atDNA Case Study: Two Parents and One Child

Updated 5 Sep 2018

Version 3: Updated and expanded on 5 September 2018

See also:

• atDNA Case Study: Four Siblings

One hundred percent of everyone's DNA comes from their parents, with almost exactly 50% from each. Autosomal DNA tests, however, aren't perfect. As the amount of shared DNA with a match decreases, it becomes more and more difficult for a testing company to draw accurate conclusions. AncestryDNA includes a "confidence" figure with every match that states the accuracy of a proposed match. In some cases, a person will be shown as a match when they are not. In other cases, a person will not show up as a match when perhaps they should.

People who have tested themselves and both parents can start to estimate how match inaccuracies might affect test results.

On 6 Jan 2017, blogger Blaine Bettinger, "The Genetic Genealogist," wrote The Danger of Distant Matches - a discussion of how small-segment atDNA matches often aren't true matches at all. Since a child inherits all of his atDNA from one parent or the other, Bettinger correctly argues that atDNA matches that are found for a child but aren't shared by a parent identify flaws in today's widely-used low-cost DNA testing chips.

I thought it would be interesting to reproduce Bettinger's methodology and compare them to the atDNA results for myself and my parents. I provided a second set of my data on 14 Aug 2017 and a third set on 5 Sep 2018. Version 3 of the article is significantly expanded.

One caveat: my father and I were tested using the AncestryDNA v1 chipset; my mother's test used the v2 chipset.

My Parents' Matches at AncestryDNA

As of 7 Jan 2017, my father has 22,867 atDNA matches, of which 1,004 (4.4%) are rated by AncestryDNA as "close" matches - estimated as 4th cousins or closer, with 20cM or more of shared DNA. All three of his sisters have also been tested - for more information, read atDNA Case Study: Four Siblings. Their maternal first cousin has also been tested, as has their paternal second cousin. Articles analyzing those matches will be presented soon.

My mother has a much smaller set of 13,910 matches; 301 (2.2%) of them are rated "close." Several cousins have also been tested on both sides; more on them soon.

My father had 74,590 matches on 5 Sep 2018, of which 3,115 (4.2%) are "close" - greater than or equal to 20cM. My mother had 45,775 matches; 913 (2.0%) are close.

Matches at AncestryDNA

As of 7 Jan 2017, I had a total of 18,020 matches; Bettinger writes that he has 16,193. On 5 Sep 2018, I had 57,884 matches - more than three times the number from 21 months ago. Note that all three "K Ireland" columns are my own results taken at different dates. The matches break down like this:

Shared With Parents

Not Shared With Parents

Here are the matches not shared with our parents; all percentages are based only on the number of matches not shared with either parent:

False Negatives - or False Positives?

For the 2018 Update, I'm adding a new step to my analysis. Although there's no definitive way to determine if a match not shared by a parent is a false negative for the parent or a false positive for the child, comparing the child's matches to a parent's siblings can provide some additional insight.

I compared my matches that I don't share with my parents with three of my father's sisters. I discovered that one or more of my father's three siblings matched to 1,534 of the 10,669 matches that I matched to but my parents did not.

Not Shared With My Parents - But Shared With My Aunts

Here's a chart of the 27 matches that my three paternal aunts and I matched, but my parents did not:

Interestingly, the match in yellow, scr*****, is a known 3C1R to my father and his sisters. Although my father is the only person who doesn't match to her, he and sisters all match scr's mother with the following shared DNA amounts:

• GI: 44.2 cM
• GP: 65.4 cM
• SI: 98.7 cM
• SM: 40.6 cM

Finally, the two matches shown in pink, Car and Ela, are members of the same line. I don't know our exact relationship to them but they both share matches with the same people we do.

Further Reading

A number of people have written similar analyses:

• Blaine Bettinger - January 2017 (cited above)
• Debbie Kennett - August 2017
• Alex Coles - August 2017
• Ann Raymont - August 2017

Send me a link to your analysis at and I'll include it here.