Elizabeth Warren has released DNA testing results after being publicly challenged and derided as “Pochahontas” as a result of her claims of a family story indicating that her ancestors were Native America. If you’d like to read the specifics of the broo-haha, this Washington Post Article provides a good summary, along with additional links.
I personally find name-calling of any type unacceptable behavior, especially in a public forum, and while Elizabeth’s DNA test was taken, I presume, in an effort to settle the question and end the name-calling, what it has done is to put the science of genetic testing smack dab in the middle of the headlines.
This article is NOT about politics, it’s about science and DNA testing. I will tell you right up front that any comments that are political or hateful in nature will not be allowed to post, regardless of whether I agree with them or not. Unfortunately, these results are being interpreted in a variety of ways by different individuals, in some cases to support a particular political position. I’m presenting the science, without the politics.
This is the first of a series of two articles.
I’m dividing this first article into four sections, and I’d ask you to read all four, especially before commenting. A second article, Possibilities – Wringing the Most Out of Your DNA Ethnicity Test will follow shortly about how to get the most out of an ethnicity test when hunting for Native American (or other minority, for you) ethnicity.
Understanding how the science evolved and works is an important factor of comprehending the results and what they actually mean, especially since Elizabeth’s are presented in a different format than we are used to seeing. What a wonderful teaching opportunity.
- Family History and DNA Science – How this works.
- Elizabeth Warren’s Genealogy
- Elizabeth Warren’s DNA Results
- Questions and Answers – These are the questions I’m seeing, and my science-based answers.
My second article, Possibilities – Wringing the Most Out of Your DNA Ethnicity Test will include:
- Potential – This isn’t all that can be done with ethnicity results. What more can you do to identify that Native ancestor?
- Resources with Step by Step Instructions
Now, let’s look at Elizabeth’s results and how we got to this point.
Family Stories and DNA
Every person that grows up in their biological family hears family stories. We have no reason NOT to believe them until we learn something that potentially conflicts with the facts as represented in the story.
In terms of stories handed down for generations, all we have to go on, initially, are the stories themselves and our confidence in the person relating the story to us. The day that we begin to suspect that something might be amiss, we start digging, and for some people, that digging begins with a DNA test for ethnicity.
My family had that same Cherokee story. My great-grandmother on my father’s side who died in 1918 was reportedly “full blooded Cherokee” 60 years later when I discovered she had existed. Her brothers reportedly went to Oklahoma to claim headrights land. There were surely nuggets of truth in that narrative. Family members did indeed to go Oklahoma. One did own Cherokee land, BUT, he purchased that land from a tribal member who received an allotment. I discovered that tidbit later.
What wasn’t true? My great-grandmother was not 100% Cherokee. To the best of my knowledge now, a century after her death, she wasn’t Cherokee at all. She probably wasn’t Native at all. Why, then, did that story trickle down to my generation?
I surely don’t know. I can speculate that it might have been because various people were claiming Native ancestry in order to claim land when the government paid tribal members for land as reservations were dissolved between 1893 and 1914. You can read more about that in this article at the National Archives about the Dawes Rolls, compiled for the Cherokee, Creek, Choctaw, Chickasaw and Seminole for that purpose.
I can also speculate that someone in the family was confused about the brother’s land ownership, especially since it was Cherokee land.
I could also speculate that the confusion might have resulted because her husband’s father actually did move to Oklahoma and lived on Choctaw land.
But here is what I do know. I believed that story because there wasn’t any reason NOT to believe it, and the entire family shared the same story. We all believed it…until we discovered evidence through DNA testing that contradicted the story.
Before we discuss Elizabeth Warren’s actual results, let’s take a brief look at the underlying science.
Enter DNA Testing
DNA testing for ethnicity was first introduced in a very rudimentary form in 2002 (not a typo) and has progressed exponentially since. The major vendors who offer tests that provide their customers with ethnicity estimates (please note the word estimates) have all refined their customer’s results several times. The reference populations improve, the vendor’s internal software algorithms improve and population genetics as a science moves forward with new discoveries.
