Somatic Mutation Rate and Cancer Genes
Harvard Associate Prof. Shamil Sunyaev on imaginative biochemistry, identifying oncogenes, and nucleotide exci...
The video is a part of the project British Scientists produced in collaboration between Serious Science and the British Council.
Behavioural Genetics has been going on for over a century. The modern era began in 1960 with a book called Behavioural Genetics but people have been interested in nature and nurture for recorded history. Animal breederы knew that you could breed for behaviorб and most animal breeding involves behavior as much as it does biology. The dog breeds are mostly bred for functional behavior – for very specific things like pointers point and scent hounds smell. It is hard to get another breed to do that: you try to get a collie to point, they won’t point, or to get a pointer to round up sheep. So for a long time we have known that genetics is important but psychology rejected it for humans. But after a century of research things have changed dramatically. Most psychologists now accept an important role of inherited DNA differences – so much so that it is no longer interesting to say: oh, here’s another trait, another measure, let’s see if it is heritable. You can save time by saying: yes, it is, it is going to be.
We have to ask more interesting questions and the big one is – can we find some of the genes? That has been a revolution and it is just now reaching a point where we are identifying some of the genes to be able to make individual predictions. This is not just psychology, this is all of the life sciences. They have all come to this position. Everything is heritable. Let’s find some of the genes, because if you can find the genes, you can understand pathways between DNA, RNA, metabolism, proteins, brains and behavior. The original goal is to find some of the genes so that you can understand the mechanisms between genes, brain and behavior. But what we have learned from these new strategies just came out really in the last 10 years or so called Genome-wide association studies (GWAS), where you have a little chip of the size of a postage stamp, and it allows to genotype hundreds of thousands of DNA markers throughout the genome, which are the 23 pairs of chromosomes, 3 billion base pairs of DNA. You can tag it throughout the genome, so genome-wide association studies is an atheoretical approach to find the genes for heritable traits.
The biggest thing that has come out of that is that the biggest effects are much smaller than anyone ever thought. The biggest effect we know for a complex trait is one gene called FTO which predicts body weight, corrected for height, called body mass index. That accounts for about one percent of the variance in weight but that is by far the biggest effect that has been found. One percent translates to this: it is a stretch of DNA that has two forms: it is called A and T. If you’re AA, you are three pounds lighter than the average, if you are AT, you are average, if you are TT, you have three pounds greater. So it is like a six pound difference, that is 1% of the variances, and that is the biggest effect. The average effect sizes are less than 0.01, it is really like 0.1% of the variance. This is an incredibly tiny effect.
What we are guessing now is that the smallest effects are very much smaller. If we are talking about: here are the biggest effects, the smallest effects account for most of the heritability. What that means is we are not looking for the gene or even a handful of genes. We are now talking about thousands, tens of thousands of genes that we put together in what we call a polygenic score, i.e. multiple-gene score. We are now beginning to use those to predict genetic propensities in a population. So for height we can now explain with DNA alone 20% of the variance, with weight about 15% of the variance. That means that it is possible now to predict that difference at birth about an individual. In psychology and psychiatry we are nowhere near that, with schizophrenia people say that they can explain about 10% of the genetic risk with DNA alone. They are called chips, these little postage stamp sized slides in a way that you use to genotype DNA that you just get from saliva. To detect these small effects you need huge samples, so the sample sizes are in hundreds of thousands. So you need hundreds of thousands of people who have been genotyped on hundreds of thousands of snips, as we call them, i.e. single-nucleotide polymorphisms. It is just the DNA variant and that is what we need to be able to detect these tiny effects. But when we get up to those sample sizes, we are detecting those effects.
So for behavioral traits that I study, like intelligence, six months ago we were able to explain 1% of the variance. We can now explain 5% of the variance. With educational achievement, though, we can explain 10% of the variance. In England we have these tests called GCSE which all children at the age of 16 at the end of compulsory school have to take, national exams. We can explain ten percent of the variance on that. We just got started and we can explain 10% of the variance, it is about 60% heritable, so we have a long way to go. But with 10% what it means is if you take the top fifth in terms of these genetic scores that we call polygenic scores, the kids who are in the top fifth are 65% likely to go to university, the kids in the bottom fifth are 35% likely to go. Even with only explaining 10% of the variance that’s a pretty big difference in terms of predicting at the high and low extremes.
