Friday, October 6, 2017

I Got It Wrong: A Retraction/Correction

Back on July 17th I published a blog with the heading How autosomal DNA is inherited: 3 siblings compared. At the time my son Jon was bothered by part of it related to the passing of DNA from great-grandparent to great-grandchild but I was too dense to comprehend his objection. Yesterday after more than six thousand seven hundred page views, Philip Gammon commented:

Hi, I'm surprised that no one has pointed out that calculation of the contribution from great-grandparents in the article is incorrect. The contribution of each pair of great-grandparents is not 25%, it is equal to the contribution the great-grandchild receives from the grandparent. This figure varies significantly as pointed out in the earlier paragraphs but this information is then ignored in the latter half of the article. 
This time I finally got the point. I wish to publicly thank Philip for his initiative in bringing this error to my attention. I would like to think that many of the rest of my readers also noticed this incorrect conclusion on my part but were too kind to bring it to my attention. However, if we truly are going to consider ourselves to be citizen scientists, we must take on the responsibility offering constructive peer review of the work of our colleagues. That is how we all learn. What follows is a revision of that original post.

Inheritance of atDNA from grandparents:

If you know how much atDNA you inherited from a grandparent or great-grandparent, can you calculate how much you inherited from that person's spouse?

By now most of us know that each of us get half of our autosomal DNA from our mom and half from our dad. But how we inherit from earlier generations of our ancestors is less well understood. We are generally told after that first generation the inheritance pattern is random. However, we often see charts that show we should expect to inherit about 25% from each grandparent and 12.5% from each great-grandparent, etc. Are there any additional rules?

Below is a small case study that explores these issues:

Inheritance from maternal grandparents are shown in the left hemispheres above and paternal contributions are in the right hemispheres.

The chart above shows how much atDNA each of three brothers inherited from each of their grandparents. Only the middle one, Grandson 2, came close to the 25% average we might expect. However, note that each grandson inherited exactly 50% from his maternal grandparents and 50% from his paternal grandparents. However, the amount inherited from each individual within a set of grandparents varied considerably. The exact percentages can be seen in the table below.

As noted above Grandson 2 came the closest to the theoretical 25% inheritance rate. The atDNA inherited by Grandson 1 was skewed somewhat from that distribution. The practical implications for even this amount of variation can be important to your genealogical research. 

In this example the maternal grandfather is Ashkenazi -- a group that statistically has been over sampled in genetic genealogy databases. He had more than eleven thousand matches in FTDNA's Family Finder. The maternal grandmother came from an area in Eastern Europe very near the origin of her mate; but she was non-Ashkenazi. She had less than five hundred matches in Family Finder. The 4% difference in the amount of Ashkenazi DNA inherited resulted in Grandson 2 having more than five thousand matches while his full brother, Grandson 1, had about a thousand matches less. This also had a noticeable difference in their ethnicity predictions. 

When Grandson 3's Family Finder results are reported, it is expected that his matches and ethnicity predictions will vary ever further from those of his two brothers -- particularly Grandson 2. As we will see below, Grandson 3 inherited more atDNA from his great-grandfather -- the father of his maternal grandmother -- than the 15% he inherited from his Ashkenazi grandfather. This variation from the expected 25% from each grandparent could have been influenced by the preimplantation genetic screening process that selected a fertilized embryo that did not include an autosomal dominant gene that leads to a potentially fatal heart defect.  

All the results above are based on actual test results. However, if one of the grandparents had not been tested, we could have calculated their contribution. Even if two had not tested we could have calculated their contributions IF we had results from each of their spouses. You have probably already figured out that the contribution of the missing grandparent spouse would be 50% minus the contribution of their known spouse.

Inheritance of atDNA from great-grandparents: 

This kind of calculation becomes more relevant in the current case when I took a look at contributions from great-grandparents. We only have actual test results from one great-grandparent who was tested a decade ago about a year before he died in his mid 90s. So we can use his actual test results and also use them to calculate the atDNA contribution of his wife. By now you probably can guess that the hypothesized contribution of the non-tested great-grandmother will be constructed by subtracting his tested atDNA amount from the tested contribution of the relevant grandparent. The latter would equal the total contribution of each pair of biological great-grandparents.

