Sunday, September 1, 2013

Solving A Mystery With Women's DNA

Many still believe that only male DNA is useful for genealogical purposes. They need to wake up and smell DNA developments of the last 3 years. I recently helped a woman unravel an old mystery in her family using her autosomal and mitochondrial DNA test results. I will call her "Allyson" to protect the guilty dead and the living innocents. Here are some of the facts:

  • The adopted daughter of Allyson’s great aunt was thought to be a family member;
  • This adoptee was born in the 1920s;
  • The adoptee’s daughter, who I will call "Bertha", is alive and has taken a mitochondrial DNA test and a Family Finder autosomal DNA test at FTDNA.

Speculation within the family included:

  • This adoptee was the result of an illicit coupling of Allyson’s grandmother and her ex-husband long after their divorce;
  • This adoptee was the result of a teenage fling of Allyson’s mother.
  •  Allyson also has taken both a mitochondrial DNA test and a Family Finder autosomal DNA test at FTDNA.

The Mitochondrial DNA results:

Mitochondrial DNA is more definitive in ruling out potential relationships than in proving them. Their results could have ruled out Allyson’s mother and grandmother as potential mothers for Bertha’s mother. However, Allyson and Bertha are exact matches over all 16,569 locations on their mitochondrial DNA. They are the only exact matches for each other currently in the database. This means that the two share a direct maternal line (umbilical cord) ancestress probably within genealogical time. FTDNA says that such a match has a 50% probability of a common direct umbilical cord line ancestress within 5 generations and a 95% probability of a common direct umbilical cord ancestress within 22 generations. In this case the results do not prove how close the common ancestress actually lived, but they do add some credibility to the belief that this was an adoption within the family. Based on this result some hypotheses were formulated.

Research Hypotheses: 

#1. Bertha’s mom was the biological daughter of Allyson’s maternal grandmother and grandfather; 

#2. Bertha’s mom was the biological daughter of Allyson’s maternal grandmother and some unknown male – at least unknown to us; 

#3. Bertha’s mom was the biological daughter of Allyson’s mom and some unknown male.

No hypothesis was defined to include Allyson’s mom AND dad being the biological parents as existing letters clearly document that they did not meet until a few years after the birth of Bertha’s mother. On the other hand it can be documented that Allyson’s maternal grandmother and grandfather remained in contact long after their divorce.

Dick Eastman has famously blogged that there is no such thing as a ½ cousin. He quoted the venerable Black’s Law Dictionary to support his assertion. Well Dick, we would never be able to solve some of our genealogical mysteries unless we are able to split some cousins and other relatives in this manner.

If we want support for hypothesis # 1 above, we would need to find evidence that Allyson and Bertha could be 1st cousins. To support hypothesis # 2 above, we would need to find evidence that Allyson and Bertha could be ½ first cousins. If hypothesis # 3 above is correct Allyson would have a ½ aunt relationship with Bertha. In these three cases Allyson and Bertha would be expected to share either about 12.5% (1st cousins), about 6.25% (½ first cousins) or 12.5% (½ aunt). These percentages can be derived from the chart on the ISOGG Wiki: 

Cousin tree (with genetic kinship)

Relation with Adoptee’s Daughter “Bertha”
Maternal Grandmother is mother of Adoptee
Allyson’s Mother is mother
Maternal Grandfather is Father
Hypothesis #1
Maternal Grandfather is not Father
Hypothesis #2
Allyson’s father is not Father of Adoptee
Hypothesis #3
1st cousin—12.5%
Half 1st cousin—6.25%
Half Aunt—12.5%
Male Cousin
1st cousin—12.5%
Half 1st cousin—6.25%
Half 1st cousin-once removed—3.125%
Female Cousin #1
1st cousin—12.5%
Half 1st cousin—6.25%
Half 1st cousin-once removed—3.125%
Female Cousin #2
1st cousin—12.5%
Half 1st cousin—6.25%
Half 1st cousin-once removed—3.125%

 Expected % of Shared Autosomal DNA 
for each Hypothesis

Oops! Using the autosomal data at hand we would not be able to distinguish between hypotheses 1 and 3. The expected amount of shared DNA would be the same in both cases. Fortunately there are three living 1st cousins of Allyson who might be willing to test. Two are female and one is male. In this case their sex would not matter because we have already established the maternal haplogroup of the adoptee. All we would need is their autosomal matches with Allyson and Bertha. These cousins would not add to our ability to distinguish between options #1 and #2 because their amounts of DNA shared with Bertha should mirror those of Allyson in both cases. Both Allyson and each of the 3 cousins should be shown to be 1st cousins with Bertha in #1 and ½ first cousins in # 2. So no increase in diagnostic ability is gained by collecting DNA from the 3 additional cousins.

But wait! We do not need them to help distinguish between #1 and #2. We might need their test results to differentiate between #1 and #3. Great! While Allyson would be a ½ aunt (sharing about 12.5%) in #3, the other 3 cousins would only be ½ 1st cousins (sharing about 3.125%) a significant difference. So it looks like it may be possible to collect enough information to solve this mystery.

The Solution:

Actually, it turned out to be quite a bit simpler than all that. As we saw much earlier in this saga, FTDNA predicted that Allyson and Bertha were 1st to 2nd cousins. For our purposes that is quite an extensive range. If they share matching segments that would be expected of 1st cousins, they would also share the amount Bertha would expect to share with a ½ aunt. But that is not the amount they share. They only share about ½ that amount. In this case it would have been easier to determine this if Allyson and Bertha had tested at 23andMe rather than FTDNA since the former reports matches in terms of % of match. FTDNA reports it in terms of centiMorgans (cMs) and SNPs matched. Although there may be handy conversion tables to translate these into % of match, Dr D had difficulty finding one. But he did find this one posted by Lindsay in her blog Confessions of a Cryokid:

According to Matt Dexter, on the FTDNA Forum, here is a breakdown of sum shared DNA ranges for various relationships:
  • Parent/child: 3539-3748 centimorgans (cMs)
  • 1st cousins: 548-1034 cMs
  • 1st cousins once removed: 248-638 cMs
  • 2nd cousins: 101-378 cMs

Although this table turned out to be very useful, it was compiled over 2 years ago. In autosomal DNA testing time that was almost light years ago. We now have tens of of thousands of more results that should have expanded our databases. Dr D would expect a currently compiled breakdown to show more overlap between the categories. 

Once the table in Lindsay's blog post was found, it showed that the 428.42 cMs shared between Allyson and Bertha were well within the expected range that would be expected for a 1st cousin-once removed relationship.

But 1st cousin-once removed is not one of the expected outcomes in the hypothesis table above. Or is it? Isn't it the equivalent of a ½ first cousin? I think we have a winner! The hypothesis that best meets the autosomal results is #2. The most likely biological parents of Bertha's mom are Allyson's maternal grandmother and an unknown male. Allyson's maternal grandfather seems to be ruled out as the biological father of Bertha's mother. Although testing additional cousins might lead to greater certainty, that does not appear to be necessary to identify the most likely candidate to be Bertha's grandmother. 

No comments:

Post a Comment