 Human genetic variation, and in fact, I'm sure most of you did it when you came in the room. You looked around, you said, are there any relatives here? Anyone that's a threat? Anyone that's a mate? Even some of you did that that won't admit to it. But behavioral genetics is a different topic. And what I, my goal today is to talk specifically about kind of the nuts and bolts of human genetic variation. To answer some of the questions, they just came up from the back. And it, starting with this as kind of a segue from Emily's talk. Anyone here from this publication? I use this because it actually highlights a little bit of human genetic variation and that there's probably some genetic variation between here and here. You'll note that there's practically no variation between hair color here. And that's probably an environmental influence, I would guess. I don't have any firsthand knowledge or anything about that. So if you look around the room, you look on this slide, you'll see that humans come in a lot of shapes, sizes, colors. And what I'm going to talk to you about today is not how that happens. Because I can tell you that we don't know it all, the whole genome association studies will help us start answering that question. But I'm going to tell you about what the raw material is in our genomes that produces these diverse phenotypes. And not spend much time on the commercial products that produce hair color. So these are the topics, the three that I'll cover in the last one is the one we're talking, we're here about today. It's a human genetic variation. What is it? Where did it come from? How is it scored? And we're here today to figure out what use it is, especially from the biomedical realm. We will dip a little bit into anthropology because there's a lot of ancillary uses to these kinds of data sets that we're generating. So genetics, as most of you know, starts here with Mendel and his peas. And if you look on this slide, you'll see there are peas that, at least from up here, don't look like they're yellow and green, but they come in two flavors. And this is a product of a cross where you get essentially peas of two kinds. Now, as most of you know, humans are not peas. If you look at the peas in here, it does bring up one human analogy which is peas in a pod, like peas in a pod. And so these humans are like peas. They're identical twins and they share exactly 100% of their DNA, at least initially. So in this case, if we have 100% of the DNA, it's the same. We produce traits or phenotypes that are very, very similar. It's a rare occurrence, but most of you probably have met identical twins in your life. I actually lived next door for about a decade to a pair of identical twins. It started when they were four years old, so I saw them from four to 14. And it was amazing how much over that period, when they were four-year-olds, I could not tell them apart. They finished each other's sentences. They dressed alike. But by the time they were 14, they started to separate. And so to underscore what Emily said, as we age, our environment starts to influence our genetic. Our genetic kind of sets the stage. Here's another pair of individuals. Obviously not the same, although from the point of view of accomplishment in their fields, they're probably somewhat equivalent in being at the best of the time. They obviously don't share 100% of their genes. They do in some way share some of their environment, but I can tell you that my dad went to school with one of these guys. And it's not the short guy like me. It's actually the tall guy. And even though they had shared environment, they didn't turn out the same way. So what explains this? The other take-home message, and I forgot to tell you, there's a quiz during this talk and some homework. So if you're typing your assignment, wait till I give it to you. Let underlies that. In fact, in human beings, we're not like peas. And almost all genetic traits, in fact, I challenge audiences of geneticists to come up with more than five Mendelian traits. So things that are inherited just like the peas, either green or yellow, that occur in humans that aren't diseases. And I won't ask, that's not the quiz. The quiz is actually easier than that. If you try to think of human traits that are influenced and inherited just like Mendelian disorders in humans, there just aren't very many. Most of them look like this. And so this is a snapshot taken 1914 of British military recruits lined up in height order. And as most of you who have had statistics at some point can kind of see, it's roughly a bell curve. And this is what human phenotypes or traits look like, is that they come in a range of flavors. And we know that if you were to, and let's just assume you could do that, cross this person and this person, the tallest and the shortest on this picture, you would not get offspring that were all shorter, all tall. What you would get are offspring that show the full range. And so what happens, the reason we get this is two reasons. One, there are little variants in the genome that have small effects and then there are also a lot of them that kind of add up to this kind of distribution. And so now I'll talk about exactly what they are. All of this that produces what you are is encoded in all of your genes, your genome. So this is a cell, this is the chromosomes from the cell just kind of spewed out onto a slide and this is them in size order. So all of the information you need to make you is stored in these chromosomes. By way that's about half protein, half DNA. We're going to talk about DNA as information part of it now. And so here's your homework assignment. This is what DNA looks like. It's just four letters repeated roughly a billion times or so depending upon what organism you are. But the entire story is told with four letters. So the homework