For the oldest brother it turns out that the tested great-grandfather had contributed 12.4% of his autosomal DNA. This is very close to the theoretical 12.5%. In this case the grandmother had contributed about 28%. This left 15.6% for the great-grandmother to have contributed.

Great-Grandson 1

atDNA from great-grandparents

Grandson 1
Grandson 2
Grandson 3
Tested Great-Grandfather






Calculated Great-Grandmother



All other Great-Grandparents




In all cases the great-grandsons received more atDNA from the great-grandmother than they did from the great-grandfather. The tested great-grandfather had passed down 10.3% of the atDNA of the second brother and 16.6% of the atDNA of the third brother. In fact the 16.6% of his overall atDNA that the third brother received through his maternal grandmother and mother was more total atDNA than this brother had inherited from his paternal grandfather (15%, see above). This result was surprising but perhaps not unusual. We are preconditioned to expect that we received about one half the amount of atDNA from each ancestor in each receding ancestral generation. In this case that expectation was confounded because of the large amount of maternally inherited atDNA that the mother of the brothers had passed down to her third son. Although the mother had inherited half of her own atDNA from each of her parents, what she passed down to the third son had disproportionately over-represented what she had inherited from her mother by a ratio of more than two to one -- 35% to 15%.

The second son inherited less of the calculated contribution from his great-grandmother than either of his brothers in part because he inherited less of his grandmother's atDNA. He inherited 13.7% versus their 15.6% and 18.4%.

Great-Grandson 2

The third son inherited the most atDNA from the tested great-grandfather and did so by a considerable ratio. Although his 16.6% inheritance does not seem like a big difference from the 12.4% of one brother and the 10.3% of the other, it is about 35% more than one and 63% more than the other. In terms of inherited ethnicity and traits, these differences could matter. 

Great-Grandson 3

The Takeaways:
1. We inherit 50% of our atDNA from each parent; 50% from each set of grandparents; then what we inherit significantly different amounts from each set of great-grandparents, etc.
2. We often inherit significantly different amounts of atDNA from each member of a set of ancestors.
3. Even full siblings often inherit significantly different amounts of atDNA from each of their grandparents and great-grandparents.
4. These differences can significantly affect the numbers of autosomal matches reported for each sibling.
5. These differences can significantly affect ethnicity estimates.
6. These differences can significantly affect health and other heritable characteristics.
7. The contributions of some ancestors can begin to fade significantly within two or three generations while those of others can remain robust much longer.
8. If you have test results for at least one grandparent or great-grandparent, you can calculate the atDNA contribution of their partners.


  1. Dave you use aDNA when referring to autosomal DNA. My understanding is that aDNA refers to ancient DNA and the use of atDNA is more appropiate for autosomal DNA used for genealogical matching. Your thoughts?

    1. Ditto, thanks Alasdair.
      Thanks, also, David for putting your observations out there.

    2. Ditto, Alasdair.
      Thanks also David for putting your observations out there for us all.

  2. Alasdair, you raise a good point. I must admit I'm guilty of equivocating between using aDNA and atDNA to refer to autosomal DNA. For most of us working primarily in the genealogical era, this has not yet emerged as a major issue. However, when we turn more anthropological and work with haplogroups going back into prehistory, this can become an issue. OK, I think you have persuaded me to standardize on "atDNA" in my writing from now on. Thank you for raising this issue.

  3. Since you brought up terminology, as I prepared material for a class I was teaching I learned that some DNA researchers have used x-DNA to indicate a type of tagged DNA used for research and so I have begun to refer to DNA results from the X-chromosome as "X-chromosome-DNA" (just to be absolutely clear). Yes, it seemed clunky at first, but as I get used to it, it's become a habit. After all, clarity and consistency is key.

  4. I think you also need to look at your data by chromesome as well. Did the gdparent/ggrdparent pass the total cMs to mnay chromosomes or each generation left out chr. 6 [pick one]. And they the comparision of such to the next generation. Similarly to our family history research, the more exhaustive analysis we do helps to link our DNA to our ancestors, our matches and the scientific community. Thanks for a great review of DNA inheritance.