So why is it important to continue to study these orphan rare diseases, although they affect a very few number of people in the world? Well, so maybe the first thing to say is that these rare genetic diseases which in French we call orphelin, for orphan, which is in fact describing the situation of these few families which are afflicted by these diseases. So it's good to remember that cumulatively, these disorders are a major public health problem and human with economic, but mainly human costs attached. Now, it's cumulative because individually the majority of these diseases are as their name implies, very rare. But I think now that the technology is improving rapidly to be able to unravel and understand what is happening, I think we owe it to these poor families to use that information to as effectively as possible give them clarity about what is afflicting their families. In a few cases, it may actually lead to effective treatments for this family. It will be the exception, but it will happen as it has already happened. So I think that's certainly also a very good motivation to do it. From a more fundamental point of view, it's a very effective way to try to determine the function of a growing number of genes in humans. So model organisms, are only model organisms. And although they will bring clarity on the function of a number of genes, because their function and their phenotypic effects are conserved across species. We already know that for a significant fraction of our genes, what you obtain by mutating a gene in the mouse, doesn't tell you a lot about what is happening when a similar mutation afflicts a human. So I think by doing that, not only do we fulfill I think a medical obligation and a societal obligation towards the families that suffer from these diseases, but in addition to that, I think from a scientific point of view when in the sense that we learn who we are and how our organism function, I think studying these families is a very rich source of information. These monotonic disorders are actually variations in nature, that give you a very severe phenotype. But that also cuts in a pathway like at one position. If you know that you have a specific part when you cut it at a specific position you can easily study the pathophysiology of that single knockout. Whereas if you would look at the multifactorial disease at the same time, that same pathway would probably be kind of interrupted at different stages and you cannot study individual functions of individual genes. And many times we have learned from families with very rare diseases. We've learned about basic fundamentals, well fundamental processes in nature which we had no clue they were existing. Take a cystic fibrosis gene, no one had an idea what was causing cystic fibrosis until cystic fibrosis hit the gene and said, oh but this is a function of a protein that has a role in Chloride transport. And I can give you tens of different examples like these. So yes they're important. That's indeed true. There is about 20,000 genes in our genome, and we know about 7,000 rare disorders. So there is some of these genes which we don't know yet what they would cause as a disease. Is it because we haven't looked at everybody yet? Could be. Because there is probably still that rare families around that have diseases or defects in new genes and we'd still find new genes every day. But there is probably a more fundamental question about this, and which is that some genes or mutations in some genes will probably not be compatible with normal life. So if you hit a mutation in that gene, there will be no human being with the mutation. Because it will be little very early in life, or very early during pregnancy. And there is examples of this as well where genes are just not compatible or mutations are not compatible with normal life. An indication of this is that some genes are not polymorphic at all. So, it means that if you twist that gene, or if you mingle around with that gene, you just kill its function and there will be no human being with that disorder. The other thing would be that maybe some genes compensate for deficiencies. But then it's hard to imagine why a gene has been conserved for millions of years to then easily be compensated by another gene. This is possible but it's rather unlikely. So, I believe that eventually if we can sequence many more people, if we can find variants, and then say okay the variants are there but we don't see offspring with those variants, then we can say this is the little form of the disease. Or if we miss this polymorphic variation, we will also say this is such a gruesome gene, that it's not compatible with mutations at all. So, why haven't we yet found more causative genes, underlying or contributing to the inherited predisposition to complex disease in human? So, I suppose what it tells us is that it's not the trivial thing to do. It's a relatively difficult thing to do. And I think also, what it tells us is that it is- that we tend to be impatient. So, the first genome-wide association studies, were essentially completed around 2005, 2006, that's 10 years ago. And although it may look long for the people that are involved in it. I think if we come back in 20 years, it will look like a very short period of time. But to be more specific. So, I think that first of all, what GWAS have told us, have shown, is that the genetic architecture of complex traits, very rarely involves genes with major effects. So the majority of the effects that we have identified are correct trials by odds ratios around 1.1, and even less. Which means that you need very, very large cohorts to be able to uncover them and detect them comfortably, which is required if you really want to dissect a corresponding risk lower size. So, one of the things that has taken time, is to accumulate the case control cohorts of the size that would allow us to do that, and that's in fact, a process that is still ongoing. So, the second thing is that, there is no, to my knowledge, no, for purist's and especially people that like genetics, there is no formal test of causality that can be readily applied to human populations, contrary to what can be done in model organisms such as Drosophila /Frances. So the reciprocal hemizygosity test, which is a beautiful test of causality of a causative gene underlying eQTL. Let's say something that is not directly applicable in humans. So today, the only formal test of causality is one form or the other of a burden test, which is to look for a differential spectrum of mutations, typically rare mutations in cases and controls. And for reasons that are unclear at the present time at least to me, this test has not really proven to be effective in human genetics. So, the big centers in the world have re-sequence at least the exol of tens of thousands of cases and controls. And to the best of my knowledge, this has not yielded as much as what people were hoping for, and one possible reason for that could be that, the only gene space where people can effectively