a16z Podcast - a16z Podcast: The Scientific Revolution of Ancient DNA

Episode Date: July 13, 2018

with Jorge Conde (@jorgecondebio), David Reich, and Hanne Tidnam (@omnivorousread) Trying to reconstruct the deep past of ancient humans out of present-day people has until now been like trying to reconstruct a bomb explosion in a room from bits of shrapnel, says David Reich, Professor of Genetics at Harvard Medical School and author of the new book, Who We Are and How We Got Here: Ancient DNA and the New Science of the Human Past. But technological advances and new tools available only in the last few years have suddenly revolutionized this field, opening up an entirely new window into the past as well as our present humanity. This conversation, with a16z bio general parter Jorge Conde, and Hanne Tidnam, dives into this new scientific revolution of the study of the ancient genome. Beginning with the so-called "black hole" of Mitochondrial Eve to the most revelatory discoveries from new knowledge and scientific tools, this episode of the a16z Podcast delves into the ways archaic humans and ancient DNA tell us not just about our biology, but about ourselves. image: Ben Casey, Wikimedia Commons

Hi and welcome to the A16Z podcast. I'm Hannah and this episode is all about the science around the study of ancient DNA with David Reich, Professor of Genetics at Harvard Medical School, and author of the recently published book, Who We Are and How We Got Here, Ancient DNA and the New Science of the Human Past, along with Jorge Condi, general partner on the biofund at A16Z. The field of studying the ancient genome has exploded over the last five years in what's really almost a scientific revolution. The conversation starts with the so-called black hole of mitochondrial, Eve and how this new revolution is enabling us to move past that black hole, to what new tools make this ancient DNA factory possible, the most revelatory discoveries that this new knowledge has given us, both about ancient archaic humans, but also about modern humanity and the ways in which ancient DNA can tell us not just about our own biology, but about our history, who we are today, and even the struggles and challenges we've survived as a species. You talk in the book a little bit about the importance of mitochondrial Eve. Can you explain who this mitochondrial Eve is and what the importance of that is in this field?

So if you go back in time, take all the 7 billion or so people living in the world today, and find how many mothers those 7 billion people have in the previous generation. It will be only 1 or 2 billion. Go back another generation, it will be a smaller number. And go back a generation will be an even smaller number. And if you go back to the time when everybody shares the same maternal ancestor, that is mitochondrial Eve, the shared maternal ancestor of everybody. So we can sequence a bit of DNA that's inherited from mother to daughter to daughter to daughter, and that's called the mitochondrial genome. And when we sequence it, we can see of closely related everybody is on their mitochondrial DNA by counting the number of differences between each pair of people.

And so in this way, we can reconstruct how long it's been since everybody shares a common female ancestor. the best guess is about 160,000 years ago, and it's almost certainly that that person lived somewhere in Africa. Kind of like a black hole of genetic information. Yeah. So if you want to learn about the history of mitochondrial Eve's own ancestors and how she was related to other women live with her at the same time, you can't

because everybody is descended from the same woman and we don't have any samples from the descendants of those other women. Everything goes dark on the entirely mitochondrial lineage beyond 160,000 years ago, and DNA variation studies can provide us information about what happened before. So in some ways, the point of my book is to go beyond mitochondrial Eve. So why is it all of a sudden possible now to move past that black hole of mitochondrial Eve? What exactly has changed?

The ability to obtain large amounts of data from ancient human bones is really a newfound ability that really became possible only for the first time about eight years ago, as techniques got rapidly better and better. So there's now a type of data that's available for looking at the past and for looking at biology that simply wasn't available more than eight years ago. It's a little bit like the invention of past scientific instruments like microscopes and telescopes and when you use it to look at it, something that's never been looked at before, it tells you many things you didn't know about.