I am really into this, because I want to be able to predict genetic strengths and weaknesses at the individual level. It is not like family risk: if you had a first-degree relative who is alcoholic, you have a five-fold greater risk of being alcoholic but so does your brother because it’s just the family risk. That does not work for people because there are big genetic differences within a family. You and your brother can be genetically quite different. But with DNA you are able to make up specific risk. We could say you are at risk for becoming alcoholic, your brother is not. What that would mean is if you guys drink the same amount of alcohol you are going to become alcoholic, your brother won’t be. In terms of the genetic prediction, it is not all genetic but it is saying that genetics can make a pretty good prediction about some of these things. When we do that, it will change research because it is so cheap to do this now. These chips I talked about – when I first did it, they cost about 3,000 pounds. They are now 40 pounds. It is cheaper than much of the stuff we do in psychological research. You need big samples to find these polygenic scores but once you find them, you can use them in small studies, like in neuroimaging studies or in psychological studies.
It is going to change the face of research, because we are going to get DNA on everyone. Francis Collins who is the director of the National Institute of Health and was a director of the Human Genome Project says that within a few years all newborns will not just be genotyped on a chip, their whole DNA sequence of three billion base pairs will be sequenced. And you can put it on a little memory stick. Once you have done that, that is the end of genotyping ad collecting DNA. You do not need DNA anymore, everyone will have all their DNA bases on this memory stick ,and then we will use that then to create these polygenic scores to make genetic predictions of risk.
For me especially, what I am interested in is reading disability and reading ability are highly heritable. What we do now is we wait till the kids to get to school, they fail at reading and then we try to do something to fix it. You could predict which kids are likely to be at risk and we already know that almost all kids who are reading disabled had language problems earlier. You cannot do reading interventions in three-year-olds because they cannot read, but you can do language interventions because there are very good language programs. But an important point is that most successful interventions are expensive and intensive. So you cannot do it with everybody, but if you could identify the kids who will be at risk for reading, you could intervene with language and it would not hurt them. It is not like giving them drugs or something, it’s only good to have people helping you with your language. And then when they go on to school, they will be less likely to have a reading problem.
So all of science in medicine is now focused on predicting and preventing problems rather than waiting until the problems occur and then trying to fix them. This is beginning to happen now. People are doing it on their own. 23andme is this company that will do this genotyping for about a hundred bucks. So lots of people are beginning to do it for their children. The reason I think it is important to get this message out now before it really hits the ground. Explaining 10% of the variance is a lot. If you want to predict school achievement, what we would you use to predict that? Nothing else will get near this 10%. But we just begаn, people do not even know about this yet. But when they do, I think, they will begin to take genetics seriously in education. Right now they totally ignore genetics. There are some very good things, good discussions that ought to begin now before this really becomes reality, even though it is getting close to reality already.
A lot of people say how terrible this is, imagine, you’re going to predict which kids have reading problems. I think it is good. Now there’s a lot of doom-mongers who say “you label kids”. Kids get labeled already. You know, there are problems. We got to discuss those problems. But I am a cheerleader for this, I think, we can do a lot of good at predicting and preventing problems from occurring. But even now before we find the genes, we have to start realizing that genetics is very important. It accounts for far more variance than anything else. And what that does is it makes us recognize that people are different genetically, that it was not all due to learning, or willpower, or personality. It is in our DNA differences from early in life and when we recognize those differences we might respect them more.
So I think, one message that comes out of genetics is tolerance. In the UK weight is a major issue, the obesity epidemic. And people are seriously saying that overweight people should be made to pay for the disorders that occur, that I have to pay for their national health service, that it is their fault: they are fat, why don’t they just lose weight? But if you recognize the genetics, 70% of the differences are due to genetics, you realize that it does not mean you cannot lose weight; if you stop eating, you lose weight. But if you are thin, you do not recognize how difficult it is for someone with a genetic propensity to put on weight and to have a difficulty losing it, it is just a lot harder. What concerns a reading disability, it does not mean that you are not going to teach kids to read. It just means some kids are going to find it much more difficult and we need to recognize that instead of blaming the education, blaming the schools, blaming the parents, blaming the kids. And, with obesity, blaming you for being a slob – you know, just get a grip, have some willpower here, lose weight! And I think, if we recognize it, that it’s easier for some people to say that, we might become more tolerant. But at least what it will do is have us realize that people are different and that suggests that medicine, you know, we talk about precision medicine – that’s basically genetics. Not prescribing drugs one-size-fits-all, not having a national curriculum one-size-fits-all, but recognize that people differ genetically and they need interventions that are personalized to them. It is one of the big messages that come out of this even before we identify the specific genes that are involved.
Harvard Associate Prof. Shamil Sunyaev on imaginative biochemistry, identifying oncogenes, and nucleotide exci...
Computer scientist Erol Gelenbe on the energy consumption in cloud servers, fog computing and why do data cent...
Historian Peter Jones on monastic education, cathedral schools and apprenticeship in the Middle Ages