Jerry and Paul

At age 16, Jerry’s mother was working in a canning factory, where she met a young man. As things happen, things happened. When the season ended, she returned to her home state a little bit pregnant. She eventually married and never saw her beau again.

Fast forward 45 years, and Jerry learns the truth: the man who raised him was not his biological father. After much denial, his mother eventually confessed and gave Jerry a name: Paul Gillen (not his real surname). After searching on his own for years, Jerry took a DNA test and contacted Janet Rinardo Travis and Lynette Bryan. They are volunteer search angels through the Facebook group Search Squad and also offer professional search services through their company DNA Now.

Within a week, Janet and Lynette had linked Jerry to the Gilland family (note the different spelling). The Gillands had 10 sons, and one was named Paul. But the timing of the discovery couldn’t have been worse: Paul had passed away 3 days prior. Normally, we could test a child or grandchild of Paul to confirm that he was Jerry’s father, but he had no known children. The alternative would be to test descendants of each of the other nine brothers to show that they were not Jerry’s father. The quest seemed hopeless.

Enter providence. Paul had one living brother left, Uncle Sedric. For reasons unknown, he had delayed Paul’s cremation. Sedric gave his blessing to the funeral home to take DNA samples. Janet had the foresight to ask for both cheek swabs and hair follicle samples (not just cuttings) and delivered them personally to FamilyTreeDNA’s lab in Houston to have DNA extracted. Paul had been dead for 6 days by then, and the cheek swabs failed, but the lab was able to get useable DNA from the hair follicles. They then ran a standard autosomal analysis.

Things Get Technical

The no-call rate for the DNA analysis was 23%, meaning that only 77% of the possible data could be obtained. That’s not surprising given that the samples were taken nearly a week after Paul passed away. But was it enough to confirm whether Paul was Jerry’s biological father?

The data quality wasn’t high enough for Paul’s kit to be included in FamilyTreeDNA’s matching database, so Janet copied the raw data file to GEDmatch. There, she was able to compare Paul and Jerry directly using the “Autosomal One-to-One Comparison” tool.

The results were not what Janet expected:

A parent and child should share about 3579 cM, whereas Jerry and Paul shared 2588 cM. That’s the expected amount for full siblings, which is simply impossible given that Paul was older than Jerry’s mother. What’s more, the largest segment shared by a parent and child is about 281.4 cM (the full chromosome 1), whereas Paul and Jerry’s longest segment was only 80.4 cM. Finally, a parent and child should share 22 segments (for the 22 chromosomes), and full siblings usually share 40–50 segments. Paul and Jerry shared 125! What was going on?

Visualizing Matches

Here’s where it helps to understand a bit about DNA matching and what typical parent–child and full sibling matches look like. GEDmatch gives the option to visualize the segments that two people share, as opposed to just listing the start and stop points of each match.

Recall that we each have two copies of each autosomal chromosome (autosome). Because a parent passes on one copy of each, and because the child inherits a second copy from the other parent, a comparison between the two should show that they match on one of their two chromosomes, but not both. The image below shows the first three autosomes (of 22 total) in a normal parent–child match.

Each chromosome diagram is made up of tens of thousands of color-coded, vertical lines, each line representing one bit of DNA (base pair) that was analyzed. Yellow bars mean that the mother and child match on one of their two bases at that spot. We call the yellow regions “half identical”. Green means that they match on both copies at that spot, which will happen by random chance sometimes in a parent–child match. And red means that they don’t match on either chromosome. For a parent–child comparison, the occasional red flecks are errors and can be ignored.

Full siblings, on the other hand, get a mix of DNA from their two parents. The pattern is quite distinct.

In some spots, they are half identical to one another (yellow) because they inherited the same chunk of DNA from either mom or dad. In other places, they have large stretches of so-called “fully identical regions” (FIRs) where they match on both chromosomes, because they inherited the same chunk from mom and dad. Those are mostly solid green, as opposed to the scattered green flecks we saw above. And in yet other spots, they won’t match at all, because one sibling inherited segment copies from, say, mom’s mom and dad’s dad, while the other sibling inherited that region from mom’s dad and dad’s mom. Non-matching segments appear as red-and-yellow patches, where the red lines are mis-matches and the yellow flecks are coincidental matches.

Artifacts from Artifacts

Jerry and Paul’s match looked like this.

That’s remarkably like the parent–child example above, with one key difference: there are a lot more red flecks. Because Paul’s DNA had started to degrade by the time it was sampled and extracted, and possibly because there isn’t as much DNA in a hair follicle sample as in a living cheek swab sample, there are a fair number of errors in the data. That’s to be expected.

Those errors trick GEDmatch’s algorithm into thinking Paul and Jerry don’t match in small regions where they really do. That is, the mismatches are artifacts of the post-mortem testing process, not real mismatches. In short, Paul was Jerry’s father.

The mismatch regions caused the anomalies we saw in the GEDmatch summary of the match, the lower-than-expected total centimorgans, the higher-than-expected number of segments, and the short longest segment. Each mismatch region reduces the total and breaks up contiguous chromosomes into artificial subsections.

Despite the tragedy of missing Paul by a few days, there is joy to this tale. The Gillands have invited Jerry to their upcoming family reunion, and they’re thrilled to welcome their newest member, 52 years in the making.

As post-mortem and artifact testing from objects like mailing envelopes and hairbrushes become more common, we will see more instances where the initial interpretation of match data doesn’t tell the full story. Professionals, search angels, and community leaders will need to educate those they help to avoid misinterpretation of the match data and false disappointment. In Jerry’s case, an overview of the match suggested that Paul was not his father, while a closer look confirmed that he indeed was. As of this writing, such SNP-by-SNP comparisons can only be done at GEDmatch.

Expenses and Timeline

For those interested in similar analyses, here is a summary of the costs and timeline associated with Jerry’s search: