Peter Donnelly on genetics and healthcare

Thanks very much for the kind invitation. It's a great honour to be here, and a huge pleasure to have the opportunity to tell you something about the excitement of modern genetics, both in terms of the science, the research and also its impact, already, and I think very much in the future, on human health and healthcare.

It's also great to have so many people who've given up part of their summer evening to come along and listen and learn about the field. I have to say, rather bravely, as you've just heard my own background is in mathematics and statistics. Statistics is a subject which doesn't have a huge amount of excitement and street credibility, so it's particularly brave of you to be here. One of my favourite aphorisms about statisticians is that they're people who like working with numbers, but don't quite have the personality skills to become accountants. And another is the rhetorical question, how do you tell the extroverted statistician from the introverted one? To which the answer is that the extroverted statistician looks at the other person's shoes.

So, I hope you'll go away from tonight's lecture with some sense of modern genetics and its impact on healthcare. But if you're more practically minded and want to take away a practical lesson, I'll get that over and done with first of all, and that's the following – it's a little tip on how to cope with those social situations we don't always find so easy, when you're at a social event and, and someone comes up, they're talking to you, and you'd actually much rather be talking to someone else.

So here's my tip for how to cope with that. Sooner or later they'll get to the obvious question, they'll say to you, what do you do? Well then try saying, I'm a statistician. The reaction will be something like the following. They won't say anything. You'll see their eyes kind of glazing over. You'll have a sense of cogs whirring as they're thinking, how the hell do I get out of this? And then, finally they'll look over your shoulder, and light will come into their eyes, and they'll say, Oh, my gosh, there's Felicity. I haven't seen her for ages. I better just go and say hello. And you'll have achieved your purpose. You'll have the opportunity to talk to whoever you really wanted to talk to, in the first place.

I've never found it very easy to explain in a sentence or two at a dinner party what I do and what my work involves, and certainly didn't in the days when that was primarily focused on statistics. But my then girlfriend did a much better job than I'd ever managed. I was living in the US, she was working for the BBC in the UK. It was early enough in our relationship that she kind of concentrated when she'd asked me what I did, and I'd said that I use mathematics and statistics to model the way populations evolve through time and the way their genetic composition changes. And so sometime later, when I was coming to visit her, one of her colleagues at the BBC said, Oh, it's great that your boyfriend's coming to visit. What does he do? And Sarah thought a little bit about the explanation about populations evolving and so forth. And she paused and said, Oh, he models things. Well, the woman who asked the question got quite interested and said, Wow, what does he model? And Sarah thought of it more and said, Genes. He models genes. And not surprisingly, she was a bit disappointed when she met me, but at least for a moment there was some enthusiasm.

So I want to try to give you a sense of genetics, and as far as human diseases go, there's a spectrum. At one end of the spectrum are diseases where genetics is the whole story. And at the other end are diseases is where it's a part, and a small part in some cases, but a part of the story about why people get sick. But at the extreme end are conditions – you may have heard of some of them, cystic fibrosis, Huntington's disease – where if you inherit in some cases one, and in some cases two changed copies, mutated copies of a particular gene, you definitely get sick.

Now, each of those conditions is typically individually quite rare, but they're also typically extremely serious. Although individually rare, it's estimated that over 1% of children born will suffer from some kind of serious genetic condition during their lifetime. Those diseases in our trade are called Mendelian, because the inheritance is quite simple. It's like Mendel and his pea experiments. It's because genetics is all of what's going on.

And the first big breakthrough in what I think of as the modern era of genetics happened in 1989, 35 or so years ago now, when for the first time, the gene involved, that bit of our genome involved in one of those conditions, cystic fibrosis as it happened, was identified.

It's kind of a sobering and salutary story that although the gene for cystic fibrosis was identified 35 years ago, it hasn't led to any direct treatments, or at least it hadn't until a year or two ago. It has just over the last year, genetics has been used to introduce a new and extremely effective treatment for individuals with one particular genetic change. That helps about 4% of the people that have cystic fibrosis. But even though the genetics hasn't led to a cure or new way of treating most sufferers, it has considerably improved their prognosis. Because we know about the gene, it's easy to test individuals at an early stage to confirm whether or not it is cystic fibrosis they're suffering from very early in their life. And that's allowed the targeting of therapies that alleviate the condition. So in the 1960s, for example, the median lifespan of someone with cystic fibrosis was about 10 years, and now it's about 40.

As I said, just a year or two ago, the first drug that really took advantage of the new genetic understanding for cystic fibrosis was developed. And it's had a hugely positive effect, but just on a subset of patients, and it gives us hope of changes more widely, but also some sense of the timescale, a realistic sense of the timescale from the scientific discovery to the treatment.

