 Okay, welcome everybody back. Nice to see so many of you again. It's a very cold and frosty morning here in Scotland. Yeah. So, I'm sorry. Let's start with the lecture just in a second. I'm sharing my screen. Okay, do you see the screen? Yes, that works. It works well. Yeah. Okay, perfect. Thank you. Okay, so we had two lectures already together. So the first one was about biological background of epigenomics. And I hope you enjoy that and you saw the opportunities in epigenomic research. I think it's a super interesting, fascinating area to study because it's on the one hand really relevant to how cells work and it's still quite mysterious and there's a lot to be discovered. So in case you're searching for a research topic, I can really highly recommend epigenomics. So on the second day on Tuesday, it was a lot more technical and I showed you some of the approaches that we are taking using machine learning for imputation and data analysis. And I think we really looked into some of the challenges in interpreting and analyzing these very, very complex data sets. But it's also very fascinating and you can apply all kinds of new techniques and think about how you model the data and so on. So, so really interesting work. Today will be a little bit different yet. So we had biology we had a bit of machine learning. So I want to put it into a bigger wider context and it's also being going to be a little bit historical and I'm a bit nervous about this lecture I have to say it contains some of the material I used in a recent genetics lecture. And I think it's it's highly relevant as well for us as scientists to think about these bigger questions. So I started a little bit chokingly as an introduction in the first lecture with this slide with my DNA with a bit of DNA sequencing. So if you introduce someone, you tell them a little bit about their past. And of course some people would think that a lot of information is already present in my DNA. So if I want to introduce myself to you I could show you my DNA. And that's of course just a choke, but it's not just a choke because you all know that nowadays you can get these kind of tests for relatively little money. You can send away a sample of your genome of your, you can send away a sample and get your sequence genome sequenced, and it doesn't cost you very much. And then you get a whole bunch of information so this is all done commercially and routinely now and there's lots of genomes collected and stored as well. So this is your private data and the results that you would get from such a test include a number of different aspects. So what does the genome tell you about me well there's the first very obvious questions I have to X chromosome so apparently my sex would be female. There are some traits that can be very easily picked up in my genome so for example skin back the pigmentation. The skin pigmentation is inherited in our other complex way and we don't quite understand all aspects of it yet. There's lots of genes involved in the inheritance of skin pigmentation but we can make a very good prediction given our genome what kind of color our skin would have. We could also get something, some information about how likely I am to consume coffee, I can confirm that. So I have to say, these results are actually I haven't done this test for myself. This is educational material that I'm presenting you with here. And of course there's some information that I am disclosing to you, not based on my genome. But it's actually quite true I do like to have my to consume my coffee. I do like good coffee so that's I would like to do PVC there in Italy but Scottish coffee is improving. And of course there's some more serious stuff that the genome tells you about me so there's a genetic health report mainly about risks. What kind of diseases could I be developing in the future. So for example, Parkinson's disease to have certain variants to have a bigger risk of developing this disease or late onset Alzheimer's disease and so on. But it's not only about me of course that this information that this data contains information about. So it goes way beyond me so it also tells you about what has been there before me. So about my past. It tells you about my ancestry, it will tell you that probably mainly German, mainly European but there's probably quite a lot of other ancestry mixed into it. And what could that be. It also tells you something about the future. So about potentially what happens if I would have children, and do I have a carrier status for some certain for certain diseases. So it tells you that my the genome tells you something about me in the present, something about me in the past my ancestry where my people came from, and it also tells you quite a lot about potential future generations. It's a very powerful data set and some people would even argue that this deep sequence with sequencing and deep learning techniques, you can even read more out of these genomes so they would claim that even before you're born you can make this prediction about for example IQ and in relation to that even fade outcomes life fade outcomes. And of course, I'm quite massive and that's that's huge big claims. But it's true so we inherit each one of us are huge library of information, which is to some extent hidden in our genome but in one more we can, we can. It's not as hidden anymore as it used to be. And when I started talking to you about epigenomics I would also introduce epigenomics as the question, what of these books do we actually read so the cell cells never access the whole of the DNA they only access certain programs of it, and that determines the particular function and phenotype of every given cell. So I've said it really matters with which book we read in order to determine which carry which phenotype with character. The cell eventually takes on whether it's a neuron cell or muscle cell or fibroblast or what have you. So it matters which books we choose to read or which books we have an opportunity to read. So of course there's a lot of information about me. That is not stirred in my DNA. Okay, so for example, about my location. So that I can disclose here that I grew up and went to school in Germany in the 1990s. I studied physics in Germany and France I hold a PhD in computer science and machine learning. My main research focus as you have guest is computational epigenomics. I have been living in Scotland for the last 10 years that was very influential on me and I'm also a parent. None of this information is stored in my DNA. And of course there are other things that are really important to know as well. It's like do I have financial interests, what is my available income, what are my housing conditions in which I live, what is my face political orientation and also. So there's a lot of things that make me make up who I am, what kind of things I'm interested in. And it's not just my genome, I think we can agree on that. So why does it matter? Why does it matter? Why should I be talking about me in the first place in our science class? So there was a very interesting article by Angelina Saini in Nature, I think it was Nature, it was a commentary in 2020 and she asked, do you want to do better science? Well, you have to admit that you're not