In this NHGRI‑hosted panel presentation, Alice Popejoy provides a comprehensive scientific overview of genetic ancestry testing. She begins by situating human ancestry within the broader tree of life, emphasizing that human genomic variation is a tiny fraction of overall DNA diversity. Popejoy distinguishes between race (a social construct tied to hierarchy), ethnicity (cultural construct linked to community, language, religion), and ancestry (biological inheritance of DNA). Using a jelly‑bean metaphor, she explains the out‑of‑Africa origin of modern humans and the progressive loss of genetic diversity as populations migrated. Principal component analysis (PCA) is introduced as a tool that reveals major biogeographic ancestry groups (e.g., Sub‑Saharan African, European, East Asian, Oceanian, American, Near Eastern, Central/South Asian). She discusses admixture, highlighting African American/Afro‑Caribbean and Latino populations as prevalent modern examples. The presentation then outlines the three primary DNA sources used in direct‑to‑consumer (DTC) ancestry tests—mitochondrial (maternal), Y‑chromosome (paternal), and autosomal (including X). Popejoy contrasts distance‑based clustering algorithms with model‑based approaches for inferring ancestry, and details the reference datasets that underpin these methods (1000 Genomes, HGDP, UK Biobank, private company panels). A major limitation highlighted is the Eurocentric bias of publicly available reference panels, which reduces resolution for non‑European groups. She notes that DTC results can vary over time due to updates in reference panels, DNA quantity, analytical methods, and interpretive frameworks, affecting personal identity perceptions. Popejoy concludes by acknowledging ClinGen’s Ancestry and Diversity Working Group and NHGRI funding.
Chapters
Thank you so much for the introduction, Mr. Bonham, and to the NHGRI for this opportunity to present our panel. Today I am going to talk to you about genetic ancestry testing and what it actually means from a scientific perspective. So when we think about ancestry, I think especially in the modern era where we have these direct-to-consumer genetic ancestry tests, we think about ancestry as humans, but I like to start by thinking about ancestry Ancestry as everything that's connected and everything that's related by DNA. So this is a tree of life according to some bacteriologists produced in 2016, and I love it because it shows the vastness of genetic diversity across all species. And a few people might already know where we live as humans on this tree of life. It's right here on this tiny branch called Epistaconta, and we share it with fungi and trees. So relative to all life that is made up of DNA, ancestry is really relative, and genomic diversity is vast across species. So the variation that we have in our genome in humans is really small, and the Human Genome Project in 2003 showed us that all humans share 99.9% of our genome identically. But that is the variation that we tend to focus on in genetics, and particularly in human genetics, that's entirely what we're focused on is the tiny percentage of our genome that actually do differ among humans. So, and this has been studied for a really long time. I love this quote by Dick Lewontin in 1972. He talks about human variation and perceptions of difference. He said, what has changed in the evolution of scientific thought and is still changing is our perception of the relative importance and extent of intragroup as opposed to intergroup variation. And these changes have been in part a reflection of the uncovering of new biological facts, but only in part. They have also reflected general sociopolitical biases derived from human social experience and carried over into quote-unquote scientific realms. So this is what we see a lot in looking at differences within and between human populations, is this immense impact of society and sociocultural factors influencing our interpretation. Of the science. So there's no standard definition of race, ethnicity, and ancestry, and this is one area in which social, cultural values and ideas and identities really influence the science. And given that there's no standard definition of race, ethnicity, or ancestry, we're really— sorry about that, I'm getting Notifications, here we go. So race, ethnicity, and ancestry have been studied extensively by social scientists. I am not one of those, but I am a fan and follower of social scientists who work on these definitional issues in different contexts. But these are the definitions I use when I talk and think about race, ethnicity, and ancestry. That race is a social construct used to classify individuals into hierarchical tiers of status and power. And it's really based on an ideology of inequality. Whereas ethnicity is a cultural construct similar to race but different and distinct in that it's often linked to community, religion, language, etc. And ancestry is the biological inheritance of DNA which can be traced through the genome. So these are the definitions that I'm going to use in my presentation today. And if you see the Venn diagram on the right here, you get a sense of really how overlapping these terms and these concepts are. I like to point out Asian as one example because it sort of fits in all three of these categories. If you think about Asia as a continent or as a sort of geopolitical construct, that's more than half the world's population, so not particularly useful from a biological standpoint. But you could think about Asian racially, right? You can make a stereotypical statement, "That person looks Asian," or you might say that I had Asian food, which is sort of an ethnic reference, right? Cultural nutritional reference to some sort of ethnic group. Whereas ancestry is really about the inheritance of of DNA from ancestors that came from Asian countries. So it can be used in all of these different contexts, and all of these terms are sort of overlapping and confusing, despite the fact that we use them all the time in science and in data collection. So let's focus on ancestry as a concept, since that's the focus of our talk today. And why is it that we can actually detect differences in our ancestry among humans, given that so little of the genome actually differs within the species? And the reason for that is that we have a single origin of modern humans somewhere in Africa. Population geneticists tend to differ in where precisely in Africa they believe that original human existed, but it's generally accepted that humans, modern