cross-contamination ?

[abuelita]

We have a situation which in classical genetics tests, with the tiny test tubes and microplates, would scream cross-contamination.

cast: son, mother, mother's cousin, step-father, and off-stage (dead) the biological father.

setting: geographically quite distinct locations for the 3 known families.

-The son and cousin were tested some time ago. This cousin is by far the mother's closest match after her son and his child.

-The mother and step-father apparently did their swabbing at the same time, though nobody was there to observe how they did it. . They say he did his own swabbing, and helped her do hers, then they mailed the kits simultaneously. The kits arrived at FT about a week apart. The results were released about 10 days apart - the step-father's first. If there was any cross-contamination it was at home, not in the lab.

-Searching the step-father's matches for his surname popped up only one match, his step-son. Okay - 1st hypothesis, the step-father is somehow distantly related to either his wife or the biological father of his step-son.

-The match was confirmed using the gedmatch one to one tool. Looking at the one significant shared segment only about 4-5 kits shared it: a couple of ancestry strangers, the stepson, and two tests done on the mother's cousin. Hypothesis narrowed down: step-father is related to the mother, his wife.

-Then the results for the mother arrived - surprise. Total confusion.

1. The mother matches her son and her cousin. She does not match her husband (the stepfather).

2. The Stepfather matches her son, and shares a segment with both the son and the wife's cousin, a segment she doesn't share (!!!) This segment is quite clear on both the FT and gedmatch chromosome views, it's about 10.7 cM

3. Matches in common?
mother-son: FT 993 (of 3127) gedmatch >100 (quit counting)
mother-stepfather (her husband): FT NA gedmatch 14 (at default 10/10 level)
mother-her cousin: FT 37 gedmatch 37
stepfather-son FT 0 gedmatch 17 (not including the mother)
stepfather-wife's cousin FT NA gedmatch 1 (the son)
son-mother's cousin FT 29 (of 2607) gedmatch 32

??? Any ideas? I would think the link is through the mother, but she absolutely does not show that shared segment. And it the link is through the biological father, how do we explain the cousin?

[John McCoy]

Best approach is first to tally up the total cM shared for all combinations. That will place some limits on the possible relationships. Are all the expected close relationships verified? The relationships you have described are so close that the total shared cM for each combination (the ones you expect to be in the range of parent/child to first cousin once removed) should give you a very clear answer.

Next, verify that X matches, if any, are consistent with the expected close relationships.

Additional small shared segments (such as one segment of 10 cM or so) involving combinations that you don't expect to be related could easily derive from a shared ancestor many generations back, such as a 5th cousin relationship or possibly even something more remote. It is also possible to find an occasional "shared" segment that misbehaves, such that not all combinations of the kits that initially seem to have that segment will actually match each other, suggesting that the match is more an artifact than something on which you can base a genealogical conclusion.

It is very unlikely that a kit could ever be contaminated with just one or two little segments. Rather, contamination, even a little of it, would show up as a huge match, suggesting parent/child.

[Emona]

With contamination, the result of the test would probably be: "Failed". It would have three or four values in many places. (Two are normal, for maternal and paternal chromosome. AT is fine, ATC is not right.)

Gedmatch has a tool named Phasing.

With mother and son available, it creates two new profiles: the paternal and maternal half of the son. You compare those halves to the stepfather, one to one.

Does the 10cM match still exist on one side? If so, you will know where it came from.

If the segment is real and came from bio-father, then probably the bio-father, step-father, and mother's cousin have a common ancestor sometime in history. 10cM can be old. By definition, it has a 10% possibility to be cut on its way to the next generation. Easily travels 10 generations or more unchanged.

[abuelita]

Thank you both. You've convinced me that cross-contamination is not a concern.

The most parsimonious explanation for this strange mix of matches is that the stepfather is related to the mother and cousin, and not to the bio father. That would mean that I have to postulate only one low probability link.

Two such events, being coincidentally related to both his wife and her ex-husband, is beyond improbable. We are talking about separate families from the Northeast US, deep south, far mid-west, originating in different countries in Europe. And 10 generations to get any two of them together may be minimal.

The only problem is ... that the mother doesn't show that segment. Or I thought she didn't. I went back through FT and I could see it perfectly clearly, looking at her with her cousin in the chromosome browser. This time I wrote down the numbers and switched to son/stepfather/cousin - quite close! - with just a little variation for interest.

In any case I'm going to try phasing for the fun of it - thanks for the suggestion. I've never tried half the tools in gedmatch.

details:
By trying different combinations of people in the FT chromosome browser I get (on 6)
136593779 - 146914378 mother-cousin
136166513 - 146914378 son-cousin
135113519 - 143111008 son-stepfather
gedmatch picked something from the same region in another, more distant, known cousin (same family)
136259803 - 147228930 mother-cousin #2