And that was the first major advance in the modern era of genetics. The second one was the discovery of two genes which are associated with familial forms of breast cancer. You'll see the names of a number of genes during the talk tonight. There won't be, even though it's a library and a place of learning, there won't be a quiz on the way out. So these two genes are called BRCA1 and BRCA2, BR for breast, CA for cancer. And they're genes with the property that if a woman inherits a mutated, a particular change in one or other of the copies that she carries of this gene, it substantially increases her risk of breast cancer. So, for a typical woman, the lifetime risk of breast cancer is somewhere between 10 and 12%. And if you have a mutation in one of these genes that risk goes up to something like 60 or 80% for breast cancer, and it also substantially increases the risks of ovarian cancer.

The genes were identified in the mid-1990s and since then they've allowed testing, for women who choose to be tested, who are in a family where there's a history of breast cancer. Some people will choose to have the test and some won't. And for women who do have the test, even if you're in a family which carries this genetic mutation, half the women tested won't have it. So that's a huge relief for them. And then the other half have choices, typically tough choices about preventative measures that they might take. And the whole issue of genetics and breast cancer was brought to prominence last year by Angelina Jolie, who chose to have a double mastectomy, and to tell everyone about it, after the fact, as a way of popularising the issues to do with genetics and breast cancer.

You've no doubt heard and many of you will know something about the human genome project, so that was a project to read one copy of our genome. So the genome is the totality of the genetic information that we carry, or actually half of it, because we get one copy of our genome from our mother and one from our father. In the case of humans, so that our DNA reads out a kind of genetic code, it's written in an alphabet of four different letters, and the human genome is three billion letters long. That sound like quite a lot of information, and it is. So each of the cells in our body, those cells are quite tiny, carries a copy of our genome, or the two copies, the one that we inherit from our mother and the one from our father.

If we were to kind of unwrap that, it's tightly wound up, if we were to unwrap it from a single cell, it'd be a few meters long. If it's not a pleasant thing to think about, but if someone took the DNA from all of your cells and laid it end to end, it would reach to the sun and back, over 100 times. So each of the cells in our body contains this instruction manual, bits of which we're starting to understand.

It's also kind of tempting to think that as humans we're amongst the more sophisticated creatures on the planet or at least many of us are. We might like to think that we'll need more complicated instructions than other creatures. By and large, that's true, but is not universally the case. So there's a species of lizard, a little salamander, whose DNA is about 30 times as long as ours is, and they probably think they're rather clever as well.

The human genome project was a massive international endeavour to read one copy of the human genome. It took scientists from six or seven countries, many thousands of them working from 10 or 15 years, and cost well over $2 billion. The draft version of the human genome sequence was announced in 2001 to great fanfare involving Bill Clinton, president of the US and then Tony Blair, prime minister of the UK. And the so-called finished version was completed two years later. And at the time it attracted quite a lot of attention and there was a lot of hype about what would happen. I think it's fair to say that we probably all underestimated how long it would take to take advantage of the information in the human genome project. But we're just at the stage where much of that is coming to fruition, and that's what I hope to give you a sense of.

So if it took a while for the impact in terms of health, how did knowing a version of the human genome help us? The easiest way to think about that is it's like a map. So we've got inside each of ourselves this vast amount of information that we don't really understand. And it's a bit like the days of early explorers trying to learn about certain parts of the world. That was much tougher without good maps. And the human genome sequence, for the first time, gave us a map. It told us which bits of our genome were where, relative to others. So it's been an indirect help. And as I said, we're just at the stage of reaping some of the benefits of it.

So, we all carry two copies of the human genome, and by and large that's the same for all of us. It's got the instructions that make us human and relatively similar to each other. On the other hand, if you compared one of your chromosomes with a chromosome of the person sitting next to you. Or with the other chromosome that you carry, the one you got from your other parent, then they'd be similar in most places, but they differ at about one position in a thousand of the letters in the DNA code. And those differences have become very interesting. Those differences are responsible for many differences between us. And in particular for differences in susceptibility to a whole range of diseases.

I said that if you compared your DNA to the person sitting next to you that they would agree at one place in a thousand. If, and don't take this personally if you're sitting next to someone, if that person were a chimpanzee rather than a human, then your DNA and their DNA would agree at 99 places out of 100. So the chimpanzees are our closest living relative, and they share much of our genetic code.

The differences, as I said, in our code, those places that do differ, they're responsible for a lot of interesting features about us, and also they're related to disease susceptibility. So remember I said that at the one extreme there were conditions where genetics was all of the story, and they were typically rare and very serious. At the other end of the extreme, at the other end of the spectrum, are all of the common diseases, diseases like diabetes and arthritis, heart disease, schizophrenia, many of the cancers. Genetics is part of the story about why people develop those conditions. We know that because we can look to see the extent to which those diseases run in families more often than they should.

