousins are quite useful as well for sorting out which line … so get a few cousins to test
Thank you. Busily recruiting!
I’m interested to know how to go about isolating my father’s DNA contribution, which I understand is called phasing. I have my mother and myself already tested at 23andme, and my sister’s results are 4/6 processed. Both sets have been imported into Gedmatch, and mine only into Family finder at FTDNA.
I was advised that using a Null or dummy entry might help, in certain phasing calculators / tools so I have extracted, matched and eliminated common bases chromosome by chromosome until I have a genome null / dummy that is filled with no calls in every position, which my mother and I have been tested at. What I don’t understand is the process involved to extract the portion which represents only my father’s contribution and his ancestors. Assuming that a reconstructed genome can be manufactured by this method, by importing the phased results chromosome by chromosome, into the null / dummy template, then copying the format of the text file down loaded from 23andme. Could I then use this to triangulate with AND at Gedmatch, initially to confirm that when compared to myself and my sister it produces Gen 1 results, and then to search for matches in general. My issue is that I’m trying to isolate only my Irish direct male line matches, which are a minority as my great grand father is the only Irish born direct male, in the four previous generations. My previous effort of filtering matches, by haplogroup and male only matches, was a limited success as most people don’t know or fill in the data with sufficient accuracy i.e. R1b M269 is no good when your looking for R1b > M269 > L21 > Z253 > L1066 . CTS9881 or R1b CTS9881 for short. I’m an admin at Z253 Y dna and am also exploring any common heritage from the nine only CTS9881 people, via Y markers , autosomal and documented family history, which is a parallel activity to searching my direct line matches, which could end up finding a TMRCA, and common ancestry.
If I understood more about phasing, and then the use of triangulation that would be of great use in this exercise.
BR MikeB
I am Cts9881. Contact me: Jasondrawhorn@gmail.com
Mike – GEDmatch has a phasing utility when it is back up to speed. Also this tool will phase for you:
http://www.y-str.org/2013/07/phasing-utility.html
But if your purpose is to trace down your paternal line and Y has not suceeded, get a few paternal side 1st and 2nd cousins to test.
We used Y for our paternal line brick wall but the road to our success involved developing a hypothesis and then testing a direct line male descendant of that possible ancestor see
http://blog.kittycooper.com/2013/03/we-have-found-our-ancestor-lars-monsen/
DNA testing is most useful when combined with standard genealogical research, else you are dependent on being lucky enough to have distant cousins on the desired line who have tested.
So my suggestion is to find the forum or mailing list where people from the part of Ireland that your ancestor is presumed to be from hang out and ask for their help on the paper trail. Then use DNA to,prove or disprove your theories.
The follow up to,our Lars Monsen story with DNA is here:
http://blog.kittycooper.com/2013/05/its-a-match-lars-monsens-ancestors-are-found/
Can a triangulation be proved when a cousin, me and my brother all share common DNA on a chromosome and segments with each other?
Possibly. If you and your brother are fully identical on that segment (have the same DNA from each parent) then NO because the match could be false, see http://blog.kittycooper.com/2014/10/when-is-a-dna-segment-match-a-real-match-ibd-or-ibs-or-ibc/
However if you are not fully identical and the segment is large enough then yes.
GEDmatch and 23andme both show fully identical segments