And what were the tools like before this? What was the lay of the land before we were able to analyze the genome? People have been successfully getting little snippets of DNA out of ancient bones. for 30 years. But what was being studied were tiny, teeny fragments of DNA that were so short that they couldn't really be able to use, in most cases, to say very much about the past. What's changed now is that we can obtain literally millions of times more data from each individual at far lower cost and access individuals who cannot be accessed before. Prior to this, people have moved around the world so dramatically and mixed so much, trying to reconstruct the deep past out of present-day

people as little like trying to reconstruct what happened after an explosion, a bomb explosion in a room from the bits of shrapnel and detritus. So if we go back, we started as it relates to ancient DNA looking at Y chromosome, we're more specifically looking at mitochondrial DNA. And now we're increasingly being able to essentially look at full genomes. For me, one of the really fascinating things is I've always assumed that DNA is a very wispy thing and the fact that it sort of survives in bone, I assume it's still highly degraded.

So how are you able to get to a reconstruction of a genome on what I assume are fragments of DNA? What does that explosion in a room reconstruction look like? What we get from the ancient genomes we have are indeed highly fragmented and degraded bits of DNA. And when we actually get DNA sequence successfully from ancient individuals, typically the fragments of DNA are really short about maybe 40 or 50 DNA letters long.

Now, that's tiny. That is tiny. And compared to the size of a chromosome, which will be hundreds of 100 million or so DNA letters long. So we need to reconstruct the sequences of large chunks of the chromosome out of these teeny tiny fragments. Now, that sounds like an incredibly daunting task, but modern sequencing technologies actually rely on very short fragments. maybe 150 or 300 DNA letters on. So while this is shorter than that, it's not incredibly shorter than that.

And what's done in modern genomes when we study a genome for medical studies is we take those 150 or 300 letter-long fragments and see what they line up to best in a reference human genome sequence. So we line them up against the reference human genome sequence, and we use that as a scaffold to hang these little fragments on. And we compare all these little fragments,

we obtained from the ancient specimen, and we then line them up with the reference genome. And do you use the same reference genome? We do use for most analyses the same human reference genome. For some sub-analyses, we'll change it for another genome. Why would you change it? How do you decide when to use those different ones? Well, the reference human genome sequence is a kind of collage of DNA from different people of known ancestries. So about 70% of the human reference genome is that of an African-American. who had about 50% European and 50% African ancestry. And then the remaining pieces of the reference genome,

the remaining about 30% comes from other individuals, some contributing more and some contributing less, mostly European and Japanese. And so it's important to know if the findings you have about biology and history are influenced by the ancestry of the person you're lining up to, maybe the person seems a little bit too Japanese or a little bit too European or a little bit too West Africa.

compared to how really the DNA looks. Yeah, that's really interesting. Now, your group developed a sort of different technique and used some different technological tools than others were doing at the time, and that directly led to some of your breakthroughs. Yeah, we did use some amazing technologies in new ways. I made a bet that it would be possible

to learn a great deal about history by studying large numbers of people who lived within mostly the last 10,000 years. So my laboratory is kind of a high, hybrid of two laboratories. One of them is Eric Landers Genomics Laboratory at the Broad Institute, where the focus was on large-scale genomics of medical patient populations. And that's where I learned how to make things efficient and to drive down costs and to think in terms of large numbers. The other laboratory that contributed to my work is that of Svante Pabo, who is the inventor of much

of the fields of ancient DNA technology. And I began working with Svante Pabébauds. in 2007, when I was invited to join the Neanderthal Genome Sequencing Consortium to help analyze the data, the focus was on trying to obtain the maximum amount of DNA from very precious, unique specimens like Neanderthals, archaic humans. And so we set up a kind of ancient DNA factory where we focused on driving down costs while maintaining quality, running many of our analyses on robots that can process up to 96 samples at once and implementing cost-saving procedures. PAYBOS laboratory had developed a technique to enrich DNA that was sequenced from an ancient sample from parts of the DNA they were interested in.

And we use this very intensively to only sequence the part of the genome from these ancient individuals that was informative about human history. And this has driven down the cost for studying ancient samples by at least 30-fold. So you've been able to extract significantly more information from any given sample of ancient DNA, where we were looking at a tiny snippet. Now we're looking at a more fulsome representation of the ancient genome of an ancient individual that's been sequenced. What's the X,000 dollar ancient genome? What were you able to drive the cost down to?