Note that major vendors in this context mean Family Tree DNA, 23andMe, the Genographic Project and Ancestry. Two newer vendors include MyHeritage and LivingDNA although LivingDNA is focused on England and MyHeritage, who utilizes imputation is not yet quite up to snuff on their ethnicity estimates. Another entity, GedMatch isn’t a testing vendor, but does provide multiple ethnicity tools if you upload your results from the other vendors. To get an idea of how widely the results vary, you can see the results of my tests at the different vendors here and here.
My initial DNA ethnicity test, in 2002, reported that I was 25% Native American, but I’m clearly not. It’s evident to me now, but it wasn’t then. That early ethnicity test was the dinosaur ages in genetic genealogy, but it did send me on a quest through genealogical records to prove that my family member was indeed Native. My father clearly believed this, as did the rest of the family. One of my early memories when I was about four years old was attending a (then illegal) powwow with my Dad.
In order to prove that Elizabeth Vannoy, that great-grandmother, was Native I asked a cousin who descends from her matrilineally to take a mitochondrial DNA test that would unquestionably provide the ethnicity of her matrilineal line – that of her mother’s mother’s mother’s direct line. If she was Native, her haplogroup would be a derivative either A, B, C, D or X. Her mitochondrial DNA was European, haplogroup J, clearly not Native, so Elizabeth Vannoy was not Native on that line of her family. Ok, maybe through her dad’s line then. I was able to find a Vanoy male descendant of her father, Joel Vannoy, to test his Y DNA and he was not Native either. Rats!
Tracking Elizabeth Vannoy’s genealogy back in time provided no paper-trail link to any Native ancestors, but there were and are still females whose surnames and heritage we don’t know. Were they Native or part Native? Possibly. Nothing precludes it, but nothing (yet) confirms it either.
DNA testing is notorious for unveiling unexpected results. Adoptions, unknown parents, unexpected ethnicities, previously unknown siblings and half-siblings and more.
Ethnicity is often surprising and sometimes disappointing. People who expect Native American heritage in their DNA sometimes don’t find it. Why?
- There is no Native ancestor
- The Native DNA has “washed out” over the generations, but they did have a Native ancestor
- We haven’t yet learned to recognize all of the segments that are Native
- The testing company did not test the area that is Native
Not all vendors test the same areas of our DNA. Each major company tests about 700,000 locations, roughly, but not the same 700,000. If you’re interested in specifics, you can read more about that here.
Everyone receives half of their autosomal DNA from each parent.
That means that each parent contributes only HALF OF THEIR DNA to a child. The other half of their DNA is never passed on, at least not to that child.
Therefore, ancestral DNA passed on is literally cut in half in each generation. If your parent has a Native American DNA segment, there is a 50-50 chance you’ll inherit it too. You could inherit the entire segment, a portion of the segment, or none of the segment at all.
That means that if you have a Native ancestor 6 generations back in your tree, you share 1.56% of their DNA, on average. I wrote the article, Ancestral DNA Percentages – How Much of Them is in You? to explain how this works.
These calculations are estimates and use averages. Why? Because they tell us what to expect, on average. Every person’s results will vary. It’s entirely possible to carry a Native (or other ethnic) segment from 7 or 8 or 9 generations ago, or to have none in 5 generations. Of course, these calculations also presume that the “Native” ancestor we find in our tree was fully Native. If the Native ancestor was already admixed, then the percentages of Native DNA that you could inherit drop further.
Why Call Ethnicity an Estimate?
You’ve probably figured out by now that due to the way that DNA is inherited, your ethnicity as reported by the major testing companies isn’t an exact science. I discussed the methodology behind ethnicity results in the article, Ethnicity Testing – A Conundrum.
It is, however, a specialized science known as Population Genetics. The quality of the results that are returned to you varies based on several factors:
- World Region – Ethnicity estimates are quite accurate at the continental level, plus Jewish – meaning African, Indo-European, Asian, Native American and Jewish. These regions are more different than alike and better able to be separated.
- Reference Population – The size of the population your results are being compared to is important. The larger the reference population, the more likely your results are to be accurate.
- Vendor Algorithm – None of the vendors provide the exact nature of their internal algorithms that they use to determine your ethnicity percentages. Suffice it to say that each vendor’s staff includes population geneticists and they all have years of experience. These internal differences are why the estimates vary when compared to each other.
- Size of the Segment – As with all genetic genealogy, bigger is better because larger segments stand a better chance of being accurate.
