# How can overlapping ~35 cM segments of matching kits not match each other?

How can overlapping 36.8937 cM (from 11,551,254 to 48,458,402) and 82.9541 cM (from 12,460,601 to 74,135,725) matches on Chromosome 18 not match each other?

On GEDmatch, I compare Alice to Bob and get 36.8937 cM and then compare Alice to Charlie and get 82.9541 cM, but comparing Bob to Charlie gets zero matching. Alice, Bob, and Charlie's paper trail show they all meet up at the same set of parents for their MRCAs.

I understand that because these kits are not phased and two values are at every position, you can get false positives, but given the very long length of 12,460,601 to 48,458,402 shouldn't that be near zero probability to be a false positive?

Does this suggest some kind of corruption in the DNA test data or blank spaces in the data? What would explain such a long false positive?

This can happen even with full siblings of the same parents. Each parent has a chromosome 18 from each of their respective parents (the grandparents). Let's call the father's pair 18pp (paternal-paternal), and 18pm (paternal maternal), and the mother's 18mp (maternal-paternal) and 18mm (maternal-maternal).

Now imagine three children who inherit from each parent as follows:

child 1: 18pp 18mp
child 2: 18pp 18mm
child 3: 18pm 18mm

In this case child 1 & 2 will show a half-IBD match on 18pp, child 2 & 3 will show a half-IBD match on 18mm, but child 1 & 3 will show no match at all on this chromosome.

Recombination effects complicate the picture, since it's possible that each child actually got some combination of the two ch.18 half-pairs from a given parent (or both parents), but the same analysis still applies at the segment level.

The simplest and most likely cause of the lack of a match is that Alice is related to Bob and Charlie through different parents. If Alice inherited the 36 cM of DNA she shares with Bob from, say, her father, and Alice inherited the 82 cM she shares with Charlie from her mother, then there would be no reason to expect any matching between Bob and Charlie. Bob would be a paternal cousin of some degree, and Charlie would be a maternal cousin, and no detectable connection between them would be likely.

• Except that the paper trail says they are related through the same MRCAs. May 29, 2017 at 23:03
• Have Alice, Bob, and Charlie run the tools to show how their own parents are related to each other? How far back are the MRCAs? May 30, 2017 at 15:10