Where does industrialization get you in terms of cost? So the unit cost for studying each sample is less than $200 in terms of chemicals and equipment. Wow, that's really cheap. That's incredible. As of the last few weeks, we've obtained genome-wide data from about 5,000 people, typically data from up to about 1 million positions in the genome, and that is enough to support the great majority of analyses of population history. That could also be done even if we had the entire genome. What's the most ancient DNA that you've been able to successfully sequence? So the community as a whole, has sequenced some very ancient DNA. The oldest genome that has been sequenced is by Ludovic Orlando

in Copenhagen, a 700,000-year-old ancient horse from Alaska. And how about in your lab? In my laboratory, the oldest material that we've sequenced is about 40,000 years old. I loved the story of the finger, the little pinky bone, the Denisovan fingerbone discovery. Can you tell us the story of that discovery and what that learning was from this little tiny, pinky bone? That was the most amazing scientific gift I've ever had in my life. So Neanderthals were known very richly from the fossil record ever since 1850s when their skeletons were began to be discovered and recognized in Europe. And it was clear that they were incredibly interesting people who had brains as big as ours, bodies as big as ours, made really

sophisticated tools, and we now know had rather complicated culture. And of course, it would be very interesting to study their DNA and study how they were related to us. So in that case, it was a fossil record and an archaeological record in search of a genome sequence, and we all knew it was important to obtain a sequence and understand, for example, whether interbreeding it had occurred. So in 2010, Svante Paibo took me out to a beer garden and said to me, our colleagues in Russia uncovered a bone, a finger bone, a little pinky bone of a little girl. We know it's a little girl because its growth plates were not formed and the genetics tells us it's a little girl.

We stumbled on this bone with our collaborators in Siberia, and they thought it was maybe a modern human or perhaps a Neanderthal. And when we extracted it, it was incredibly well preserved. And the DNA sequence, when we looked at it, was not that of a Neanderthal. It was not that of a modern human. It was something else entirely, an entirely different human population we had no idea about before. That no one had seen. No one had seen, and no one had even expected.

So whereas the Neanderthals, I think of them as a fossil, record in search of a genome. In the case of the Denise Evans, it was a genome in search of a fossil record. Now we have to go out and find the skulls and the bones and the tools that people of this DNA type made. And we don't yet know what they were like. Is there a hypothesis as to why there hasn't been a fossil record found? Because you would expect that would be, you know, easier at a spot, obviously, than the genome. Well, I think it's almost certainly the case that there has been a fossil record found and it's staring us in the face and museums all around East Asia. So I think if you go to Southern in China or you go to Southeast Asia or you go to parts of Russia, you will see bones in museums

and storehouses and university collections everywhere. And I believe many of those will be from Denisovans. The Denisovans must have been as capable as Neanderthals in many ways of all the sophisticated behaviors we know Neanderthals had. And in fact, it's now clear from the genetics that there were other ancient archaic populations that were in a very distant way related to and with holes in modern humans, but distant altogether that were spread all over Eastern Asia. Has the computational work been done to determine whether or not if we are the modern human genome has denivisian DNA in it? Yes. So when we analyze the denizen DNA, we saw another amazing signal of statistical evidence of interbreeding with modern humans. People in New Guinea, in Australia,

and the Philippines have about three to five percent of their DNA derived. from Denisovans, archaic Denisovans who interbred with modern humans as modern humans settled these places. That's on top of the approximately 2% of the genome that these populations receive from Neanderthals. And so Denisovans have left a non-trivial impact on the ancestry of some populations in the world, and all East Asians have maybe a fifth of a percent of DNA from Denisovins, but there's very little in Europe. The Neanderthal, I think it's a fascinating example where there was a rich historical fossil record, and you describe it as, you know, a fossil in search of a genome. We have the genome now. We have the incredible findings that there's

been intermixing between Homo sapiens and Neanderthals. Is there anything in terms of the analysis of the Neanderthal genome itself that is aided and added to our understanding of who they were? There are so many things we've learned about the Neanderthals in their relationship to modern humans or a relationship to other Neanderthals just from studying Neanderthal genomes. One of them is that Neanderthals were of an incredibly small population size, marginal population size for a half million years since separation from modern humans, persistently low population size. If you look at across the entire animal kingdom at how low genetic diversity is, modern humans

