25 years of the Human Genome Project

0:01You stopped waiting for permission and started building yourself, your dreams, your credit. Experian Boost could turn your on-time payments into instant FICO score gains. Every bill you paid could become proof of your power. Self-made? Absolutely. Self-funded? Experian could help you get there. Download the Experian app today. Results will vary. Not all payments are Boost eligible. Some may not see improved scores or approval odds. Not all lenders use credit information impacted by Experian Boost. See Experian.com for details.

0:49Hello, welcome to the Naked Scientist podcast, the programme that brings you the biggest breakthroughs and also talks to the major movers and shakers in the worlds of science, technology and medicine with me Chris Smith and this week 25 years of the Human Genome Project we've just passed the anniversary but what is it and what has it achieved over the last 25 years from Cambridge University's Institute of Continuing Education this is the Naked Scientists

1:27Well, it's been a quarter of a century since the first draft of the Human Genome Project was unveiled, a magnificent milestone that promised to transform our understanding of human biology. In fact, some even dubbed it medicine's moon landing. Here's the then US President Bill Clinton when the announcement was made. More than a thousand researchers across six nations have revealed nearly all three billion letters of our miraculous genetic code. With this profound new knowledge, humankind is on the verge of gaining immense new power to heal. Genome science will have a real impact on all our lives.

2:02It will revolutionise the diagnosis, prevention and treatment of most, if not all, human diseases. In coming years, doctors increasingly will be able to cure diseases like Alzheimer's, Parkinson's, diabetes and cancer by attacking their genetic roots. But that dramatic moment in the year 2000 was the product of work that had begun much earlier. Long before powerful sequencing machines or global databases, a small number of researchers began to imagine a radical possibility. What if we could decode not just individual genes, but, as Bill Clinton said, the entire 3 billion genetic letters of the human genome?

2:40Much of that early vision took shape here in Cambridge, the city where DNA's double helix was first described in 1953. By the time the Genome Project was gathering momentum, Cambridge scientists, including those at the newly founded Wellcome Sanger Centre near Hinkston, were at the heart of the effort to turn a dream into data. And one of those scientists was Richard Durbin. A mathematician by training, he became a key figure in the development of genome sequencing and analysis, and he was involved in the UK's contribution to the Human Genome Project. I met with him at Queen's College, Cambridge.

3:13We knew about genes and some of their functions because we can damage them, We can make mutations in them, and then they don't work properly. And it's a bit like saying if you break the chain on a bicycle, and then however hard you pedal, you never get anywhere. So it's involved in transmitting power to the wheels. So that sort of genetics was done for the whole of the 20th century, really. And then Watson and Crick, actually here in Cambridge, discovered that the genetic material was made of DNA, and they worked out that genes were these regions of DNA sequence. and so to study them better and understand what was going on we wanted to get hold of these genes the one that made the chain work and find out what it actually was and that allowed you to both do experiments and gives a lot of insight actually into how it works at a molecular level.

4:03Did you start with the human though? Did someone just walk into work one day and say let's sequence a human or was the field already going in that direction and the human was the next step? We're all interested in human biology or interested in how we are made and how we work. But we're very complicated. And for a long time, people have also worked on simpler organisms. And famously, there's two very simple organisms, the fruit fly and a little worm, C. elegans. And in that system, people had started in the 1980s to find the genes in the genome. It's a much smaller genome than the human genome, one-thirtieth of the size.

4:40But it was a big endeavor. It took any one person several years to find a gene in the 1980s and to find its sequence and to start to study it. And really, people began to have the vision that we could speed all that up if we did a proper job to start with and made a library where you could go and just pull out of the library the book of this particular gene and start work immediately on it. And that was the vision that grew in the 1990s and we started with these simpler organisms worms and flies. Did we have the technology then in the early 90s to read a genome end to end like that like you would pick up a book and just pick your way through all the letters in the words or did that take a step forward in technology to even make people begin to conceive of doing that?

5:27Well people knew that there was a sequence from the 1950s and 60s but it was only in the end of the 1970s that Fred Sanger also here in Cambridge worked out how you could read decent lengths of sequence. So you could read hundreds to thousands or so DNA bases at a time. It was that that people were starting to apply in the 1980s. And in principle, there was no reason not to just scale that up. You could do something on this scale of hundreds to thousands. We knew that we had to get up to millions. The C. elegans genome is 100 million bases long. So it was a challenge, but you take one step at a time and you can walk 100 miles.

6:11So the idea of sequencing a human came out of doing the same thing for a worm, but a human is a totally different ballgame. I mean, that's 3 billion genetic letters long. That was pretty awe-inspiring at the time, I would think. What was the original timescale that was conceived when someone put up the idea of doing this? Originally, I think the sort of official kickoff date was in 1990. with a goal of 15 years. You know, in fact, we did it under time. It was a big spend, though. I mean, it came in at something like three billion. That would have bought you a large Hadron Collider or put you a reasonable space project in space, big telescope or something.

6:50So this was a big spend. How did scientists convince policymakers and the people with the readies that they should spend a lot of money on this? It is a lot of money. not everyone did buy into taking on this very big industrial scale project a lot of scientists tend to forget how opposed they were who now I spoke to some Richard I spoke to some at the time when this was coming out because I was a medical student at the time there were a lot of naysayers who said this is all fanfare it's all hot air this is all hype this hasn't done anything well we're now 25 years on and I think it's completely transformed biological science and I think it's now pretty clear that we're living in the era of sequencing genomes.

7:35I believe in a hundred years or a thousand years people will look back and say this was a key point in science like the discovery of gravity or you know the discovery of quantum mechanics at the start of the 20th century. Some people say this is like biology's moon landing. Yeah I think it's it's completely transformational for our the science of life. One of the things that was said at the time is that it's one thing having a map of all the genes and where they are but we still don't know what they do have we kind of got there now have we gone through all the 20 000 genes do we understand what they all do or is there still a huge amount to find out no we absolutely don't understand what they do i mean we know more about what they do than we knew 20 years ago or 50 years ago but it's the job of the present and the future to understand what they do and how they work and how we can work with them For example, we know that now in a much clearer way how cancer is a genetic disease, how mutations in cells in our body that cause those cells, which now have a different genome because of mutations, to change their behavior and start to grow and act in ways that are not in the interest of the rest of the body.

8:45And we can study them using the power of genomics. But we do need computers and data analysis tools to study them, to use that information. genomics goes hand in hand with computing and we'll be hearing more about the next step in the relationship between computers and genomics in just a moment that was richard durbin talking to me at the university of cambridge about how he and his colleagues got the human genome project going the welcom sanger institute was one of the global powerhouses behind the sequencing project which launched in 1990 and was completed by 2003 along the way the genome project revolutionized genetics and medicine because the investment that flowed in unlocked new technologies that are now commonplace in diagnostic and research laboratories around the world.

9:32At the same time though, skeptics wondered whether the project would achieve its ambitions and hit the target on time. So the next call I made on my genome journey was to meet with Matt Hurles. He's the director now of the Sanger Institute and he describes the place as the genetics equivalent of Bletchley Park. we got together in the orchard to talk about how it was competitive pressure from commercial organizations that really drove the race to sequence the genome in record time in the first place so i think they two major things that that competition really helped stimulate the first was it coalesced the public funders to really make sure the project had the funding it needed to deliver and secondly it pushed the scientists themselves to re-evaluate their technologies and to look at how they could speed them up and make them faster.

10:19What happens then? So I'm sitting in that lecture theatre. Sydney Brenner, who's to get the Nobel Prize two years later, says tomorrow, this is 25 years ago, almost to the day that we're sitting here, there's going to be a really big announcement in one of the world's best science journals announcing the first draft of the human genome. How transformative was that? Some people have said that was almost like the biology equivalent of the moon landings. I mean, do you think it was that enormous? I do think it was enormous. Essentially, it's the blueprint that we use today. And in a world where you don't have a blueprint, you're incredibly limited by what you can do.

10:53I think one of the things that's important to realise is the human genome is a bit like having an incredibly detailed map of the world at a point at which you can only really walk around the world. And so to make the use of that knowledge, you then have to invent a whole suite of technologies to allow you to use that knowledge in different ways. And how has the way we do this changed? How has the enormous investment, about$3 billion went into the Human Genome Project, what has that done to the industry and what can we now do that we wouldn't have set a chance of being able to do before? So I think the diseases where we knew that DNA was important, all of them had been revolutionised.

11:32Our understanding has been revolutionised. So in rare diseases where we knew that single gene mutations can cause disease, we've discovered many thousands more diseases than we had previously appreciated. and those technologies that were used to sequence the human genome and their successors are now being used in clinics around the world to diagnose those diseases much more rapidly. It's a very similar story in cancer where we knew that cancer is a disease of mutations acquired in DNA in cells throughout the course of their life. We've discovered many hundreds of cancer genes now and many drugs are now targeting those particular mutations.

12:08in common diseases things like type 2 diabetes or rheumatoid arthritis or asthma it took a bit longer because the genetics there is much more complicated but now we have a pretty accurate picture of the genetic architecture of those diseases and that information is being used by drug companies around the world to design better drugs to new targets that weren't previously appreciated it's moved on in a number of ways technologically as well hasn't it because on the one hand it's a way of reading dna we've got a number of very very fast methods for doing that now but it's also a computer problem isn't it people have now got better computer programs very powerful computer systems that can comb through dna piecing little bits of dna together to work out sequences and and also spot patterns in certain genes linked to certain diseases which we couldn't do that before so the technology to scale up dna sequencing is really being at the heart of those changes because as soon as you can generate data that's a thousand or indeed a million times greater than before all of your computer programs which worked on your previous data no longer work on your current data and you can also answer questions that you couldn't previously even hope of answering before so the advent of new technologies which generated much much larger data sets then drove a huge amount of computational innovation so the the sanger institute that was set up to initially lead the charge getting the human genome read once that job was done which by about 2003 it was pretty much done what have you pivoted into doing now so the key thing was the realization that that different way of science was as important for delivering the benefits of human genomes as it was from sequencing the human genome in the first place So we pivoted to understanding rare diseases and common diseases and cancer and using those new technologies generating vast amounts of data to enable us to kind of answer the kinds of questions that I just mentioned.

14:06but also more broadly to use similar technologies to map all the cell types in the human body and create a reference of all the different cell types and increasingly start to use those same DNA sequencing technologies in an industrial scale to generate reference genomes for all of the species on the planet. Including the microbes because you're beginning to link microbes to diseases as well. Absolutely so all kinds of pathogens have genomes as well whether they be viruses or bacteria or others they're much much smaller genomes and so we can generate much much larger numbers of them in fact the genome we've generated the most sequence of is the covid genome and at one point in during the pandemic we were sequencing 70 000 covid genomes a week good grief and where did all that data go so a lot of that went into the public global repositories as being used to develop a lot of the science around covid which has been transformed from being a pathogen nobody knew about to the pathogen on which there's the biggest amount of sequencing data of any species on the planet.

15:09What's next? So what's next for us is starting to think about how we can use the kind of technologies of genomics to engineer new biology that can do new useful things for humanity. Like what? So that could be enzymes that break down pollution, it could be more effective ways of doing photosynthesis, it could be new materials that do new jobs for humans but are actually biological rather than inorganic. You're working on all that here? So we're working on the foundations of those latter parts. So that whole engineering of biology, I think, will be a huge direction for the next, well, the rest of the century.

15:45I think there'll be many, many different organisations, public, private, that'll be generating that. But we're trying to build the genomic data sets that give you the foundational models that allow you to design what you want to build. Matt Hurles talking to me at the Sanger Institute and looking to the future there.

16:2659 different sizes in really soft fabrics and really good fits. Because at the end of the day, your bra should hold you in and not hold you back. Visit Aerie.com to find the right bra and your perfect fit.

16:39The Naked Scientist podcast is produced in association with Spitfire. Cost-effective voice, internet, and IP engineering services for UK businesses. Find out how Spitfire can empower your company at Spitfire.co.uk.