- Academic Phasing – A methodology academics and vendors use in which segments of DNA that are known to travel together during inheritance are grouped together in your results. This methodology is not infallible, but in general, it helps to group your mother’s DNA together and your father’s DNA together, especially when parents are not available for testing.
- Parental Phasing – If your parents test and they too have the same segment identified as Native, you know that the identification of that segment as Native is NOT a factor of chance, where the DNA of each of your parents just happens to fall together in a manner as to mimic a Native segment. Parental phasing is the ability to divide your DNA into two parts based on your parent’s DNA test(s).
- Two Chromosomes – You have two chromosomes, one from your mother and one from your father. DNA testing can’t easily separate those chromosomes, so the exact same “address” on your mother’s and father’s chromosomes that you inherited may carry two different ethnicities. Unless your parents are both from the same ethnic population, of course.
All of these factors, together, create a confidence score. Consumers never see these scores as such, but the vendors return the highest confidence results to their customers. Some vendors include the capability, one way or another, to view or omit lower confidence results.
Parental Phasing – Identical by Descent
If you’re lucky enough to have your parents, or even one parent available to test, you can determine whether that segment thought to be Native came from one of your parents, or if the combination of both of your parent’s DNA just happened to combine to “look” Native.
Here’s an example where the “letters” (nucleotides) of Native DNA for an example segment are shown at left. If you received the As from one of your parents, your DNA is said to be phased to that parent’s DNA. That means that you in fact inherited that piece of your DNA from your mother, in the case shown below.
That’s known as Identical by Descent (IBD). The other possibility is what your DNA from both of your parents intermixed to mimic a Native segment, shown below.
This is known as Identical by Chance (IBC).
You don’t need to understand the underpinnings of this phenomenon, just remember that it can happen, and the smaller the segment, the more likely that a chance combination can randomly happen.
Elizabeth Warren’s Genealogy
Elizabeth Warren’s genealogy, is reported to the 5th generation by WikiTree.
Elizabeth’s mother, Pauline Herring’s line is shown, at WikiTree, as follows:
Notice that of Elizabeth Warren’s 16 great-great-great grandparents on her mother’s side, 9 are missing.
Paper trail being unfruitful, Elizabeth Warren, like so many, sought to validate her family story through DNA testing.
Elizabeth Warren’s DNA Results
Elizabeth Warren didn’t test with one of the major vendors. Instead, she went directly to a specialist. That’s the equivalent of skipping the family practice doctor and going to the Mayo Clinic.
From the report, here are Dr. Bustamante’s credentials:
Dr. Carlos D. Bustamante is an internationally recognized leader in the application of data science and genomics technology to problems in medicine, agriculture, and biology. He received his Ph.D. in Biology and MS in Statistics from Harvard University (2001), was on the faculty at Cornell University (2002-9), and was named a MacArthur Fellow in 2010. He is currently Professor of Biomedical Data Science, Genetics, and (by courtesy) Biology at Stanford University. Dr. Bustamante has a passion for building new academic units, non-profits, and companies to solve pressing scientific challenges. He is Founding Director of the Stanford Center for Computational, Evolutionary, and Human Genomics (CEHG) and Inaugural Chair of the Department of Biomedical Data Science. He is the Owner and President of CDB Consulting, LTD. and also a Director at Eden Roc Biotech, founder of Arc-Bio (formerly IdentifyGenomics and BigData Bio), and an SAB member of Imprimed, Etalon DX, and Digitalis Ventures among others.
He’s no lightweight in the study of Native American DNA. This 2012 paper, published in PLOS Genetics, Development of a Panel of Genome-Wide Ancestry Informative Markers to Study Admixture Throughout the Americas focused on teasing out Native American markers in admixed individuals.
From that paper:
Ancestry Informative Markers (AIMs) are commonly used to estimate overall admixture proportions efficiently and inexpensively. AIMs are polymorphisms that exhibit large allele frequency differences between populations and can be used to infer individuals’ geographic origins.
Using a panel of AIMs distributed throughout the genome, it is possible to estimate the relative ancestral proportions in admixed individuals such as African Americans and Latin Americans, as well as to infer the time since the admixture process.
The methodology produced results of the type that we are used to seeing in terms of continental admixture, shown in the graphic below from the paper.
