What does it mean for two DNA half-matches to not match each other?

Question

I'm looking at my gedmatch.com kit A951428 and found close matches at the same overlapping location of chromosome 3:

Kit     Chr Start Pos   End Pos     cM      SNPs
F236166 3   25,435,600  38,396,796  14.3    3,053
F392491 3   30,260,499  42,524,347  11.9    2,767
F381005 3   30,531,357  42,392,580  11.4    2,643

Gedmatch says that F236166 is a half-match (yellow) in this region and that the F392491 and F381005 (who are closely related to each other) are also half-matches with a few scattered full matches (blue).

I compared F236166 and F392491 and got no overlap, so I assume that means they half match my kit on opposite sides of the DNA. Does that suggest that the two kits with no overlap are completely unrelated and thus will not help refine the relationship between my kit and F392491?

I'm interested in tracking down the unknown relation that is suggested to be a 4th level cousin between my kit and F392491 which only match at this location in chromosome 3, hence, I found overlapping matches at that location and only found this F236166 kit for an overlapping position.

How might I use these DNA tools and data to get additional data points to help triangulate exactly how we are related? I believe the relation is through the Austria region and have no paper records for my connections there, hence, DNA is my best hope at this point.

Answer 1

It means that you do not have a triangulation group. In this case, F236166 is a match to A951428 on chromosome 3 between 25M and 38M and close relations F392491 and F381005 match to A951428 on chromosome 3 between 30M and 42M, but F236166 does not match F392491 and F381005 on chromosome 3. Therefore F236166 is matching the DNA that came from one parent of A951428 while F392491 and F381005 are matching the DNA that came from the other parent of A951428.

This can occur because the autosomal DNA technology discovers two values at every position along each chromosome, one contributed by each parent, so you can match two people along the same segment but have those two people not match each other along that same segment because they are unrelated to each other, each matching the other half of values at each location.

In this case, you need to find additional matching kits to search for a true triangulated match between 3 people who all match each other along the same segment and thus most likely share a common ancestor.

If you have at least one parent tested and can phase your autosomal results (such as by the tool GEDmatch.com offers), then this issue goes away since you now only have one value at each location and if you were to match two people along the same segment, they would necessarily match each other.