16:58Music in the programme is sponsored by Epidemic Sound, perfect music for audio and video productions. This is the Naked Scientist podcast with me, Chris Smith, and today we are exploring how far research into the human genome has come in the 25 years since it was unveiled for the first time. So let's turn our attention now to some of the applications of the genetic know-how and technology that we have at our fingertips. One of the revolutions we've seen is in the rapid diagnosis and treatment of life-threatening conditions like sepsis. And this is what Emma Davenport, who is group leader at the Wellcome Sangre Institute, works on.

17:33When different groups of people are fighting a particular infection, there can be a real variation in the way that they respond. And so as well as the particular infection that they're fighting and maybe comorbidities that the person has, we know that their DNA sequence can make a difference as well. Therefore, if you can read it, you can make predictions. is this anticipatory before a person's even got near an infection or is it once an infection has taken hold at that point you're beginning to ask these sorts of questions? So for us and my particular team it's very much once they have an infection what information can we use about their genome to actually understand how they're going to respond to that?

18:10Is it just their genome or is it the genome of the thing that's doing the infecting as well? Absolutely yes so we know that the genome of their bug is also important and that's something that we're now trying to integrate into our projects as much as we can too. So it's sort of building a system that can integrate the host, us, our genetic makeup, the genetic makeup of the thing that's doing the infecting and then the response and presumably the outcome as well. Yes definitely and for us that's what's really interesting is there's lots of different outcomes you can use so it could be how severely ill somebody is or whether or not they survive the infection.

18:46How are you doing this? So we work with a lot of our clinical collaborators and we are able to collect samples from patients when they're undergoing a particular infection. We bring those samples onto site here at the Sanger and we're able to extract the DNA and the RNA from their blood samples. And then we use that to generate these large data sets of DNA sequencing and RNA sequencing. That gives you a genetic code, a readout. How do you marry that up with what the disease is doing? So our team in particular focuses on gene expression. So this is a bit like a dimmer switch in the light. So a gene can be switched on or off or anywhere in between.

19:23And we know that an individual's DNA sequence can regulate how much that gene is turned on or off, but also their environment. So if they're fighting a particular pathogen or a particular bug, we know that that can have an influence too. So we look at across the genome, how much in a person these genes are turned on or off. and we compare that across all our individuals and we link that up to their survival rates as well and you can start seeing signatures of genes and start correlating those with then the outcomes of interest. Do you need a computer to comb through that data? Is that how you do it?

19:55Because it sounds like when you've got 20 ,000 genes in the average person to play with and you've got hopefully a big group of people you can study, that pretty quickly inflates to extremely large numbers of variables. Yes, it definitely does. And we are working on cohorts in the hundreds to the thousands. So we're really sort of building up really nice big data sets. And so all of my team are computational biologists and we spend our time running our experiments on our laptops or on our big servers that we have available at the Institute. What particular diseases are you focusing on? So we particularly focus on sepsis.

20:30So this is when the body is responding to an infection, but it ends up injuring its own tissues and organs. Generally, people are admitted to the intensive care unit when they're that sick. This will tell you something prognostically about what's going to happen to that person. But does it do the other important thing is, can it throw open the door to possible leads for what the best treatment might be? Or new ways to research new drugs that attack the body or attack the disease in new ways. So we can basically get a handle on it from a different direction. Yeah, this is an area that we're particularly excited about.

21:02So historically, sepsis has been really difficult to treat because generally the options are to give patients antibiotics to try and fight the infection. But we don't always know exactly which bug that individual has. And then organ failure can be treated by, for example, if someone has kidney failure, they're put on dialysis. But what we think is that actually targeting someone's specific immune system and the way that that immune system is responding could help us to think about the way that new drugs could be used to target this. therefore you might get a situation where there are certain changes in the way the immune system works in certain people and if you know of drugs that will make that happen less often they're less likely to have that vulnerability is that the sort of direction of travel when they're sort of coming in with this really bad outcome we want to be able to look and see how their immune system is working at that moment and decide whether certain drugs may move them into one sort of trajectory or another so we can see if they potentially have too much suppression of the immune system and they're not able to fight the infection anymore then we may need to boost it or if their immune system is sort of acting too aggressively we may need to dampen it to try and reduce the organ failure.