assignment is your next story. File it using only four letters. You can choose whatever you want. So you have 26, if you're writing in English, you have 26 to choose from. This is what codes us. And if you look at the information on the screen, it doesn't really tell you a huge amount. The other important thing about this system is that it self replicates. And so this is another amazing property other than the fact that you can convey all the information to make a human being using four letters. It actually has enough information to copy itself. And that brings up the one feature of DNA that we're going to talk about today in that DNA replication is essentially almost perfect. And some of you probably had to take a information theory class in college at some point. Information theory tells us that if we have a self replicating system, it can be no better than making one mistake per replication event. And in fact, DNA and its self replicating actually approaches close to that theoretical limit. Every time a DNA molecule replicates, it makes a couple of mistakes. So every time it copies itself, it makes one or two mistakes out of a couple of billion. Now it does that by a complicated system that Emily kind of talked about. It's not only is there people who type, there's copy editors who take out the mistakes as they go, but the result is some mistakes are made. Now mistakes aren't necessarily thought of usually as a good thing. The reason I say luckily they've been made because imagine primordial organism that first started encoding itself in its information in DNA. What if it never made a mistake? We wouldn't be here because the mistakes lead to changes in DNA and raw material for selection. And these kind of mistakes or errors in self copying kind of come in three flavors, the good, the bad, and the indifferent. And I will tell you now that we tend to focus on the bad because those are the ones that cause disease. But almost 99.9% of all the mutations or variants in our genome are essentially indifferent. They don't matter. So you could imagine if you were typing on your laptop and you had two keys for the letter T, and sometimes you hit one and other times you hit the other key, it wouldn't really make any difference. It wouldn't change the meaning of the word, it wouldn't change what happens. Most of the errors in DNA come in that flavor that they don't make a difference. And that's good because we actually have a lot of them. It also means that we all walk around with differences. And so that everyone in this room is different from everyone else in this room because of these changes in their DNA. So what I'd like to do is to stage as to how these happen. And this is the only math in the talk. But think about a new mutation. They don't happen by direction, they happen by chance for the most part. And so what happens if this is a population and this is time measured in generations so your distant ancestors are here and today we're about up here. And this is the frequency of mutation. All the mutations that we have for the most part started in one person. And when that happens and they pass it on to their children, that child can either pass the mutation on to their children and their children and so on. Or what happens most often is the mutation might get up into the population at some frequency and almost all of the mutations then die out of the population, the frequency goes back to zero. Occasionally mutations, so if this was one variant here, say this was a C at this position, over time it could be completely replaced by a T. And so that we would all have, this could be a very successful variant and in fact it could have some advantage so that everyone who had a C at this position, if I come back years later they know I'll have a T. What happens more often is that these things just kind of drift around a bit. And so if we come in today and ask what do we see in the genome today, so we take a slice through it, we see variants at all different kinds of frequencies. If we were to come back a long time from now in our population because it's so large, those variants would probably still say at the same frequencies. If we go and someone was asking about isolated populations, in isolated populations because they're small, the variants can change frequency a lot more readily. But this is what we look at when we're looking at variants in the genome. We're taking a cross-section looking at things that are currently variable. As Francis mentioned they come in two flavors for the most part. The SNPs are single nucleotide polymorphisms which we're going to spend most of the time on today and put duplications, deletions or segmental variants in your genome on your radar screen because we think they might be important but we don't necessarily test for them very well and we don't necessarily understand what they do yet. So these are cases where instead of having the same address but different color doors, in this case there's whole blocks, city blocks missing. And what we've discovered over the last couple of years is that people in this room might have genomes that are different sizes. Some of your genomes might be missing, thousands of kilobases of DNA, some might have maybe not thousands but hundreds of bases of DNA and others might have more or less. But today we're just going to focus in on single nucleotide polymorphisms because that's what we're using in these studies that you'll hear a lot about. And why are we working with single nucleotide polymorphisms? The first you already heard about, they're numerous and I'll tell you how numerous the second. The second is that they're stable and I'll talk about how they are stable and how stable they are. The third and probably the biggest reason is they are very easy to score using carrying technologies and I will show you some