Genetics is part of the story, it's certainly not all of the story. For many of those conditions, there are other factors, factors to do with our lifestyle and our environment, and so on, which also affect our risk of disease. So part of the story is genetics, as it were nature, and part of it is the environment we live in, as it were nurture.

And although we've known for a long time that genetics is part of the story of susceptibility to these diseases, it's only been relatively recently that we've started to understand exactly which changes, which differences in our DNA, make some people more likely to get heart disease. And others, for example, more likely to get arthritis or perhaps schizophrenia.

So I'm going to take you through to give you a sense of the pace of discovery, a series of cartoons like this. So each one of these stripy things is a kind of representation of one of the 23 human chromosomes. And they're numbered from largest, that's chromosome 1, chromosome 2, chromosome 3 and so on, to smallest.

And we're going to start the series of pictures in 2005. Before 2005, there were a handful of examples that we knew of changes in our genetic code that had been definitively associated with people's susceptibility to particular diseased. And then in 2005, there was this discovery. So what does this mean? The little lollipop here stuck somewhere on chromosome 1, what that means is that, and I'll show you more of these, during 2005 there was a discovery of genetic variance in that position in chromosome 1 which affected susceptibility to a particular common disease. In this case a disease called age-related macular degeneration, an eye disease that people can suffer from as they get older.

So before 2005, we had a handful of examples in total. There's one discovery in 2005. There are a few more in 2006, actually one or two more were the same disease, so if the lollipop's the same colour, it's the same disease. And a number of others in a version of inflammatory bowel disorder. That was 2006.

In 2007, the position started to change dramatically. There were 50 or 60 discoveries that year, and we thought, and I'll come back to this, that that was huge progress. That was 2007; that was the position a couple of years later in 2009, 2010. That was 2011, by which stage we knew over 1500 letters in the DNA where changes affected susceptibility to diseases. And the range of diseases, I won't expect you to read this nor will there be a quiz on this either, you'll be relieved to know, but the range of diseases covered things like heart disease, arthritis, schizophrenia, diabetes. And also kind of ordinary conditions, things that aren't diseases like height and weight, where again, genetic changes are responsible for some of the differences between us.

That's a position represented in a slightly different way a bit more recently. So there are now 2000 or 3000 positions, where we've done the research to learn that if you have an A rather than a T at this place in your DNA, it might make you more likely to get say, depression or arthritis or diabetes.

As I said, in 2007 we were pretty excited about what we saw then as a sea change, and Science magazine, which is one of the major general science magazines, each year kind of rather pompously looks back on the developments during the year and picks one break through or set of break throughs in all of Science, to highlight, and in 2007 it highlighted this then fairly substantial expansion in our understanding of genetics and common diseases.

I want to tell you how we did those studies. It's not too hard, it's not rocket science, as it’s tempting to say. Kind of obviously not rocket science, but it's genetics and biology rather than astrophysics. It's also not rocket science in the colloquial sense; the basic idea is really easy. So what we did in these studies is to take large numbers of people who had the disease of interest. Let's think about, say, heart disease, so we might study 2000 or 3000 individuals who have heart disease, and measure lots of their genome. We couldn't measure all of their genome easily in those days, but we could measure half a million places, say, in the genome, relatively inexpensively, so Australian currency in those days, maybe $300 or $500 per person. So we measured half a million places in the genomes of the people who had heart disease, and the same positions in the genomes of a random sample of the population, most of whom won't have heart disease, and we look for differences. We try and find places, so a place in the genome, where there are two different letters that different people carry in the genetic code, maybe an A and a T, with the property that the A is more common in the people with heart disease, relative to the set of controls, the random sample from the population.

So here's why that's relevant. Suppose it were the case that the A predisposed you to heart disease, it increased your chance of getting heart disease, then it would be more common if you took a sample of individuals with heart disease compared to the healthy people. Now we're measuring lots and lots of places across the genome and so these are very large data sets and that's the way those rather undervalued statistical skills that I had come in handy and making sense of the large amounts of data that we have.

Why do the studies, and what's the impact been? There are two different sorts of reasons, and I think most of us would argue that the first reason is rather more important than the second. The first reason is the following. So for most human diseases, in spite of very extensive study over many, many years, hundreds of years in some cases, we know depressingly little about what goes on to trigger the disease, about what goes on inside people biologically that leads to, say, heart disease, or schizophrenia, or arthritis. We try to study diseases kind of from the outside in, from understanding the symptoms and going down to the basic biology.