22:12Is it bearing fruit yet? So we've managed to identify two gene signatures which we use to differentiate patients into two different groups and we from our experiments know that there's two very different immune responses and they're also associated with different survival rates. So our collaborators at the University of Oxford are now working closely with a company to try and turn this into a point of care diagnostic test. So one of the main challenges with something like sepsis is this happens really quickly, like hours and minutes can be really important. And so the key challenge here is being able to take a blood sample, extract your RNA and measure the levels of those genes really, really quickly so somebody can react to that within the next couple of hours.

22:53That hasn't been possible until sort of more recent times and the changes in technology now mean that that's now an opportunity for us. And now we've got to figure out on the computational side which are the best genes to measure and how we come up with some sort of score that the clinicians can use or the doctors can use alongside the other clinical measures that they currently have to help them make a decision. Sepsis is obviously an extreme example. There are many, many other health conditions where it would be useful to know what's going on as soon as possible. Do you foresee this finding a place in primary care?

23:25The average GP will start to do things like this in the future. They'll take a sample and we'll have a way of registering a signature of a disease and therefore pointing us towards the right sort of treatment for a whole raft of different conditions. Yeah, I really hope this is the future for us. And you can see, you know, for example, in areas such as autoimmune diseases, which might include things like rheumatoid arthritis or lupus patients again have very variable symptoms that they present with and often it's very trial and error to work out exactly what kind of drug might suit them. Emma Davenport group leader at the Wellcome Sanger.

23:59Now as Matt Hurles was telling us earlier the Sanger Institute is pioneering a project to identify and catalogue each of the 400 different cell types in the human body but these are all uniquely susceptible to different disease processes and therefore different drug treatments potentially. Moreover all these different cell types are exchanging information with each other through various chemical conversations so when we think about a disease we actually have to consider what's happening both to the affected cell types and how this is influencing or influenced by other surrounding cell types and also what effects drug treatments might have on all of them.

24:34Are there for instance any emerging and consistent patterns that can flag up attractive drug targets. This kind of pattern recognition is something that AI systems are brilliant at solving, and it's an approach that's being developed by Mo Lott-Falahi, who's building AI models to unpick on a cell-by-cell basis the mechanisms of diseases and their treatments. My lab and I work specifically on a technology called single cell genomics. So you basically try to isolate a cell, and you try to extract the molecular information. So all those informative data that is happening in the cell, you try to extract it.

25:11So you have information about DNA, RNA, all those interesting bits and pieces in the cell that would give you information about disease and how a disease cell would look like. Is it that you're choosing an individual cell because then you know it's just that discrete cell? I'm not getting blurriness because of what's going on in another part of the body, which just happens to have added some molecules. I've got just that cell, it's a discrete unit, I'm interrogating its genetics and its biochemistry. And you just keep doing that many, many times for a cancer, for example, and you learn how that particular disease tends to present genetically in those sorts of cells.

25:50Yeah, we phrase that with a term called heterogeneity. So heterogeneity means that if you get a disease, so let's say you have a flu, right, and then your lung is affected, but not every single cell in your lung is affected. So that's why we have to look at individual cells, because they respond to the disease in a different manner. And then cells, they don't work in isolation. They actually, they're like a social network. They talk to each other. So they're affected by each other. And what we're trying to understand is, like, by looking at individual cells, but also how they interact with each other, is A, what are the cells that have been affected by disease?

26:26B, what is causing the disease, and then understand the network of cells, i.e. the neighbors of those cells that are diseased, and understand what caused that. The first step is training the model, teaching it what normal or even abnormal looks like. So how do you do that first and foremost? The most dominant technique here is called self-supervised training, which is simply what we do is say that we have a picture and that picture constitutes a face, a nose, an eye, blah, blah, blah. And what we do is we feed that data to the model. We basically take away part of the picture and we ask the model to predict that missing part.

27:10So this way, the model learns, A, what is the cell? What is the information content within the cell? And then how would they change upon disease? How do you then tie that once the model understands what a cell looks like genetically and so on? How do you tie that to a mechanism of a disease or a treatment for a disease? We train these models based on healthy data. So we take samples from healthy donors. We also have disease data. We also have data from disease patients treated with a drug. So this way the model can distinguish between a healthy cell, a diseased cell, and a diseased cell treated with a drug.