Matching test takers against the genetic locations that can be identified as either Native or African or European informs us that our own ancestors carried the DNA associated with that ethnicity.
Of course, the Native samples from this paper were focused south of the United States, but the process is the same regardless. The original Native American population of a few individuals arrived thousands of years ago in one or more groups from Asia and their descendants spread throughout both North and South America.
Elizabeth’s request, from the report:
To analyze genetic data from an individual of European descent and determine if there is reliable evidence of Native American and/or African ancestry. The identity of the sample donor, Elizabeth Warren, was not known to the analyst during the time the work was performed.
Elizabeth’s test included 764,958 genetic locations, of which 660,173 overlapped with locations used in ancestry analysis.
The Results section says after stating that Elizabeth’s DNA is primarily (95% or greater) European:
The analysis also identified 5 genetic segments as Native American in origin at high confidence, defined at the 99% posterior probability value. We performed several additional analyses to confirm the presence of Native American ancestry and to estimate the position of the ancestor in the individual’s pedigree.
The largest segment identified as having Native American ancestry is on chromosome 10. This segment is 13.4 centiMorgans in genetic length, and spans approximately 4,700,000 DNA bases. Based on a principal components analysis (Novembre et al., 2008), this segment is clearly distinct from segments of European ancestry (nominal p-value 7.4 x 10-7, corrected p-value of 2.6 x 10-4) and is strongly associated with Native American ancestry.
The total length of the 5 genetic segments identified as having Native American ancestry is 25.6 centiMorgans, and they span approximately 12,300,000 DNA bases. The average segment length is 5.8 centiMorgans. The total and average segment size suggest (via the method of moments) an unadmixed Native American ancestor in the pedigree at approximately 8 generations before the sample, although the actual number could be somewhat lower or higher (Gravel, 2012 and Huff et al., 2011).
Dr. Bustamante’s Conclusion:
While the vast majority of the individual’s ancestry is European, the results strongly support the existence of an unadmixed Native American ancestor in the individual’s pedigree, likely in the range of 6-10 generations ago.
I was very pleased to see that Dr. Bustamante had included the PCA (Principal Component Analysis) for Elizabeth’s sample as well.
PCA analysis is the scientific methodology utilized to group individuals to and within populations.
Figure one shows the section of chromosome 10 that showed the largest Native American haplotype, meaning DNA block, as compared to other populations.
Remember that since Elizabeth received a chromosome from BOTH parents, that she has two strands of DNA in that location.
Here’s our example again.
Given that Mom’s DNA is Native, and Dad’s is European in this example, the expected results when comparing this segment of DNA to other populations is that it would look half Native (Mom’s strand) and half European (Dad’s strand.)
The second graphic shows Elizabeth’s sample and where it falls in the comparison of First Nations (Canada) and Indigenous Mexican individuals. Given that Elizabeth’s Native ancestor would have been from the United States, her sample falls where expected, inbetween.
Let’s take a look at some of the questions being asked.
Questions and Answers
I’ve seen a lot of misconceptions and questions regarding these results. Let’s take them one by one:
Question – Can these results prove that Elizabeth is Cherokee?
Answer – No, there is no test, anyplace, from any lab or vendor, that can prove what tribe your ancestors were from. I wrote an article titled Finding Your American Indian Tribe Using DNA, but that process involves working with your matches, Y and mitochondrial DNA testing, and genealogy.
Q – Are these results absolutely positive?
A – The words “absolutely positive” are a difficult quantifier. Given the size of the largest segment, 13.4 cM, and that there are 5 Native segments totaling 25.6 cM, and that Dr. Bustamante’s lab performed the analysis – I’d say this is as close to “absolutely positive” as you can get without genealogical confirmation.
A 13.4 cM segment is a valid segment that phases to parents 98% of the time, according to Philip Gammon’s work, here, and 99% of the time in my own analysis here. That indicates that a 13.4 cM segment is very likely a legitimately ancestral segment, not a match by chance. The additional 4 segments simply increase the likelihood of a Native ancestor. In other words, for there NOT to be a Native ancestor, all 5 segments, including the large 13.4 cM segment would have to be misidentified by one of the premier scientists in the field.
Q – What did Dr. Bustamante mean by “evidence of an unadmixed Native American ancestor?”