of it. There's actually show and tell in addition to the quiz and the homework. I'll show you some of the technologies that are used to score this. And the third is that sometimes they're actually in genes and we can interpret them. So this is what a SNP looks like. It's not very fancy. Here's, let's pretend that this is Larry Thompson's DNA, this is my DNA, this is Terry's DNA. If we were to sequence along the same stretch in the chromosome, most of the bases would match. But occasionally we would hit variants where, and I think this was me, where I have a T, Terry has a C. Over here you can see, was this you Larry? This is Larry, he has an A, the rest of us have a G. These are simply nucleotide changes that happen due to this replication error, lack of DNA repair. They are all over the genome. In fact, if you compare any two chromosomes and all of you have for each part of your genome, you've inherited two copies, one from your mom and one from your dad, those chromosomes will differ about one every thousand bases. And so I don't know at what rate errors creep into the columns you write, but I'm sure that your copy editors pointed out to them when you make one. One in a thousand, you guys probably do better than one in a thousand. But our genomes have these errors, sorry. If you compared more chromosomes there'd be even more variants. So it's not that every fixed thousand bases, so if I count out to base 500, there might be a variant there. That doesn't mean the next one's not to 1,500. There could be another variant over here that is just in different populations. So the more, the more chromosomes you look at, the more variants are possible. And so it's been estimated that there could be 20, 30 million different variable sites just out there in the population. And not all of them are common. And when Terry will discuss how we use these in studies, we're going to focus mainly on the common variants. This is the quiz. So the concept of linkage disequilibrium, which I agree is an awful name and a hard concept, is also linked with this other awful name and concept which is called haplotype. And it's actually hard to describe but easy to understand if we do it in the form of a quiz. So shown here is this stretch of a chromosome and shown here is the stretch of chromosome from someone else. And in that case, if I know they have a T at this position, if I look downstream and I see an A and I realize I look at a lot of people and see every time I see a T here, I see an A here. That means that those two are in disequilibrium. They don't have to be, but they are because we've observed them. And so here's the quiz. What color socks do I have on? I can't tell. Christine, what color socks do I have on? What color sock do I have on my left foot? Maggie? That's linkage disequilibrium. Once you know through probability that if I have a gray sock on, on my right foot, I'm most likely to have a gray sock on my left foot, will that hold true for all populations? I have an eight-year-old daughter. And in her population, she often goes to school wearing two different color socks, which is a relief in the morning and maybe fashionable, I don't know, but at least we don't have to find matching socks in the morning. So the degree of disequilibrium you see between markers can vary between populations. I'm going to discuss a little bit why, due to the population structure, it actually does vary. But simply it's a probable ability statement that if I know what's here, I can guess what's here. Now some of you may be hardcore investigative reporters and you'll have to wait until afterwards to look at my left foot. But most of you will take it for granted that I have a gray sock on my left foot. And that... Oh, you guys don't ask each other. You can independently verify. Yes? How far away in the stretch of DNA does the other one have to be to count? It's a good question. The closer they are, the more likely they'll be the same, and knowing one, you'll know the other. As you get further away, that relationship becomes less certain. In human genome, in most populations, that can extend from hundreds to thousands of bases so that if I know where this T is and I look in a specific population, I may know exactly what all the snips are, actually a wrap around the room, almost a thousand bases. And we don't have to guess because that cover with the fancy diagram that Francis showed you is the hat map data, which we've gone out and looked at several major populations in the world and mapped this relationship already so that we can pick the markers we want and they will report. And it varies a lot. There are some places where, maybe not on this scale, where an A here would not be associated with a T here because there's been ancestral recombination, we call it. So it varies. We actually had to go out and map it in the genome before we could use it. And that was the hat map project. Now that we have that, we could say, if we're in an area where the distances are very tight or the relationship is tight, we'd say, we only have to genotype this one and then we'll know that one. Other areas we'd have to put markers in much more densely. And that's what these tools that I'll finish off with actually do. They've taken this haphalotype relationship into account. I want to diverge a bit for the last five minutes on where this stuff came from, which will have some bearing into the issues that Vence is going to touch on and does, hard fact, a little bit to the hat map. As humans, we're actually a relatively young species or population. We tend to think that modern humans came from a core population that existed about 100,000 years ago. It probably was a relatively small population for long periods of time. We use this awful term, effective breeding pool, which I don't know whether you're in the pool or not in the pool. It sounds awful, but because the populations were small, that allowed different things to rise to frequencies using that graph that I mentioned. And then at some point we got much more stable food supplies, coincident with agriculture. Our population kind of exploded. And things changed much less readily in large populations. And so now we've reached somewhat of a stable degree of genetics except for these smaller, isolated populations. And again, we're coming in here, and this process I'm showing here to go from this point, from zero frequency to 100% frequency, we know statistically takes about four times the effective breeding population, which in humans would be about 40,000 generations, which means it would be about a million years to go from zero to 100% for a particular variant. So if we're looking at intermediate variants, some of them may have been around in the population for a half a million years or 100,000 years. So these SNPs that we're genotyping for these assays have been around in our population a long time. And that gives rise to the prediction that you would find them across the world, which in fact you do. And the reason you do find them across the world is because if you go back 100,000 years ago, all of our ancestors lived in Africa. So the pool of all human genes existed in Africa. Now remember, the SNPs are older than this. So almost all of the SNPs that we're going to look at were present before we migrated out of Africa. Well, and what these colors show, and this is a slide bar from Ken Kidd, what these colors show you is range of genetic variants. And so Africa is the source of all the genetic variants we have. And so you'll see lots of different colors. Now what we think happened is that a subset of this population, a small subset, left the continent of Africa. And what happens, remember that lesson, if the subset is small, you might have more rapid changes in frequencies of variants. And then eventually this population spread throughout the world in different continents. And what's shown here is that the variants, although the variants are all the same, the frequency of particular variants may vary across the globe depending upon what we call continental origin. And this explains why we see different frequencies of different variants in different populations. It has nothing to do with evolution, selection, probably 99% of the time it's just chance and who happened to leave Africa. And one time I showed this slide and someone said, well, these people must be smarter because they knew to go this way instead of that way. And if you think about 100,000 years ago, this is random wandering. This has nothing to do with selection. There are no GPSes 100,000 years ago. And so this out of Africa theory kind of explains the spectrum of genetic variation and the frequency differences in variation we see throughout the world. The last thing I want to finish up on, and Emily, if you could pull out, there's a plastic bag in my bag right down there, is... Nope, not that one. I have more than one plastic bag. That one, yeah. Is how we score this thing. And Francis mentioned the technology that has changed over the last five years has been amazing. So I'm going to pass... These are glass slides that can break. Usually with high school kids, I warn them about safety, but for you guys, I don't. The other thing that will be going around and you can just spread them out are chips. There's some glass in there, but it's in plastic, so I can trust you. What I'm passing around are examples of the platforms. You can take the slides out that are in the box. If you look at those slides, which are shown over here, or those chips, which are shown here, there are two different systems for doing SNP genotyping. And within a year, both of these systems will allow us to do one million SNPs in a single experiment. And for those of you that have been covering the genetics beat for a long time, that may be stunning to you. It's even more stunning to me, because we used to look at them one at a time and it used to take years. So in one experiment, what we would do is either wash it over the slide, or if you look at those DNA chips, squirt it in the back, there's some holes in the backs of those chips. In both cases, there are things that will be on the surface of glass where the genotypes will be scored. And in both cases, and these are two different platforms, and I don't have any, in fact, I'm not allowed to have any affiliation with any of these companies. One of the plastic encased ones with glass slides are from a company called Illumina. In both cases, what we do is we wash over the person's DNA, and the chip has specific features that will tell us what the genotype is. Do you have a C or a T here? And in the end, in both cases, what has happened is whether you have a T or C is highlighted by a fluorescent molecule, and a camera and a microscope objective takes an image of the chip. And so in both of these, what's generated is a giant image file. And so for those of you who do digital photography, you've seen the megapixels go creep up and up and up. The data for these experiments are giant image files, which are then analyzed. And Terry will show you some of the crunched numbers. But what is actually analyzed is how bright these spots are. And for those of you that use laptops, I'm sure most of you know they've gotten smaller, better, faster. All of the genotyping's gotten smaller, better, faster, cheaper, but it's also been coupled very tightly with advances in computational biology and storage. The raw data for a project of 1,000 cases, 1,000 controls that Francis just mentioned in these image files covers about a terabyte of data. And when I first heard the word terabyte, there were a few people in the country, a few places in the country that could deal with a terabyte. Now you can buy this portable, not disposable, portable hard disks that'll hold a terabyte for a couple hundred dollars, which is pretty amazing. Both of those things have pushed this whole field forward. And so I'm happy to close with this and remember the homework, we got the quiz right, and I'd be happy to take questions. Do both the chip platforms have the same snips on them? Or would you find something different in an mathematics chip from an aluminum chip? And also... Chips on them. The same snips on them. Sorry. I'm sure there's some overlap and I haven't looked at the data files. People can get this and compute comparisons. They're slightly... The problem is that just like software, there's different versions all the time. So the working versions of the aluminum platform, the working versions, the commercially available ones now were designed slightly differently to cover areas. There are also issues in how you design the assay that will event one from working in one platform and will work on the other. And one more follow-up. It's Bobby Ergoth. Who... These chips used to be used years ago mostly for expression levels. Who first said, hey, let's use these for, you know, whole genome associations. Was that the chip company's idea? It just kind of happened? Or did you guys or someone in the generalist community says we really need you guys to do this? The chips, especially the apometrics chip, was invented to determine sequencing originally. Now, it failed at that. In fact, it doesn't work very well to determine new sequences. But it can determine re-sequencing. And so one of the things we did, and a lot of this is done, we talk to these people all the time and there's public and private official partnerships as well as unofficial partnerships. They knew that there would be a real need to do this. And these are the two platforms that have risen at least now to the top. There's other ways of doing synaphenotyping that have been out throughout the years. But the ability to synthesize in the api case or in Illumina to do this combinatorial process on very small surfaces has really been the big plus. Hi, it's Joe Palcott, NPR. I wonder if you could go back a couple slides to the one that's titled, Fate of New Mutations. That's so pretty. Okay, so I take it the arrow is sort of, you can think of that as today. Yeah, in this case. The problem that I've been trying to wrestle with, and I know this isn't particularly the focus, but since you brought it up in your talk, I'm going to ask anyway, you've got data up at the top of the slide that says, well, by fiat or by experimental evidence or by something, the data about the change of the mutation rate. But I'm still confused about how you infer backwards to what was from what you see in a single population today. I'm not sure what you mean. Well, you're making some assumption about, for example, in the next slides about where people moved to. You're making an assumption about how long ago they left Africa based on the frequency of certain mutations today. And I still find that, I mean, I can get it, but every time I try to come up with a good analogy for it, I get a little lost. So I was wondering if you could try to come up with one. Part of the problem is that what really is very important, it's based on modeling, and people model the different scenarios and come up with something to stop them. And what's really important to know, which we don't know, are these population histories' effective breeding sizes. They're essentially estimates. And these conclusions are drawn in conjunction with archaeological anthropological evidence as well. So from the analogy point of view, how do we know that out of Africa happened just looking at mutations or looking at variants in this case? And if we want to talk, I use mutations and variants interchangeably. There's a slide in there that discusses the formal definition of both. And I think the easiest way to do it now is with looking at hat map type data in that, and this gets to the question from over here, in that the areas of disequilibrium, if you go in and look in African populations, they tend to be smaller, implying that they've been there longer. So if, and if I go back to this slide, what occurs when these two mutations are next to each other over time due to reproduction and recombination is they get separated eventually. And the more generations that happen, the more chances that an event will separate them, so this will get recombined and be on this chromosome now. And so when we look at African populations, the areas that we see of disequilibrium are actually smaller, implying that there's been more time to have them divided. When we go out to European populations, then we go to the more isolated populations. We find the area of disequilibrium gets larger and larger, implying that they've been separated in shorter amounts of time. Does that help? No, it's not a... I mean, it's definitely, I mean, it's helpful conceptually, but in terms of trying to... I mean, without using the word disequilibrium, I'm trying to come up with a comfortable analogy because I've been trying to explain the genetics of population movement through Europe, for example. And at some conceptual level, I get several steps down the road and then I get to linkage disequilibrium and I see ice glaze over. So I'm hoping at some point to come up with the perfect analogy, and I'm not sure I've found it yet. It's actually one of the things that we'll talk about at lunch because I think the language in this field is... I hope my nice colleagues will forgive me, but it's dreadful. And it's really, really hard for me to even talk to my mother about it. And she knows what she's been listening to me for years, so... Thank you for that, Vodacamp. What's your life in the already given example? End up in the