are at the low end, but they're not off charts. Neanderthals are off charts. They were a population that must have been going through repeated population crashes, difficult environments and just hanging on and losing diversity all the time. Another thing that we can see is that the late Neanderthals, the ones that modern humans encountered sometime after 50,000 years ago when modern humans spread from Africa, the late Neanderthals really descend from another population crash,

one particular group of Neanderthals that then re-expanded. So one of the reasons I liked the Denisovan story so much was because it introduced this sort of surprise new factor that we hadn't thought about at all and in fact was hiding in plain sight all over the place. But can you tell us about a time where something changed fundamentally and turned on its head

our understanding of what had commonly been accepted as a major sort of historical narrative? One of the most amazing examples is prior to 2015, the assumption had been that once farmers got to Europe about 9,000 after 9,000 years ago, from Turkey, present-day Turkey, there weren't any major movements of people from outside, little movements of people, but not big ones.

We obtained lots of DNA samples from Germany ranging in time from 7,000 years ago to 3,000 years ago, and also DNA samples from Russia and Spain from the same period. And what we found was that in this period, beginning about 4,500 years ago, there was an absolutely huge population turnover, a massive migration from Russia, into central and Western Europe, where in Germany there was a minimum of 70% population replacement from people who came basically from Russia. And that these people kept sweeping west

so that, for example, 100 years later in Britain, they replaced at least 90% of the population. And the population that was just finishing building Stonehenge at this time, the last stones had just gone up, was at least 90% replaced by these newcomers. And these newcomers are the primary ancestors of people in Northern Europe today. That is astonishing.

So I'm just going to ask the challenging question, what do we learn about these archaic encounters for modern humans? If you kind of had to bracket out the biggest categories of learnings that the study of this ancient DNA gives us about not just where we came from, but who we are now, the kinds of information about humanity as a whole that we get from the study of this ancient DNA? Yeah. Yeah. I mean, in the last three years, I think it's not too much to say that we've come to an entirely new understanding. of the nature of human variation. I think it used to be reasonable, even as recently as 2015, to say that the people who live in each place in the world today might be descended from the first modern humans who got there in the past, so that the first spread of humans to America or to South Asia or Australia

or different parts of Europe gave rise to much of the ancestry of the people who live in those places today. But what we now know is that that's almost never true and that the people who live in any one place today are almost always the result of layers and layers of mass migration and population mixture convulsively replacing and changing each population over time and that there are almost no exceptions to this. I'm just going to read a little quote where you say you had assumed that the big genetic clusters of populations we see today reflect the deep splits of the past. But in fact, you go on to say, there was never a single trunk population in the human past. It has been mixtures all the way down. That feels like a very profound shift in how we think about even the basic understanding of a family tree, of a population tree. How do we start to begin to think about that when we think in those terms of mixtures all the way down? I think we have abandoned the idea of a family tree for thinking about human populations.

I think that that model is just so hopelessly wrong that we actually have to think about new mental pictures to think about our relationships to each other. A concrete example of this where we know best what's going on is what happened in Europe where people are called white or Caucasian because they're genetically pretty similar to each other and relatively more different on average from people in East Asia in other parts of the world. But if you go back in time 10,000 years ago with the time machine of ancient DNA and you study genetic material from people who lived in different parts of West Eurasia, we now know from ancient DNA that that region harbored four groups. Each is different from each other as Europeans and East Asians are today. And what happened between then and now is that those populations, none of them went extinct or disappeared. In fact, they all contributed substantially to people living today. And it's that mixture that produced the population we see today. We also know that those source populations themselves were the results of earlier layers of mixture. You also talk a little bit about using this science to understand more about the spread of languages, which sort of surprised me this idea that you could, you know, by studying DNA, then we understand more about how languages change. How does that work?

Well, you know, we all have DNA in us, but we also carry another heritage, which is the language we saw speak. So by reconstructing how languages are related to each other, for example, Portuguese. and Spanish are pretty similar. We can reconstruct relationships from shared languages. Linguists have identified groups of related languages, like these Austronesian languages that spread over the Pacific over the last 3,000 years, or Dravidian languages spoken in the southern part of India, or Bantu languages in Africa today. An interesting question is how these language groups got spread. And so one of the things we've been doing with ancient DNA as a community has been finding movements of people in the past that likely correspond to the spreads of languages

we see reflected today. You mentioned at the beginning that a lot of ancient DNA has been based on trying to essentially reconstruct what happened in history going backwards in time. If you could wave a magic wand and we had the genome sequence of every person on the planet that lives today, how would that impact your research in ancient DNA? One way that it would be amazing would be that it would make it possible to reconstruct in exquisite detail the movements of people that connect present-day people over the last few hundred generations. But as I mentioned before with mitochondrial DNA, information about the deep past becomes mudier and muddier the further you go back in time because it's cloaked by these great mixture events that have occurred. It's not possible to replace by just huge numbers of present-day DNA sequences, which is... DNA from times far in the past before some of these mixture and migration events.

So the timescale does matter. By co-analyzing DNA from the past and the present, we can learn some things that we would not otherwise be able to learn. But I think you can never use a brute force approach of present-day people to replace the ability of ancient DNA to study, for example, events that occurred 5,000 years ago. You've done 5,000 ancient DNA samples. I assume that number will continue to increase. What do you think is sort of the sweet spot number that will help lead, you know, ancient DNA into a second or third revolution?

So I think we are right now in a very Eurocentric phase where the great majority of samples, maybe 85 or 90%, are from Europe or far Western Asia. What we need to do is a community is to obtain ancient DNA with large sample sizes everywhere in the world where people have lived. Europe is a small corner of the world, not by any means the most important one. We need tens of thousands of samples from ancient people spread over time. And it really is like a voyage of exploration, a map building, an Atlas building project. It's a little bit like for Europeans who were discovering the world in the 1500s, parts of the world that other people had already discovered, but from the European perspective, there were these ships and expeditions mapping the ocean, the coastlines and the interior and rivers of the world, and they created these amazing maps. Here, we need to do the same thing with variation of people, not just in the slice of time that we have now, where human variation is so rich, but also 100 years ago, 500 years ago, 1,000 years ago, 10,000 years ago, 20,000 years ago. And with this movie of humans' variation over time in all different places the world, we'll have a far richer and better understanding of our path.

I think you describe it at the end as an ancient DNA Atlas of Humanity. I love that. You also talk about something you call the genomic signature of inequality in human history. Can you walk us through what you mean by that? What the problem that you're wrestling with is? Yeah. So this type of data we're studying actually amazingly captures the ancient signatures of past inequalities. So African Americans have most ancestry to enslaved Africans who are brought forcibly to the Americas from Africa by Europeans in the 16.

17th, 17th, 18th, and 19th centuries. And African Americans today have about 80% ancestry from West Africans and about 20% ancestry from Europeans with a lot of variation. But the European ancestry in African Americans comes four to one from the male side. So very little of it comes from European women and almost all of it comes from European men. What that's reflecting is the history of inequality in this country where European ancestry males would have children with African-American females, and those children would be raised as African-American. The genetics is giving you a genetic signature of that terrible history of inequality and slavery that occurred in this country. So that's an example we know about. If you look in people in India today, India's today

are the result of a massive convulsive mixture between two groups that occurred around three or four thousand years ago. And we now know that from lots of lines of evidence, including ancient modern DNA analysis. Now, that process was a very sex-biased event. And the people who have West-Eurasian ancestry related to Europeans and Central Asians and Iranians and near-easterners, more of the ancestry from those people was contributed on the male side than on the female side. And so that's telling you about an event of inequality. It's likely that the people of these two ancestries were mixing in unequal situations. That's really fascinating. How do you account for those in your studies? What are the signals that that's happening?

when there's something so far back that we don't know any of the history around it? Well, one of the most powerful types of ways we look at this is with the X chromosome. Women have two copies of it and men only have one. So it unusually weights the history of women as a two-thirds female chromosome. And so if you compare that to the rest of the genome, which is only half female, you can actually see what is happening more with the female ancestry than the males. So that's a really powerful tool. Another example is through analysis of Y chromosomes, which are inherited from father to son to son to son.

If you look at all the Y chromosomes of people today in most of the world, there are many, many shared male ancestors in each generation until about 5,000 years ago. And 5,000 years ago, there as a collapse down to a very small number of male ancestors of the great majority of people in Europe or in East Asia or in South Asia or in other parts of the world today. And what's very clear is what's happening is that this is a period when people for the first time began to be able to concentrate large amounts of power. Men with all that power began to have very large numbers of children and were able to pass down their high social status to their own male children who in turn had many children. And as a result, large numbers of people today in each of these regions descend from those male ancestors. If we were to roll the tape forward in human history on the order of tens of thousands of years, then the future it's possible. that we will be, you know, directly editing our DNA, whether it's through CRISPR or anything else.

If the David Reich born 30,000 years from now, if he were looking back in time and trying to understand ancient DNA, how would the fact that we can, we might be directly impacting our evolution, our genetic evolution through things like genome editing, would that break the lineage of what is studiable? Or does that just sort of require sort of a different lens of thinking about how we analyze something like this at the population level. Yeah, I think that that raises a number of different questions.

You can always trace the ancestry of the DNA that the person is carrying, even if the DNA has been edited. And so if there's an editing process or a kind of splicing in of one piece of genome into a person that came from another population or even species, we would just be able to trace that. And it would be a funny loop in the relationships, and you would see the impacts of it. Turning your question a little bit on its head,

I think that we are living in special times when medical care is allowing people to live who would not have done so before through in vitro fertilization or through having glasses despite correcting very bad eyesight. And as a result, it's probable that there are mutations building up that would not have been building up in the past due to the random reign of bad mutations onto our genome. There's an idea that the rate of worsening of the genomic health of human populations is maybe akin a little bit to global warming. It's not getting worse too perceptibly on the span of a generation or two, but over the span of a few hundred years, there might be very severe consequences. And so I think that we are in a time of profound

environmental and lifestyle change. Our lives, our own personal lives are too short to really appreciate the impact of that on our biology or even very much on our environment. But over a scale of a few hundred years, I think that these may and are likely to have some kind of impact that we don't fully understand. And study of ancient DNA of past populations in their genomic health in comparison to modern populations might be informative about that. Do you have a view on how we should collectively be thinking about the study of ancient DNA to help inform how we analyze genomic information for the purposes of understanding health and disease around the world? I think the area where ancient DNA will be most powerful for understanding health and disease is for understanding how

our species has changed biologically over time. So if we can sequence the genomes as we're now doing of hundreds of ancient farmers and huntergatherers and compare them to the genome sequences of people living today in those same places, we can see what changes have occurred have mutations that allow, that make people prone to diseases like diabetes or autoimmune disease or that change people's height or dietary patterns changed systematically over time in response to the environmental changes and lifestyle changes that people were experiencing. So by using the time series data that's provided by ancient DNA, we can track the experiment of nature that's occurred in multiple human populations around the world and understand how our biology as a species or as a population

reacts to profound lifestyle changes. And that might provide us with some insights. And that might provide us with some insights about how to deal with the new challenges we're facing as our lifestyle further changes today. That's so interesting to think of using this study to sort of unearth biological experiments over thousands of years. That's really a cool idea. Yeah. To me, one of the most astounding things about genomics, whether it's ancestry.com or 23 and

me, has sort of been this deep interest in the general public to understand, you know, their history. sort of not who we are and how we got here, but who I am and how I got here. How does that shift or change what I would learn from an Ancestry.com or 23 and me at the individual level? It's a little bit like one of these New Yorker cartoons of the world where, I don't know, you're sitting having dinner with your spouse, and you're in New York, and you see the towers of New York, and then you might see Florence on the horizon and Paris, and that's it, because that's all that matters to you. And you're not really aware of the complexity of the world around you. And so what Ancestry.com tells you is about your cousins and the people who are related to you in different parts of the country and some breakdown of your ancestral contributions. But it doesn't give you an understanding of the nature of humanity, how people relate to each

other, the deep sources of ancestry and mixture processes that produce people today. And so if you take this broader perspective that's not driven by your interest in yourself, but rather interest in the population as a whole, I think you can have a much more profound understanding of the past. And if you really want to understand the world around you, it might be good to try to think not about yourself, but about everybody. I like that very much. Thank you so much, David, for joining us on the A16Z podcast. Thank you.