Genes R US — Pulling a rabbit out of the genomic-genealogy hat

Genes R US — Pulling a rabbit out of the genomic-genealogy hat

Posted: Wednesday, April 11, 2012 5:00 pm

OK, I will admit it. Both Barthélemy and I were disappointed when he got his 23andme DNA results back. We did not share enough DNA to be identified as blood cousins. 
According to the number one website for genealogical research, ancestry.com, Barthélemy and I are seventh cousins, twice removed, through our paternal Acadian sides. Specifically, I am Michel Hache dit Gallant and Anne Cormier’s eighth great-granddaughter and Barthélemy is their sixth great-grandson. (http://tiny.cc/rd5jcw) 
Any shared ancestry more distant than five generations is not likely to show up using single point DNA measurements, such as SNPs, which is what 23andme uses. The SNPs studied only compose 15 parts in one thousandth of one percentage of our total DNA, and that is a small number. To save words, let’s call cousins identified through SNPs-based DNA testing that are related within the last five generations DNA-based cousins. 
So why would I think that Barthélemy and I would show up as DNA-based cousins, anyway? Well, it started with David way before I got into ancestry.com. David and I are fourth DNA-based cousins identified through 23andme. David and Barthélemy have the same famous Acadian last name, Hache dit Gallant. Furthermore, the Acadians are famous for cousins marrying cousins. 
I waited several months to contact Barthélemy after David’s name appeared in my 23andme’s cousin list. Barthélemy and I were classmates for 12 years, with most of it in a small school. Although well-acquainted, I had not spoken to him in decades. 
Barthélemy was happy to submit a sample for 23andme DNA testing. When we did not show up as DNA-based cousins, Barthélemy challenged me to pull a bunny out of a hat to show that we do share a DNA connection.
Loving a challenge, especially if it has anything to do with data, I began thinking about how to do this. I needed some bunnies (people) and a hat (an experiment design). So here is another kind of Easter bunny story.
If person A is known to be a DNA-based cousin of person B and person B is a known to be a DNA-based cousin of person C, then persons A and C have a 50 percent chance of being DNA-based cousins, too — at some level.  Hence, over a large number of comparisons, we would expect persons A and C to be DNA-based cousins about half the time.
The Cousin experiment: Count the number of shared DNA-based cousins between Barthélemy and Nancy. Count the number of shared DNA-based cousins between Barthélemy and several random people. If Barthélemy and Nancy have more common DNA-based cousins, then they should have a higher probability (over a large number of cousins) of having a “real” DNA relationship back more than five generations. No guarantees, just a higher chance.
Each participant selected for my cousin experiment had more than 800 DNA-based cousins identified through 23andme. Only subsets of 23andme-DNA-identified cousins are willing to share their names publicly; and I need names in order to uniquely identify cousins for my comparisons. Let’s call a DNA cousin, who shares his name publicly, a public DNA cousin. These are my bunnies.
  So, besides Barthélemy and me, I selected all two of my other available candidates: my husband, Doug, and a friend of ours who also lives in the mid-South. None of us were identified as a DNA-based cousin to Barthélemy through 23andme. I am not a DNA-based cousin to Doug but my friend does show up as distant DNA-based cousin to Doug. That was a surprise.
Barthélemy has 186 public DNA-based cousins. I counted the number of public DNA-based cousins that he shares with each of the other three of us: 
Shared DNA-based cousins     Stats                         
Barthélemy => Me        6 out of 168 ~ 3.6%
Barthélemy => Doug    6 out of 226 ~ 2.7%
Barthélemy => Friend    3 out of 228 ~ 1.3%
My baseline, to interpret the above results, comes from counting the DNA-based cousins between each of the other three of us:
Shared DNA-based cousins     Stats  
Nancy => Doug          2 out of 226 < 1%
Nancy => Friend          1 out of 228 < 1%
Friend => Doug        6 out of 226 ~ 2.7%
By the way, Doug’s surprising number of DNA-based cousins with Barthélemy and my friend is due a strong common Scottish/Irish ancestry.
Based on my tiny experiment, Barthélemy and I have more potential DNA-based cousins with each other than either Doug or my friend.  We can expect no more than half of them to be real. Maybe none of them, if the numbers are too small. 
In summary, until whole genome sequencing drops down to something affordable, this is only bunny that I can pull out of my hat. 
Happy hopping till next week.
———
Editor’s note: Nancy Miller Latimer has worked in scientific research and development for 27 years. She blogs at NeuronalBeauty.BlogSpot.com
Published in The Messenger 4.11.12

Leave a Comment