A – Unadmixed means that the Native person was fully Native, meaning not admixed with European, Asian or African DNA. Admixture, in this context, means that the individual is a mixture of multiple ethnic groups. This is an important concept, because if you discover that your ancestor 4 generations ago was a Cherokee tribal member, but the reality was that they were only 25% Native, that means that the DNA was already in the process of being divided. If your 4th generation ancestor was fully Native, you would receive about 6.25% of their DNA which would be all Native. If they were only 25% Native, that means that while you will still receive about 6.25% of their DNA but only one fourth of that 6.25% is possibly Native – so 1.56%. You could also receive NONE of their Native DNA.
Q – Is this the same test that the major companies use?
A – Yes and no. The test itself was probably performed on the same Illumina chip platform, because the chips available cover the markers that Bustamante needed for analysis.
The major companies use the same reference data bases, plus their own internal or private data bases in addition. They do not create PCA models for each tester. They do use the same methodology described by Dr. Bustamante in terms of AIMs, along with proprietary algorithms to further define the results. Vendors may also use additional internal tools.
Q – Did Dr. Bustamante use more than one methodology in his analysis? What if one was wrong?
A – Yes, he utilized two different methodologies whose results agreed. The global ancestry method evaluates each location independently of any surrounding genetic locations, ignoring any correlation or relationship to neighboring DNA. The second methodology, known as the local ancestry method looks at each location in combination with its neighbors, given that DNA pieces are known to travel together. This second methodology allows comparisons to entire segments in reference populations and is what allows the identification of complete ancestral segments that are identified as Native or any other population.
Q – If Elizabeth’s DNA results hadn’t shown Native heritage, would that have proven that she didn’t have Native ancestry?
A – No, not definitively, although that is a possible reason for ethnicity results not showing Native admixture. It would have meant that either she didn’t have a Native ancestor, the DNA washed out, or we cannot yet detect those segments.
Q – Does this qualify Elizabeth to join a tribe?
A – No. Every tribe defines their own criteria for membership. Some tribes embrace DNA testing for paternity issues, but none, to the best of my knowledge, accept or rely entirely on DNA results for membership. DNA results alone cannot identify a specific tribe. Tribes are societal constructs and Native people genetically are more alike than different, especially in areas where tribes lived nearby, fought and captured other tribe’s members.
Q – Why does Dr. Bustamante use words like “strong probability” instead of absolutes, such as the percentages shown by commercial DNA testing companies?
A – Dr. Bustamante’s comments accurately reflect the state of our knowledge today. The vendors attempt to make the results understandable and attractive for the general population. Most vendors, if you read their statements closely and look at your various options indicate that ethnicity is only an estimate, and some provide the ability to view your ethnicity estimate results at high, medium and low confidence levels.
Q – Can we tell, precisely, when Elizabeth had a Native ancestor?
A – No, that’s why Dr. Bustamante states that Elizabeth’s ancestor was approximately 8 generations ago, and in the range of 6-10 generations ago. This analysis is a result of combined factors, including the total centiMorgans of Native DNA, the number of separate reasonably large segments, the size of the longest segment, and the confidence score for each segment. Those factors together predict most likely when a fully Native ancestor was present in the tree. Keep in mind that if Elizabeth had more than one Native ancestor, that too could affect the time prediction.
Q – Does Dr. Bustamante provide this type of analysis or tools for the general public?
A – Unfortunately, no. Dr. Bustamante’s lab is a research facility only.
Roberta’s Summary of the Analysis
I find no omissions or questionable methods and I agree with Dr. Bustamante’s analysis. In other words, yes, I believe, based on these results, that Elizabeth had a Native ancestor further back in her tree.
I would love for every tester to be able to receive PCA results like this.
However, an ethnicity confirmation isn’t all that can be done with Elizabeth’s results. Additional tools and opportunities are available outside of an academic setting, at the vendors where we test, using matching and other tools we have access to as the consuming public.
We will look at those possibilities in a second article, because Elizabeth’s results are really just a beginning and scratch the surface. There’s more available, much more. It won’t change Elizabeth’s ethnicity results, but it could lead to positively identifying the Native ancestor, or at least the ancestral Native line.
Join me in my next article for Possibilities, Wringing the Most Out of Your DNA Ethnicity Test.
In the mean time, you might want to read my article, Native American DNA Resources.