Wednesday, October 1, 2014

DNA News: "Finding Your Roots" Review And Very Low Confidence Comparisons

I enjoyed the 2nd episode of the new season of "Finding Your Roots" on PBS. The last show focused on finding out more about absent fathers' lines. This show focused on sports heros. I could relate to Derek Jeter in the fact that I didn't know the origins of my Kapple surname until I did some research in the 1990's, which started my genealogy obsession. When you have an unusual surname people always ask where the name is from? It was great that it was possible to determine the origins of the name Jeter. That's a great gift for the Jeter family.

The only reservations I had about last nights episode was the language used when presenting DNA results. When Rebecca Lobo was told exactly which line her Ashkenazi admixture came from, I thought Dr. Gates was over stepping a little. It does seem, most likely, she got the admixture from the stated line (an Austrian line). Documentary evidence leads to that belief too, but you never know for sure unless you test more family members. Working with my DNA admixture results I was fooled a few times when trying to make an educated guess about where an admixture came from. It wasn't clear to me that I was wrong until I had more relatives tested. Ashkenazi is one of the admixtures which is easy to identify. The percentages are tricky though. She is predicted to be 10% Ashkenazi, so Dr. Gates surmised from that percentage that this came down from a particular great-grandparent based on rough inheritance percentages. Since inheritance is random, the only percentages we know for certain are that we inherit 50% of our DNA from each parent, after that generation there can be more variation in inheritance. For instance last week we learned that someone's father was substantially Ashkenazi according to available documents. This person's DNA results came out to be 12% percent Ashkenazi. So if we based our prediction of where this DNA came from we would guess farther back in time than parent based on estimates using average inheritance. Admixture percentages can fool you, and you can get admixture from more than one line. There is no way to say with 100% confidence which ancestor you got a particular admixture from, unless you test many relatives. When I watched this show before I got involved in autosomal testing and I heard, for instance, Dr. Gates say that a tradition of Native American ancestry was wrong based on DNA testing I believed that to be true. Since then I've learned that we don't inherit DNA from all of our ancestors; so it's possible a family tradition of Native American ancestry may be true even if you don't see it in your DNA results. So it's still possible Billy Jean King has some Native American roots?

Dr. Gates did use the word "suggests" when talking about results sometimes. The word "suggests" is often the best word to use when talking about results.

I'm convinced that Derek Jeter is descended from the family that owned his family. Y tests are more reliable than autosomal (autosomal testing is most reliable for close relatives). He is also most likely descended from the slave owner as stated. However, it's not 100% positive. I believe Dr. Gates did say that this connection was most likely the case, but didn't say 100% certain like he did with Rebecca Lobo. You can't look at DNA and say it came from a particular person or line with 100% certainty. With Y DNA you can only say you are definitely related or definitely not related using the test. It would be tricky to do a TV show and explain the intricacies of DNA testing. The general public probably isn't interested in learning about predictions either. They want to hear about definite conclusions. They might come away disappointed if they took a test after watching some of these episodes? Otherwise, I love the show and listen to the DNA test results with great interest.

Another point of interest is the DNA tests the guests took were blood tests. I wonder which company they used?

I was so happy to hear from someone in the Brenneman family line at Facebook. He and some of his family members had  their DNA tested. He, his sister and father were low confidence matches at AncestryDNA. We compared at GEDmatch and I found out he shares a 7 cM segment with my Mom. His sister and father shared an 8.9 cM segment in the same place. I would say low confidence is a good prediction based on the amount of DNA we share. I'm always interested in comparing at GEDmatch with AncestryDNA matches because the phasing process that produces the confidence levels doesn't always work. I'm very interested in the Brenneman line. I've found trees connecting my family to this one, but I can't find documentary evidence? Hoping DNA will help?

I'm hoping these segment on Chromosome 16 are from Brenneman line?


Your Genetic Genealogist said...

Hi Annette,
I would hope that you would realize that since I am doing the DNA research for the show, that I would be very thorough and careful about the conclusions we reached. Of course additional relatives were tested before we came to those conclusions! I spent scores of hours on each of the guest's DNA research (sometimes hundreds). This included testing lots of additional family members and theorized family members, often with more than one type of test. I can assure you that the conclusions stated on the show were done so with very high confidence levels.
The DNA tests were not done with blood. Those were just "stock shots". We used the three major DNA testing companies that I recommend for everyone.
Although I would love it if we could show more of the process of the DNA research, we could never explain the intricacies of the genetic genealogy research in the small amount of time allotted (and would lose most of the audience if we tried).
I'm glad you enjoyed the show.