Genetics and the series of 1500 or 2000 findings that I have described, they allow us to go the other way. We can start from the kind of fundamental building blocks from the genetic code. Suppose we know, for example, if you have an A at this position you're more likely to develop heart disease than someone who has the T, then we can ask, biologically what's the difference? What's going on inside people with the A, compared to the T? And maybe it will be the case that there's a particular protein produced by a gene which is different, or more commonly for those findings, there’ll be a particular gene which is turned on when it shouldn't be, or more often than it should be, or less often than it should be, in producing its protein.

So the genetics gives us a whole new foothold, a whole new insight into the biology underpinning the disease. And one of the really striking features of that whole series of discoveries I have described is that for almost all of them, they pointed at genes and bits of biology which we wouldn't have guessed, and which the experts in the field wouldn't have predicted, were part of the disease process in question.

So all of those discoveries are a whole series of clues which can allow us to understand the diseases better. If we do understand that biology better, then we can do a better job of developing new drugs and new treatments. That process will take time, tens of years probably, but I think it will be one of the major impacts of the series of discoveries that I've described.

The other thing that we could try and do, which is much more to do with individuals, instead of using the genetics to learn about biology, and I think that, as I said, that'll be the major impact, we could take specific individuals, we could measure all those variants. See whether you got the A rather than the T in that position, so that might increase or decrease your risk of heart disease, and so try and make predictions for individuals about their disease risk.

So at the moment we can't do that very well on the basis of the genetic discoveries we've found, although every one of those 1500 that I had described affects the risk of the disease in question and we've got solid evidence of that. Usually the effects are very small. If you have the A rather than the T, it might change your risk of heart disease by 5% or 10%. So, it's certainly not definitely going to determine who will get heart disease and who won't. It's part of a complex picture of risks. And that's the genetic part, the part which is complicated. And then there's the environmental and lifestyle factors, which also contribute.

So actually at the moment, although we've learned all of these positions in the DNA which are related to disease susceptibility, at the moment the best predictor of your likelihood of getting heart disease, or diabetes and so on, is your family history. It's something that doctors will routinely and appropriately ask for. And it gives us a better sense of the genetic risk factors you have, than the variants that we've learned about.

So I want to share a little bit about the science. We were, as you've heard, involved in one of the very early studies; something called the Wellcome Trust Case Control Consortium. And as a scientist, what you hope is that your research will be interesting enough to one of the major science journals, and in this case, it was, and we published in Nature in 2007. What's a bit more scary as a scientist is what the wider media might make of it. But in this case, they also found it interesting. So the Independent, one of the broadsheets in the UK, they devoted all of their front page, or almost all of their front page, to this first wave of discoveries that I've been describing. Although rather interestingly, we shared the front page with a woman called Tracey Emin, some of you may know her as one of Britain's more controversial avant-garde artists.

Here’s another newspaper, part of the respected broadsheets in the UK, the Daily Telegraph, so there we share the front page not with an artist but with the English soccer, as we would say in Australia, but football as they would say in England, team. Obviously something massive had happened for them to get on the front page, and as an Australian I rather enjoyed this, what they actually managed to do, they'd reached the dizzying heights of beating Estonia three–nil in a qualifying match for the Euro 2008 competition, which they subsequently went on not to make anyway.

So I'll spend just a moment or two actually showing you some of the data from the studies. So here's probably the most interesting finding of our original study and it relates to genetic variance associated with heart disease. So what's this picture about? This represents a position on a particular human chromosome, in this case human chromosome nine. And each of these black dots is one of these genetic positions that we measured. And what the plot shows is, how far the point, the dot, is above this axis, so in this direction, is a measure of how the statistical evidence reflects potential differences in frequency between the sick people, in this case with heart disease, and the healthy people. And what you see here and here is kind of background noise. And then in the middle there's a huge leap to very, very compelling statistical evidence of a difference. As I said, this is one of the most interesting findings of our study. A number of others reported it at the same time.

So here's a variant which is definitively associated with heart disease, and the first thing one tries to do is to work out how that might be affecting biology. So, you look for any genes in the region. So, genes are the bits of our genome that produce proteins. There are two genes in this region, marked here. They're called, as I said, genes don't have very memorable names, CDKN2A and CDKN2B, so no one studying heart disease had thought to focus on these genes. But actually they were quite well known in another biology, they’re very well known to cancer biologists as so-called oncogenes, genes that play a crucial role in the development of cancers.

Again, it’s a salutary tale. This affect is reasonably large. Each copy of this, whatever the letter is, an A say, you could have zero, one or two copies of it on your two chromosomes. Each additional copy increases your risk of heart disease by about 30 or 40%. So, if you have two copies of the A compared to two copies of the T in your genome, your heart disease risk goes up, it's not quite doubled but it is somewhere near that. It is comparable with the risk of, in terms of heart disease that cholesterol, one of the risk factors that we currently measure, is. As I said, most of the variants we've learned about have much smaller effect on risks than that.

So no one had thought of looking at these genes for heart disease at the time and, actually in the six or seven years since, and it's another warning, that in the six or seven years since, we still haven't got to the bottom of what the biological role of these variants are, and exactly how they affect heart disease.

This is one other example that was quite interesting for reasons that I will explain. So the same kinds of picture. So here is the regional human chromosome 16 around a gene called FTO where genetic variance that we were studying affected people's risk of type 2 diabetes. That’s a type of diabetes that you get typically in late middle age, if you're to suffer from it. And it turns out that we'd been studying diabetes along with collaborators doing similar studies and when we compared our results with theirs it was quite scary because they didn't see anything here. And that's the worst possible thing as a scientist. You think something must have gone wrong with our study, otherwise why didn't they find it? Well, there was a bit of detective work that went on. And to cut a long story short, what we found was that this variant does affect the risk of type 2 diabetes, but it does it indirectly. What it affects directly is people's weight. And higher weight, obesity, is a risk factor for type 2 diabetes. So if you have two copies of, whatever the letter associated with this position is, if you have two copies of that, you’re on average about three or four kilos heavier than someone who has two copies of the other variant.

So, as the first genetic variant that have been found to be associated with human weight, as you can imagine, it attracted rather a bit of attention in the press. Headline writers had a field day. So even the Financial Times, that's the kind of doyen of serious British reporting, was interested, both in the finding, and actually in a nice way, they devoted some editorial space to the fact that we, as most of the others doing these kinds of studies, made our results available immediately, rather than trying to protect intellectual property, our feeling and that of our funders was to get the results out there to allow as many people to work on them as possible. So the Financial Times was interested, but I think this is the favourite of their headlines.

So I’ve talked about a series of studies and shown you their impact where we typically look at large numbers of people with a particular disease and compared them with controls, measured lots of the genome, so how has the field moved on? Well, the field has moved on in a number of different ways. We're looking at much, much larger studies, typically tens and in the case of a large UK study that we're involved in, 500,000 individuals. And instead of just looking at individuals who have heart disease or don't, these are individuals who've been fairly extensively studied, so we can measure genetic information extensively on them. Currently in the large UK study, which is called UK Buyer Bank, we're measuring about 800,000 positions in their genome, but within a couple of years we'll be able to measure the entire DNA sequence. And all of those individuals have had many physiological measurements. They have consented to be part of a very big study, they have had many physiological measurements taken on them; there's lot of information about lifestyle factors, diet and so on. And soon, all of that will be connected to their electronic patient records.

So, you've heard, and it's one of the buzz phrases, about big data. The idea of modern technology tying together lots and lots of different pieces of information. These kinds of studies are an example in biomedicine of big data, where linking genetic information, many measurements and healthcare records will really allow us to understand, we hope, if we can find the right ways of teasing information out of the data, will really allow us to understand a lot more about human disease and the causes of that, both genetically, and in this case because they're measured, environmental lifestyle factors.

So why all the change? I said we had a few examples before 2005. We're now a wash with them. What's happened? Well, there have been changes in technology, actually there have been changes in the way we can extract the information, and, probably more obviously, changes in terms of the biology.

So, here's a graph showing how much it costs to read an entire human genome. Remember I said that the human genome, the first one of these that we did cost several billion dollars. By the early 2000s, the price was down to a mere $100 million, and what you see is a fairly steady – this is a logarithmic scale, so each line here is a decrease by a factor of 10 in price – there's a fairly steady decrease according to something called Moore's law, which you may have heard in a computing context. And then from about 2007 there was a massive drop.

So to give you a sense, between 1990 and now, the price of reading an entire human genome has decreased by a factor of about 100 million. And between 2007 and 2011, it went down by a factor of about 10,000. So it's been a massive change in the ease and the affordability of getting genetic information at scale on people. So we're all aware of the Biblical advice that one should know oneself. I don't think it's meant quite this way in the Bible, but if were meant in terms of our genome, we can now know ourselves rather more cheaply than we used to in the past.

So this is a picture of one of the modern DNA sequencing machines. I said the first version of the human genome took thousands of scientists 10 years or so. One of these will read a human genome in about 24 hours. Current costs in Australian dollars are about $3500, and just in the last couple of weeks there have been announcements of technology increases, which will reduce the price this year to about $1000 per genome.

So that's great in terms of being able to get the information, but lots and lots of data is generated. So, here's an example, this is a kind of a caricature of a genome. Actually, this is one one-millionth of a genome, and we've got two copies of it so it's only half of that. And what you can imagine in these studies, although it's all done by computers with the algorithms that we write, is looking in much, much more information than this in many thousands of individuals and trying to look for patterns that are common amongst the people who are sick with a particular condition or have higher cholesterol or lower glucose or whatever, compared to the control individuals.

So it's pretty challenging at a research level. It's what makes the statistics in this field actually pretty exciting and important. We have to move from the research to the healthcare setting, and enable healthcare professionals to make sense of this kind of information in ways that are rather more sophisticated than these cartoons, I'd suggest. And that's another one of the big challenges for us, I think.

You'll have heard about the idea of personalised medicine, using this kind of information to target medical treatments, and I want to give you a couple of different examples of that and how, in many of these cases, it's actually already here.

So one of them is an area called pharmacogenomics. I've given you the sense, I hope, that many things about us are affected by the genome that we inherit from our parents, the two copies of that genome. One thing that's affected biogenetics is the way we react to drugs, whether or not we have side effects, and also the dose we might need to have a particular effect. That's to do with the pharmacokinetics, the way in which the drug is taken up, how long it stays in the bloodstream, and so on. So many of those things vary between people because of genetic factors. Knowing about the genetics can allow us to get doses more accurately or to prescribe drugs that will work for one individual and not another. So there's a very good example at the moment, drugs against the virus hepatitis C for people who have a particular genetic change in a particular letter in the DNA, there's a drug which works well, and if you have the other letter, it works much less well. Taking the drug is not fun, it has all sorts of unpleasant consequences, so it's good to be able to do the test in advance to work out whether that six- or eight-month course of the drug will work or not.

It's also the case that side effects from drugs can be controlled by genetics, and I'll give you just one example. A drug called Abacavir, which is one of the so-called anti-retrovirals used to treat HIV, the virus that gives rise to AIDS. There's a serious side-effect to this drug. It's called hypersensitivity, which can even be fatal. It turns out that the side effect is pretty rare so the drug was developed, it seemed to be working well and then it was noticed that about 5% of people suffered from this side effect. So if you introduce a drug and 5% of people have a serious side effect, you have to withdraw the drug. Unless you've got some way of predicting in advance which individuals will and which ones won't suffer the side effect. So it turns out in this case that there's a particular genetic variance, something called HLAB 5701, it's got a frequency of about 5 to 7% in Caucasian populations. It's only the individuals with that genetic variant who suffer from the side effect.

So whereas previously there was a drug that works for most people, but not all, and had serious side effects that had to be withdrawn, now because of that genetic understanding, we can do a genetic test in advance. If an individual carries this variant, you don't give them the drug, if they don't, you can, and a drug which was very effective for most people, but would have otherwise had to be withdrawn, got rescued. That's one example and there are a number of others of those, and I think the list will continue to grow.

Another area where personalised medicine is really here already is cancer. So genetics is involved in cancer in two different ways. It can affect susceptibility to cancer. But actually cancer itself, the disease, is all about genetics and genomics. I mean that in the following sense, so, as I said, each of the cells in our bodies typically carries the same copy of our genome, the ones we inherited from our parents. What happens in cancer cells, the cells that make up tumours, is that their genome, if you want to think of it as their instruction manual, has gotten messed up. So there are some changes in the genome in those cells which causes them to behave differently, typically to proliferate more than they should.

So what goes on in cancer, think of cancer tumours as cells whose genomes are messed up. They've got a different version of the instruction manual, and instead of behaving in the orderly way that the rest of our cells do, they proliferate. And that's why tumours grow, and eventually spread throughout the body.

So it's now possible, to read the genomes in those cancer cells. And we can compare the genomes in the cancer cells with the genome in all the other cells in that individual's body and try and look for the particular changes which are driving the growth of that tumour.

That helps us to understand cancer as a process rather better, but for a particular individual it can allow us to target drugs. So, both in terms of research and clinical medicine, it's already having a substantial impact.

So let me try and explain what this picture's aiming to show. What I'd like to do is to give you a sense of how different the genome is in a typical cancer cell, or at least one set of cancer cells, compared to the ordinary genome of an individual. But the ordinary genome is three billion letters long, so is the genome, typically, in the cancer cells, and it would take us a while to go through it position by position. So this cartoon tries to represent that information rather more succinctly.

If we started in the middle and go outwards, each of these purple lines – so the human chromosomes are now around the outside – each of these purple lines represents a very gross change in the genome of the cancer cells. It's a chunk of one chromosome which isn't where it should be, it's stuck onto another chromosome. These green lines in the centre are very big changes within a chromosome. So a large chunk of a chromosome which might be missing, or instead of having two copies of it, you might have three or four copies in the cancer cells. In this particular tumour, so this is taken from a malignant melanoma, a skin cancer, in this particular tumour there are about 37 of these very large gross changes.

There are another 66 examples of quite large bits of DNA that are either missing. Normally we'd have two copies of it, so one or both of those might be missing. Or sometimes, instead of having two copies, you'd have three copies or four copies or five copies in the tumour cell.

And then, around the outside, there's a kind of graphical representation of single letter changes. And, again, in this tumour there are about 33,000 of them. So, I want to give you a sense that it's not just that the genome and the cancer tumour has kind of one letter that's gone wrong, or a little bit of a gene that's changed – they're grossly different from our ordinary genomes. And one of the tricks, and we're getting better at this in the field, is to work out which of these changes are actually responsible for driving the growth of the tumour, and which ones are kind of accidental and not directly related to its cancer biology.

That's the example of one melanoma tumour. I'll show you six more of these pictures. Each of these is taken or represents a genome in the same kind of way of a breast cancer tumour. So, six pictures from six different women. And all I want you to observe, even from the back, is that these are very different.

So there's a massive paradigm shift in the way we now think about cancer. We use to classify cancers according to whereabouts in the body they were. Skin cancer in the skin, breast cancer the breast, colon cancer in the colon, and so on. Now we can look at cancer tumours individually and learn about the genetic changes that are driving that cancer. And actually it might turn out that the genetic changes in one person's breast cancer have got more in common with the genetic changes in someone else's lung cancer than they do with a different breast cancer patient. And so, genetics allows us, now, to try and learn what's gone wrong and target treatments to that.
Actually in breast cancer, for some time that's been the case. Some aspects, not the whole genome, but some aspects of the genetics of the tumour are used now for prognosis and for choice of drug therapy. So Herceptin's a drug you may have heard of that's used in connection with breast cancer. It works on breast cancer with certain genetic changes and not others.

I want to go back to I think one of the most exciting examples of the idea of focusing medicines on particular changes. It's back to melanoma, which of course is a really important disease in Australia. So about 10 years ago it was observed that many melanomas, so think of each melanoma tumour as a set of cells with a messed-up genome, many of them have a particular genetic change. So this is how we derive it in genetic shorthand, BRAF is one of the 20,000 genes in our body. And this is saying that in 600th position in the building blocks which make up the protein produced by the gene, there's one of those building blocks in amino acid, which is replaced by another one. So that was learned as a scientific discovery about 10 years ago. You can think of it informally, so genes produce stuff, they produce proteins and the effect of this change is that a lot of this BRAF stuff is produced. And that promotes growth so that's what causes the cells in the tumour to proliferate. And a fairly obvious idea from a drug therapy point of view is to develop a drug. We can't yet develop drugs which change the underlying genome. We can try and correct the consequences. So the idea here was to develop a drug which kind of mops up this BRAF stuff which is produced too much in the tumours.

So we had the scientific discovery about 10 years ago. The selected BRAF inhibitor was developed about three years ago. And here's a picture from the first paper reporting trials on 15 individuals. So here's one of the individuals in the trial. This is a scan before they were given this new experimental drug. You don't need to be an expert on scans to realise there's lots of stuff going on here. Actually, everything that's not the bladder or the brain that's kind of green or orange is tumour. So this person had very, very late-stage malignant melanoma. They had tumours covering their body. That was what they looked like before the trial. And the next picture shows a similar scan two weeks after this new drug that mopped up BRAF had been given to them. It was a huge change. Almost all of the tumours had gone.

So these are reports from the earlier studies, not quite as straightforward as this. It was a very exciting development, because as you probably know, there aren't many very good treatments for malignant melanoma. So this one showed promise. Cancer and cancer cells are rather cunning things. It turns out that these individuals were followed, and nine or 10 months later, the tumours started to come back because the cells in the tumour had evolved, so that this drug was no longer effective against them. So, it was promising, and, as I said about 10 years ago the scientific discovery, three years ago the first drug trials. At least in the UK, this drug is now a front-line treatment for melanoma. It doesn't solve the problem, but it buys between six months and 12 months as other drugs are being developed, so there's a kind of evolutionary race. The cancer cells are trying to evolve to get around what the drug's doing to them, and we're trying to develop new drugs which catch them when they've done that evolution.

So this is one example, and I think over time there will be many in terms of cancer, of being able to target treatments specific to the changes of individual cancers.