look at the present time is decoding space. And I think it's not impossible that this differential spectrum of mutations is actually going to affect the gene switches, rather than the coding part. And one possible explanation for that would be that, it is the gene switches are tissue specific, while the x zone space is not tissue specific. So mutations, in the protein coding part would affect the function of the genes across all the spectra of tissues. And I think it's not impossible that these complex diseases are due to tissue specific perturbations of gene function, which will require mutations that act in a tissue specific way. So we may have to start looking at the regulome, the regulatory space, the gene switches, and at the present time we are not yet armed as well to do that, as we are for the protein coding part. So in a nutshell, I think it will increasingly become the attention of a growing group of very talented scientists that will certainly in the near future find new ways on how to do that most effectively. I really think it's the frontier of the study of common complex diseases which is attracting a lot of attention now. And I'm very hopeful that things will change in the near future. Well, first of all I must emphasize that, I strongly disagree with the statement that success rate is low. Well, probably this is because I started working in this field about 20 years ago, and in 1995 it was kind of holy grail for complex disease to reliably find the genes or even genomic regions which would be associated, or linked to complex disease and which would replicate and you could see that in one study, in all the study. And up until the 2005 or 6, we couldn't find much. And actually, if you think of majority of complex disease, if you think of quantitive traits, before 2005, we knew only a handful of genes and variants which would be reliably associated with this disease and traits. Really a handful. Right? I could count just a few. For type 2 diabetes it's probably three, and for common height population, population aberration of height it was zero, and so on and so forth. So, the progress we've made is totally incredible. Again, thinking of type 2 diabetes, it's more than 40 genes now. If you think of height, it's hundreds of different size. So the progress has been incredible. Now the question is like, why didn't we find all the variations which is involved in the control of, well, we can think of any specific complex trait? Why we don't find everything? And I think the question is actually almost fundamental. So imagine that, so first we can think like, we don't have yet sample sizes big enough which are powerful to resolve very, very small effects. But now think of this. human population is finite, it's huge. it's in billions, but it's finite. So even if you had genomes and phenotypes of all the people on earth, you still can postulate a very small genetic effect, which you have no power to find. Right? And you can postulate easily, that those very small effects exist. So basically, even theoretically, there is a limit to the genetic defects which we can identify. But now the question is, well actually do we need to know everything? Because there is this theory of ubiquitous pleiotropy which tells that well any variations in genome affects every trait, right? I think that the key is that we don't have to know every variant which is associated with every disease, right? And again in principle, we can postulate that every variant is associated with every disease, although the effect may be very, very small. So, I think you don't need to know all of them. You just need to know enough, to understand biology behind these traits. So the question is, whether we could use genomic information to predict disease. This is a difficult question and answer is actually yes and no. So we could think of two extreme scenarios, and one scenario is of monogenic disease, and there we can predict the risk of development of such disease very well using genomic information. And on the other side of the spectrum, you could think of some trait with negligible heritability. So it's almost entirely on the control of environment, and it's quite clear that for such trait or disease, use of genomic information is well, almost of no use. However, all the real situations are somewhere in between. And it depends on the heritability of the trait how well we could in theory, after years and years of research and studying all these finding out, all these genes which are controlling this trait. But the heritability is putting the upper limit of your predictive power. Well, one thing this is interesting, and this is something very unique for genetics, is that genetics provides you with very long term prediction. So you can take a baby which is just born and try to predict the risk of development of certain disease by the age of 50. And in a way this will be a right prediction, but it will be also very inaccurate prediction for most of the situations. But interesting thing is that, it's very long term prediction. And then you can think that, so now, we developed all of this mixed techniques and we can assess the levels of thousands of different molecular species in organisms. And you can think of constructing biomarkers which would be predictive of the short term risk. And we already have a good examples of that, classical examples, like fasting glucose level is providing very good prediction for the development of diabetes in the next five years, or also, very well-known example, cholesterol levels for the risk of cardiovascular disease. So you can think that actually, you would like to smartly combine the knowledge of genetic predisposition with the system which would assess the levels of current risk using some panels of biomarkers. Right? And smartly combine it in a way to predict the risk of development of disease. However, I think that actually, the main value of genetics for human disease is not in the risk prediction. At least in the long term, I think way more benefit to humankind if you wish, is from use of genetics as a tool to understand biology of this disease because this can really dramatically change the picture. We can come up with new strategies for treatment of this disease and prevention of this disease. I would instantly think about cancer in this case. Because even if a cancer occurs in the same tissue, it may have a different genetic cause. So some cancers will have mutations in, aka rest or unrest, other mutations have other cancers type mutations in p53. And depending on the mutation, we can now direct specific therapies. So it's very important to sequence cancers, to know which therapy, which track will work. And also, if you then repeats the sequencing after some time of treatment, you can see whether the cancer is escaping the therapy and you would be allowed to redirect to another drug. So from a therapeutic standpoint in cancer, that genomic information is huge and important. So the future of genetics. While this is a very broad topic and of