dryer? Sharon Bagley from Newsweek. Does SNP analysis generally find the variation in both copies of a particular chromosome? And similarly for a whole genome analysis, does that generally run both chromosomes and does it matter for an estimation of the increased risk of a disease? Yes, the answer is yes, yes, and yes. So when we do these analyses, we take a blood sample from your finger and we isolate DNA from it. They're no longer in these chromosomal packages. They get all broken up. So your maternal and your paternal copies are all kind of mixed. And when they get put on this chip, they're all mixed. And it's very important that we query both flavors and there'll be a spot. We won't know which one came from which unless we have your parents, too. But it does answer... It is very important that we get both copies recorded. There are some cases where there can be interference and we might miss one and the people who produce these platforms have gone out of their way to avoid those issues. And yes, it's very important to know both in the whole genome association that we're scoring broken copies. It's not as important... This is the breakthrough in the past analysis of how to study animal husbandry by pulling people from the population no longer in animal studies. What we're doing is combining this deep relationship of human families when we do this whole genome association study. So we don't need to know parental copies when we're just studying them individually. In fact, they shouldn't be related. So yes, Sharon, that's a really important question because I think sometimes the way the data analysis can be confusing. So each SNP has two possible spellings. Let's say a T and a C, two alleles, as we say. But because we all have two copies of everything, unless you're talking about X and Y, that means there are three possible genotypes. You could be AA, you could be AT, or you could be TT. Some people, when they do the analysis in a whole genome study, will simply say, is the T allele more common in the affected than the unaffected, and therefore it's a risk? And some will instead say, do you see TT individuals, homozygotes, more commonly in the affected than in the unaffected. So they're looking at genotype instead of just allele frequencies. And sometimes the result can be sort of tweaked a little bit by which of those methods you choose. One is more looking at a dominant effect. One is more looking at a recessive effect, actually. And I think when one looks at the details of those papers, it's probably important to see exactly what was it that they claimed represented that association. Was it just the allele, or actually the genotype? Yeah, David Brown from The Post. My question is a little bit like Joe's, if I understood his, and that is that the... What confuses me is that when sort of driving the relationship between a founder population and a descendant population, there seems to be the suggestion that one of them stays the same and one of them changes. But of course, for any given locus, the founder population in Africa has been breeding, and thousands of years have gone on, and it has just as great a likelihood of that locus changing as the ones that went out and founded another population. So I guess my question is, do you always need at least two and many different ones in order to get the sense of stasis in one population and change in another population? I think in the context of these kind of association studies, it doesn't matter so much. In the context of reconstructing population history, it doesn't matter as much either because these snips are older than the populations. And so there is a rate of change, but the rate of change is that the introduction of new snips at high frequency is slower than the migration of the populations through modern human times. So it doesn't really factor in that much, and that all of the variants, and there's two things that I want to... There's two nuances that I want to make sure that you get, is that we oversimplify a point to Africa and say it's everything. There is a microcosm of lots of different populations, and one that we haven't plumbed a lot of what the variation is. We just had some evidence that a lot of the variation came from. The other extreme is the homogeneous populations that you asked about earlier, that homogeneous is maybe a bit of a strong word as a member of one of these homogeneous populations. I know we're not all the same, and there's just chunks of our genome that we tend to share that you can see when you look by standard. But I don't think the changes that have happened over the last 100,000 years on the SNP landscape, and we used to do things on microsatellites with which change over that area. On the SNP landscape, they just don't change that much, so that if I find something where I find variants in Japan, and I find populations that are T or C here, if I go to Africa, I'll find that there's people in Africa that are T's and C's in there. So I'll do a lot of them. When you do these genome-wide studies, are you looking at the nuclear DNA only, or the mitochondrial, or both, or what? It's a good question. So as far as I know, the SNPs, and maybe Francis or Terry know, I don't think the mitochondrial DNA is on most of the SNP chips. We're missing 17 kilobases, and an important 17 kilobases out of 3 billion. The problem is there's a little bit of a mismatch. There's about 1,000 copies of your mitochondrial DNA in each cell, and only two copies of your nuclear DNA, and so they would have to balance the probes in some way. But there's no reason they couldn't put them on there, especially because the variants in mitochondrial DNA are well-described, and there aren't very many of them. Now, some of them are length polymorphisms, which makes it harder, but for the most part, we'll serve a nuclear DNA.