Using "shared DNA" column in 23andme for triangulation?

Question

On the 23andme site when you compare dna with a relative there is a column labeled "Shared DNA" and the description on their site says

Shared DNA means that you and your two genetic relatives share a portion of the same DNA segment. By clicking on the “Yes” links below you will be able to view where that overlap occurs in the DNA View of DNA Relatives tool.

several websites say that this is a true triangulation feature; for example https://dnagenealogy.tumblr.com/post/144966221253/relatives-in-common-quick-reaction

I have 2 questions:

1. If 23andme has a "yes" in this column is there a way to find out how much overlap? The viewer shows the segments and hovering above them shows the length of each segment but so far it looks like the only way to calculate how much overlap is to eyeball it.

2. Is there an efficient way to use this tool? I can see spending way too much time on this; first I need to select each relative I have, and then I need to go thru the dozens of pages that show up for each relative because of the 2 columns (one labeled "You" and one labeled with the relative I am comparing to) the column for me is in order of closeness but the one for the relative is not. So I might need to go down several pages before I discover that a 3rd cousin to me is a 2nd cousin to my relative I am comparing to. Each page click takes time to load...

It looks like I have a lot more relatives to work with on 23andme than I do on gedmatch which is why I'm trying to figure out how to do this.

Answer 1

Jim Bartlett recommends using the 23andMe's "Yes" feature to develop your Triangulation Groups.

He suggests you download all your segment matches and sort them in a spreadsheet. Then he says to work down your list starting from the first match, click on the hyperlink that takes you to your match's page, use his ICW list and mark each "Yes" in your spreadsheet.

Jim explains the whole procedure in his Triangulation at 23andMe article.

I'd highly recommend anyone working with segment matches and triangulation to read all of the articles at Jim Bartlett's segmentology.org blog.

And I wouldn't worry too much about the amount of overlap. As long as 23andMe is indicating that there is overlap and that it is triangulating, then the overlap is significant enough to be able to put it into a triangulation group.

Answer 2

Jim Bartlett recommends using the 23andMe's "Yes" feature to develop your Triangulation Groups. He suggests you download all your segment matches and sort them in a spreadsheet.

As long as 23andMe is indicating that there is overlap and that it is triangulating, then the overlap is significant enough to be able to put it into a triangulation group.

Thanks all for the advice.

I found that 23andme will return relatives and shared matches in json format if you send them the right string after logging in such as this to get matches

https://you.23andme.com/tools/relatives/ajax/?offset=0&limit=2000

I took my list of matches and then shared relatives for these matches and wrote a few scripts to process them. This way I can quickly play with parameters such as how much DNA we must share, although I'm sure you can do the same in spreadsheets if you know how to use them.

Without using the "Shared DNA" as a filter I end up with a big jumble, but when I use it I get nice graphs that hopefully are showing me real relationships. Now if I can just figure out who any of these people are!

Here is one graph I made by looping over each of my 2nd cousins and displaying links to everyone who is a 3rd cousin or closer to me and with at least 0.9% shared DNA to me. Everyone here is a match to me: salmon is a sibling, green is a 2nd cousin, yellow is a 3rd cousin, pink is a 4th cousin, blue is any other relationship. These colors are used for both nodes (eg. people) and relations between them. Each drawn relationship is a triangulated relation according to the "Shared DNA" field. For example 23andme says 11 is my 2nd cousin, 18 is my 3rd cousin, the relationship between them is 3rd cousin, and on the page for cousin 18 in the section showing our relatives in common the line for cousin 11 "Shared DNA" says "Yes".

I'm guessing graphs such as this are telling me something about my family tree but I'm not sure how to interpret it. One thing I find a bit puzzling is I can have a pair of my 2nd cousins such as (2, 17) that triangulate but their mutual relationship is only 4th cousin. I can't draw an example family tree that has that relationship so I wonder if some of the relationships are over/under predicted. Or am I mis-understanding what triangulation means?