humans, were originated in Africa and then spread out across the continents over time. And I like to think about the dispersion of genetic variants across the globe in terms of a metaphor. Explaining this is that you think about all the genetic diversity that existed in that original population of modern humans as a giant jar of jelly beans with all of the colors and all of this variation. And then you sort of randomly take a handful of jelly beans and you throw them into the next continent, and then you take a handful of that subset of jelly beans and throw it to the next one. So the further away you get from Africa, you get less and less genetic diversity. You have fewer colors of jelly beans, let's say. But you can in fact detect where some of these ancestors came from by sampling those genetic variants that are typical of different ancestral populations. And now some researchers have used an analysis tool called principal component analysis to determine what the major biogeographic ancestry groups are across the globe. So you see American in green, and this refers to pre-colonization Americans. So we might think about this as American Indian. This includes North and South America. Then you have East Asian, European, Central or South Asian, Near Eastern, and Sub-Saharan African. And then bottom right, you have Oceanian. So these are the basic biogeographic ancestry groups based on that dispersion of those genetic variants across the globe. But that doesn't represent everybody, right? Not everyone comes from one of those populations. In a lot of cases, there is what's called admixture, and I'll talk about that more in a second. It'll be important for Dr. Rothchild as well, that there's some sort of lineage in the past, and then those populations diverge as they go away, and then they come back together and they create what's called an admixed population, or one that has ancestry from different biogeographic ancestry groupings. So in this paper by Hudart et al. in 2019, the two admixed groups that are also highly representative of modern populations are the African American/Afro-Caribbean group and Latinos. So this is ancestry admixture, where you have an ancestral lineage, as I mentioned, the population split and then they come back together. And so over time you have this admixed population where the genetic makeup is a combination of two different ancestral lineages. So how do you actually get an ancestry test from a direct-to-consumer company? What does that test entail? Well, there are different data sources that you might use to do your test. One is a mitochondrial DNA test, so you're testing for your maternal lineage. You could do Y chromosome testing where you're looking at paternal lineage. And then there's the autosomal DNA, all the other chromosomes and the X chromosome. Now there are different algorithms that different companies and different research teams use to infer ancestry or estimate it. You have distance-based clustering algorithms where you're actually looking to find those clusters or those groupings of individuals based on their genetic variants, based on the differences between them. There are also model-based algorithms where you're basically coming up with an idea and you're testing whether that model fits your data or not and what that looks like. And then you have reference data. As I mentioned, ancestry is relative, so you have to have something to compare somebody's DNA to in order to give them an estimate. So where do we get the reference data from? Well, you have publicly available resources like 1000 Genomes, the Human Genome Diversity Project, the UK Biobank, and others. There are often private or other proprietary collections that people use. And then, of course, direct-to-consumer companies actually have data from their customers, which they feed into their algorithms and use as reference data to improve their estimates. But the problem is that most reference data, especially those publicly available resources, come from research. And the majority of genetics research has been done on people of European ancestry. I did an analysis in 2016 that showed we have some improvement in diversity in genome-wide association study samples. Samples since 2009, but unfortunately, the majority of that growth in diversity is due to studies in Asian countries conducted at Asian institutions, and those data are not often made publicly accessible. And also, this publication was in Nature, which is a UK-based journal, and most GWAS are coming out of Europe and the United States, so it doesn't really make sense to include the diversity of Asian samples that are being studied elsewhere and not necessarily included in our reference datasets. In this part of the world. So there are some limitations, right? One of them, a huge one, is that reference panels lack diverse representation. So if everything is relative and you're making these comparisons, you don't have the ability or the granularity in these non-European populations to give you accurate and very specific ancestry results. Also, different companies and different teams use different methods for testing genetic ancestry. So these tests differ on the type of DNA that you're using, the algorithm that's being used to analyze the data, the reference panels that are used to estimate the proportions of ancestral populations, and then interpretation. It's subjective and it's personal. So what does that actually mean when somebody— a company says you're this percent this? What does that mean and how does the result impact our identity? I'm really looking forward to hearing Dr. Roth address that very question in her talk. So why does ancestry change over time? Well, there's 3 main reasons. One is the data. The reference panels can be added or changed or updated over time. The type of DNA that's being analyzed or the amount. The method that you're using. Sometimes companies change methods and then all of a sudden they have new estimates. It doesn't happen very often, but they're often tweaking or adjusting methods as well, which might change the results. And then interpretation. Companies might update their interpretation of your results based on new information or new data that they have. And then also your own personal interpretations might change based on new information that they're being presented with. So with that, I'd like to thank my funders, and I'm the co-chair of the ClinGen Ancestry and Diversity Working Group, and the NHGRI funds ClinGen and is hosting this panel. Thanks very much for having me.
