Sunday, September 26, 2010

Known Relative Studies with 23andMe: Expected Percentages

I have had quite a few known relatives test with 23andMe recently. In the coming months, I will be writing a number of posts based on these results. As an introduction, I wanted to review some of the information that 23andMe and FTDNA provide in their FAQs regarding their autosomal DNA tests, Relative Finder and Family Finder.  All of these percentages are estimates and will vary due to the random nature of genetic inheritance. The following statistics apply to both 23andMe's Relative Finder (RF) and FTDNA's Family Finder (FF).

Expected percentage of shared DNA between:
Parent/child/siblings = 50%
Grandparent/grandchild/aunt/uncle/nephew/niece/half-siblings = 25%
1st Cousins/great-grandparent/great aunt or uncle/grandnephew or niece = 12.5%
1st Cousins once removed = 6.25%
2nd Cousins = 3.125%
2nd Cousins once removed = 1.563%
3rd Cousins = .781%
4th Cousins = .195%
5th Cousins = .049%
6th Cousins = .012%
7th Cousins = .003%
8th Cousins = .001%

The likelihood of detecting shared DNA with a known relative significant enough to show up in RF or FF:
1st Cousins and closer: 100%
2nd Cousins:  > 99%
3rd Cousins:  ~ 90%
4th Cousins: ~ 45% (FTDNA says > 50%)
5th Cousins: ~ 15% (FTDNA says > 10%)
6th Cousins and more distant: < 5% (FTDNA says < 2%)

In some families, there will be situations that will complicate the above predictions, such as cousin marriages and half-siblingship, but, in general, they are the guidelines that should be used in analysis and comparisons.

Further reading from this series -
Known Relative Studies: Great Grandchild Inheritance
A Success Story and the Randomness of Autosomal DNA Inheritance (Fourth and Fifth Cousins)
Known Relative Studies: Second Cousins 
Known Relative Studies: More Second Cousin Comparisons 
Known Relative Studies: Second Cousin Comparisons, Allen Great Grandparents
Known Relative Studies: Second Cousins or Half Second Cousins 
Known Relative Studies: Identifying DNA from Great Grandparents Using Second Cousin Comparisons
Known Relative Studies: I Found My Third Cousin Today! 
Known Relative Studies: A Third Cousin Comparison and More Random Autosomal DNA Inheritance
Known Relative Studies: Purdy Fourth and Fifth Cousins
Autosomal DNA Matching and the Importance of Testing Multiple Family Members (Ninth Cousins)
Ratekin Seventh Cousin
A Second Cousin Adds to My Chromosome Map and Answers a Nagging Genealogical Question


  1. When I first got my results at 23andme, I expected my close relatives to conform to the above expected percentages. It was interesting to see how they differed. I share 54.30% with my sister, 31.39% with one paternal half sister and 23.67% with the other (this was the most surprising!), 24.64% with my paternal aunt, 14.85% with my dad's double first cousin, and 10.18% with my maternal great uncle.

    I still have four kits from DNA day to give away: one is going to another one of my dad's double cousins, two are going to my maternal aunt & uncle, and one is going to a maternal great aunt. I really need to get on that and talk everyone into spitting :)

  2. Yes you do! I sure wish I still had kits. The more I learn, the more I want to research... :-)

  3. I am descended from Terrance Travis, who moved from Ireland to Leeds. Is the book about your Travis relatives available?

  4. I am descended from Terrence Travis, (1805-1902) who moved from Manorhamilton, County Cavan, to Leeds with his wife Mary McCann (1820-1902). Their children were Catherine, James, John, and Terrence. They came to Chicago when their children emigrated in the 1860s. Is the book about your Travis relatives still available?

  5. What you have documented is the sensitivity of the genetic matching algorithm, in other words the number of true matches actually detected. Is there a similar figure for specificity i.e the number of truly unrelated pairs that are not matched by the algorithm. In other words, what is the chance that a detected match by the algorithms (say for 4th cousin) is actually a "false positive" and the two are in fact unrelated (or much more distantly related). In medical diagnosis it is important to compute both sensitivity and specificity.