course, if I try to answer it, if I try to play a prophet, I'm going to be biased towards the position I'm standing in. But if you think of this, then like two major classes of problems in genetics are, so the first problem is, you can call it a problem of ontogenesis, a problem of development, or a problem how the genetic information translates into an organism. Right? And this organism has developmental trajectory, it grows, and then at certain point, it start aging, right? So how the genetic information on the individual level translates in all these traits and some of these traits are diseases. So this is one from the mount of think. And then other fundamental thing is of course, is about evolution. Right? So how does the genetic information now, not in the context of specific individual, how does it change with time? How does it evolve? Now, I'm trying to be a little bit more specific. I'm thinking of this area of genetic analysis where we study how genetic information is translated into phenotypic diversity, into phenotypes. I think that is one very important step which we are going to make, I hope in the future years, because if you think of this in the area of complex disease genetics and complex trait genetics, quantitative trait genetics, may have accumulated a lot of knowledge. We performed a lot of Genome-Wide Association Scans and we have all these regions of associations. However, for most of these regions, we have very little idea how these statistical results, all these peaks and P-values, how do they translate into biology? So we really need to understand better the function, we need to understand better how it works. So this I see as major next step in the development of human genetics for the future years. I have a very specific view about this. Maybe not specific in the sense that it's mine. It's shared by many people. But I think genetic will become the basis of medicine. There will be genetics in all different specialties of medicine, be it cardiology or fertility. I think we will sequence a lot of patients, try to explain why they have this disease, and then respond accordingly. This will be called genomic medicine. The term already exists. What we are waiting for is to see this kind of introduction of genetics in all the different specialties. I strongly believe that this is the future of genetics. It will become genomics as we call it now. And I also see that increasingly, we are sequencing patients. We can now sequence entire genomes with the novel technologies of massive parallel sequencing. So eventually, I believe that everybody will be sick as a bird. You may say this will not happen. People will say, "What are you talking about?" It's not my opinion, it's what I anticipate that will happen. I think society will want us to sequence everybody at birth, to try and predict disease, to try and prevent disease, and so forth. So let's say that in 10 years from now, genetics will be in every house not just in medicine. So it seems to me that resequencing our genomes, that's certainly imminent revolution. And it's primarily, I think we can say that because the cost of sequencing has already gone down spectacularly and because there is no reason to believe that this tendency is going to stop. So the cost of sequencing a genome will be of the same order of magnitude if not cheaper than many of the medical tests that clinicians are relying on a daily basis today. And I'm confident that there is sufficient information hidden, if I may say, in our genome or genomes that it will justify a nearly systematic use of this technology. So I think we can say with near certainty that our children will all have their genome sequenced on multiple occasions during their lifetime. So this will primarily be used, I think, to address problems with newborns, so severe problems with newborns that are possibly due to inherited, to mutations with major effects. It will be used extensively, I imagine, in the context of cancer. So I think we are rapidly approaching a time where the DNA of every tumor will be sequenced because in a number of instances, it may suggest therapeutic approaches which are personalized, which target the genetic defects that characterize that mutation. Where it's more difficult to know where things will go, is for multifactorial complex traits and complex diseases. So, I think that in a second wave of GWAS, the phenotypes will not be the diseases per se, but things like response to treatment. So a former pharmacogenomics, evolution of the disease, and that kind of information may help the doctor to make choices, make decisions about how to approach a given patient, what treatment to choose, whether to go for surgery or not, for instance, in the case of inflammatory bowel disease. So I think there will be a number of clinical decisions that will gain from the genomic information. With regards to prevention of disease and management of disease, there in fact, my view at the present time, is that the information from the individual is very, very difficult to use. I think that if we were to look at our individual genomes, we would all see that we are a bit at higher risk for disease A and a bit at lower risk for disease B. And so if you go through all the diseases, we'll all be somewhere equal with few exceptions, where all of us will have some risk that is increased compared to the mean and some that is decreased, and I doubt that it will easily affect our behavior. I know that I should eat less and do more sports to live longer, and I don't do it and I don't believe that reading in my DNA is going to affect my decision. So from the individual, it is something where I think the information is, for most of us, of limited value. The difficulty is that for the group, for society, it could be an information that can potentially be used in a way, for instance, to be more cost-effective in the way we handle our dollars, I want to say, or the rubles that we spend on our health as a community. So, if we could stratify the population by the risk to different diseases, we may if we have economic strains on the system, decides that we reserve some diagnostic screening, preventive diagnostic screening, to the people that are at the highest risk of developing the disease. And I think mathematically, one can show that this may ultimately save quite a bit of money for the society even if it is with methods that have low sensitivity, which means that you will miss a lot of people who you should have given this option, and that you will actually force the tests or allow a large proportion of people to do the test, which in fact didn't need it. So despite the fact that this test will not be perfect, when you average it over a large population, you are going to save money. So I think that the difficulty will be to balance the interests of the individual versus the interest of the group. And it is likely that different societies will actually approach that in different ways. And we can just hope that it will be done in the most ethical way possible.