I'll just give you two more examples quickly. So I've talked almost entirely, apart from my little excursion into salamanders and chimpanzees, about genomes in humans. All creatures have genomes. Bugs, germs have genomes. Bacteria and viruses, they've got genomes, they've got genetic information. And they're easy to read. They're actually much easier to read than our genomes because they're much smaller. So, even today to read the entire genome of a typical bacteria costs, maybe, $20 or $30. So, I think very soon, in terms of the impact of genetics in clinical medicine, very soon microbiology will be done by genomics. So someone'll be sick. They'll have some kind of infectious agent. A sample of that will be taken. Its genome will be read. We can look at the genome of the bug that's infecting them. The first thing that we can then do is to confirm whether it is what the doctor thought or not. Is it really C difficile, or is it something else? The second thing we can do is to just look at the genome, and read off which antibiotics that bug will be resistant to and which ones it won't. So we'll be able to talk about antibiotic use much better.

And the third, there are already examples of all of these, the third is that for diseases which naturally spread in ways which are worrying, at a number of hospital infections C difficile is one, MRSA is another one, we'll be able to use the genomes as a kind of fingerprint. And we'll be able to work out that the man who's sick with C difficile in this bed was actually infected by the version of the bug that the lady in the next ward has. Or maybe it's from someone upstairs. So we'll be able to trace, in real time, the spread of epidemics and try and intervene to stop them.

My final example moves on to whole genome sequencing in the clinic. In Oxford, we undertook what was at the time a rather large study of 500 patients in treatment in clinical medicine in collaboration with Illumina, one of the technology companies in the field, to try and work out where, in the short term, genomes could impact clinical medicine. And we studied individuals who had rare diseases, individuals who had cancers, and individuals with particular extreme immune phenotypes.

I'll give you just one example. So we sequenced the DNA of a trio. There was a young child who had, I'll show you some pictures in a minute, really severe problems, both physical problems and problems in terms of intellectual development. We suspected in that case, that what had happened was that she'd been very unlucky, as had her parents, and she'd inherited a random chain. So she had a piece in her, a letter in her DNA that was different from what was passed on from what her mother or her father had, of that position. We tried to find that by reading the entire genomes of the child and both parents.

So here's a picture, this was taken when she was much younger, of the child. You'll see she has an unusually shaped head. And a scan shows that whereas most of us have solid skulls, hers had all sorts of problems in terms of its development and binding together. She had a condition called craniosynostosis. And by doing the sequencing, we are able to find the genetic change, the single letter amongst the three billion that we thought was responsible for the condition. In this case, it didn't directly affect her treatment but it had a number of indirect and important consequences.

So this was the first child for those parents. They were very worried, they suspected and had been told it was probably a genetic effect. They're very worried about risks to future children. By being able to show that the change was one of these chance events that happened in either the sperm or the egg that was passed on by her parents, it followed that if those parents had subsequent children they wouldn't be at risk of the same conditions. So that was hugely important for them and actually there was a very mundane but nonetheless important effect: it gave a real diagnosis. The clinicians were able to point to something which was responsible for her condition, which had all sorts of downstream consequences in terms of bureaucracy and her becoming, because there was a kind of name change and a definitive cause for the disease, affecting her ability to get special help and special needs help in education. And it attracted some attention in the UK as one of the first examples of whole genome sequencing within the clinic.

So within the UK, this idea spread rather more widely. We sequenced 500 individuals. The National Health Service in the UK are undertaking a much larger version of this study and, and this use of genetics in clinical treatment. Over the next few years, there'll be 100,000 NHS patients who have their entire genome sequenced, principally individuals who have either reg diseases, like the example I showed you of the little girl, or cancers. So these would be individuals who are already sick, their doctors will say to them, look one thing we could potentially do here is to read your genome, or in the case of all individuals with cancer, your genome and the genome of your tumour, and use that to prescribe or to adapt clinical treatment. Individuals will have the chance to consent or not. And the first possible outcome, we would hope, is that for at least some of them, it'll impact on their clinical treatment in positive ways. And also, for individuals who consent, their data, their entire genomes and their health records, will be collected together with that of others as a research resource to enable us to learn more about diseases and the role of genetics and other factors.

So I hope I've given you some sense of the excitement in the field of genetics. There are lots of exciting changes in the times we live in, in terms of technology. Things like smartphones, laptops, iPads and so on, we probably didn't imagine 20 or 30 years ago. Our understanding of the physical universe has changed dramatically, whether it's at subatomic scales, the Higgs boson on so on, or galactic scales with our understanding of the early events in the universe, that's changed as well.

But I think when we look back on the early years of the 20th century and the developments, it'll be genetics and the changes in our understanding of genetics that stand out. It'll be, I think, seen as the time in which we first really started to understand human biology, and human disease biology. When we first started to understand the language in which our DNA code was written, the language of our genes.

Thanks very much.