Jean Shea said...

I too enjoy the show and watch it whenever I can. It is true that learning more about your ancestry can provide crucial missing pieces, but it has its limits.

When you say "Since then I've learned that we don't inherit DNA from all of our ancestors…," is that to say that it's possible that an ancestor's DNA can pass out of a person's genome? If so it presents an interesting scenario in which genealogically they remain an ancestor, but genetically they are out, is that correct? It's an interesting reverse of the case in which you have a significant genetic component of a group or race but your genealogical and social-cultural links are gone. I have a Native American component of roughly 50% but for all practical purposes, I am not Native American. My genetic results sort sent me reeling for a bit, but then I realized I might have the genetic bona fides for a NA ancestry, but that alone doesn't make me Native American.

Still, I am glad for the introduction of genetics into genealogical, it can provide some objective perspective.

I will add one caveat I believe to be true. I'm no genetic expert, but from what I understand the database that provides the basis for Native American ancestry is based on 108 individuals. That hardly seems like enough to make any statements in that regard with much confidence.


Your Genetic Genealogist said...

Hi Jean,

You make some very good points.

Blaine Bettinger wrote about our genetic family tree versus our genealogical family tree back in 2009:
Ancestors don't start to "fall off" our genetic family tree until about the 3rd great grandparent level, so all recent ancestors do contribute to our genomes, but not our more distant ones.

You have a huge amount of Native American DNA! I am wondering if you might be of Mexican, Central American or South American ancestry since that it the only time I have seen that high of a NA result...?

There are three different databases that we use for genetic genealogy (4 if you count NatGeo) and they all have different reference databases. I don't know the exact number, but across all of those, I am positive it adds up to more than 108 individuals. In the case of Billie Jean discussed in this post, we tested her at all 3 companies and none showed any Native DNA (or any Asian) at all. So, we came to that conclusion using more than only 108 reference samples. While it is true that she could have Native ancestry further back than 3rd great grandparent, she could not have had a "full-blooded" Native American ancestor more recently, which is what we were addressing in the episode.

With that said, without a doubt, we need more North American Native people willing to test before we can reach the kinds of conclusions that I and many in the community would love to see. It is a difficult challenge though since most do not wish to test and ancient remains are returned to the tribe by law. Differentiating among tribes and determining just how admixed Native Americans were a couple hundreds years ago remains elusive.


Annette said...

Thanks so much for the interesting comments Jean and CeCe! I'm interested in the Native American results because I'm supposed to be descended from Half King Tanacharison (Half King was a nickname used by George Wasington)? We're noticing some of his possible Owens line descendants have trace amounts of Native American DNA, and some none at all. I've read up and listened to all I can on the subject and learned it is possible, at that generational distance, that perceptible DNA from this ancestor may have disappeared. Another problem is there are few Native American samples, like Jean said. After reading Jean's comment I looked at MyOrigins to see how many Native Americans they used in their reference population. They didn't use any from their own data collection. Sounds like they used the Human Genome Diversity Project, Estonian Biocentre project, or the International HapMap Project? I wonder if all the companies use the same project, or projects, for Native American admixture? I wish they would tell us how they are coming up with the Native American results, so we can evaluate the credibility of the results?

Your Genetic Genealogist said...


The 108 number apparently came from 23andMe. For most "ethnicities" they use their own proprietary samples mostly put together from their customers, but for Native they also use HGDP samples. Here is what they are using (not sure if this will copy well):

Population Source Sample Size
Maya HGDP 25
Pima HGDP 25
Karitiana HGDP 24
Surui HGDP 21
Colombian HGDP 13

AncestryDNA uses the extensive proprietary Sorenson database that they purchased and NatGeo uses the reference samples that they collected themselves (also proprietary), so they are not all using the same samples.

Annette said...

Thanks very much CeCe for the interesting information! Great to see some numbers! Central and South America are fairly well covered. North America remains unrepresented, except for Pima. I don't blame these tribes for not trusting the researchers since previous generations of researchers hadn't treated them with respect.

I'm thinking the Genographic project might be the best place to test for Native American? Maybe they have some North American DNA samples?

It's interesting that my family's Mayan percentages range from 1% to 12% (the higher percentage represents my mother). I tend to believe the higher percentages since my maternal grandmother was Nicaraguan. My cousins Native American was 3% at one point at 23andme and is down to 1%. My Mom's Native American was 10% at FTDNA and is now down to 3% with MyOrigins. AncestryDNA is likely the most correct with 12%.