objective, that your science is not done in isolation, that you come with some baggage and if you admit that you're not objective you will be doing better science. So what does that have to do with that lecture? Well, I think it has to do a lot with the lecture because the question we ask as scientists and the answers we find are shaped by who we are and the roles we play in the world. And I think it's very important to understand that. It is also important to understand that what I teach is to the best of my knowledge based on well-established facts and reproducible science. But how I presented to you the selection that I make, the ordering, the weights I give to different points is very highly determined by my own framework. So as you follow my teaching, you will see the scientific facts through the lens of my filters. Okay, so I cannot claim that I'm giving you objective scientific facts. You will see them presented through my filters and that's true for all of the lectures that you've been hearing in this winter school it's been true for my epigenomics lecture and also for my understanding of deep learning interpretation of the data. And that brings us to another question, of course, what about data? So if you're kind of taking the human factor out of the picture, then surely we can do objective science, right? And so one of the very influential books for me was this book by Judea Pearl, the book of Phi and where he's trying to move beyond correlation and introduces causality. And he says in that book, data alone are hardly a science regardless how big they get and how skillfully they are manipulated. So, and the reason for that is of course that the very act of measuring and collecting data involves interpretation and value judgments. So this problem about objectivity cannot be completely removed out of our study subject if we are just following the data because we are still measuring, collecting and interpreting data. So the human factor is very important here. And I think it's very important to keep in mind also that when we are studying genomics or epigenomics, we are really studying not just a disease, not just a cell type, not just a mechanism, but to some extent we are our own study object. So it has a lot to do about who we are in the world, how we connect to others and so on. So it is very important that we keep in mind that in this area in particular, we come with a lot of, we come with a lot of judgment and values and we have to question those if we get along to do better science. So, I think, apart from our personal perspective, our indians, the fact that we are, we are approaching science as individuals, there's a history, and we have to also think about the bigger picture that science doesn't happen, that we are not isolated, but it happens in the context of the past, of past scientific events, past historical events in the context of politics. And again, so science cannot be treated completely independently or independent of that. So, I have already said, we have been talking about genomics and all the data that is available, individual, the personal information that is stored in our given epigenome. Now we are looking, I've been looking at epigenomics. And for example, we are trying to find maybe epigenetic biomarkers to predict diseases or to, to use them as their poetic targets. And these are, you offers huge opportunities, of course. However, we have also come to understand that the epigenome to some extent is potentially even more powerful than the genome, because it does incorporate sources of variation that are coming from the genome. So they include lifestyle components. So for example, our individual decisions, whether we smoke or not smoke has an influence on local epigenomics of cell types. It has an environmental exposures, can change our epigenome, these can be individual exposures or having to do with choices or things that we can't avoid. We also have the genetic background. So all of these together play a big role in epigenomics. And then we have also scientific limitations. So in many cases it's really difficult still to infer causality. So to invoke causal associations and not really significant statistical associations. And it's really hard to distinguish drivers versus passenger AP mutations. It is also really difficult to establish a reference epigenome. I think I have that in the next slide. So as I have tried to show you is, we do not have a unique epigenome to our lifetime. We have every cell has its own epigenome. But in addition, it changes over time. It changes due to aging, it changes due to exposures to environmental stimuli or pathological situations. So that means that this is quite a challenging topic. In addition to that, usually we are trying to compare a normal state versus a pathological state in order to infer causes to some extent. And here it's really difficult if we don't have a unique epigenome during our lifetime. And it is also quite difficult to constitute what what is a normal reference epigenome. So we have talked about encode and encode challenges to predict our reference epigenome. We found that it's actually quite difficult to validate that these, which of these data sets are the correct ones. There's other initiatives, for example, the International Human Epigenome Consortium, which is also a global consortium with the primary goal of setting up high resolution reference human epigenome maps for normal and disease cell types. So there's a lot of data sets already available. But in order to make to make this jumping into to make the translational jump to make epigenomic research really valuable for the individual. And we need to generate huge large volumes of personal data that currently is a common that would be accumulated in secure databases. And with that, of course, we do have to think also about potential harms of misuse. I don't want to say that I have already said in the beginning, I think epigenomics is an amazing field to study it's amazing research field. It has amazing opportunities. And of course, I think as we go along, we have to also consider, is there other potential harms, and is there potential misuse cases, and how can we, how can we inform the, how can we design the our research questions and frame it such that we can avoid most of these harms. And we have already seen in the genomics in the in other areas where machine learning is used that we do get harmful and miss harmful consequences and misuse. So for example, this is a 2020 paper where they show that when you use. So an algorithm that is used very widely in the US, which affects millions of patients, and that this algorithm exhibits significant racial biases. It is potentially important also to know that these biases arise not, not just from the biases in the data, but also from the way the algorithm is, is, is the purpose of the algorithm and the training of the algorithm that bias arises because the algorithm predicts healthcare costs rather than illness. And, and that generates an inequality of opportunity for already disadvantaged parts of the communities. So, how does that look like in the field of epigenetics. So, this is a growing field of research, but it's, it's not as well studied yet I think also the concerns are also present and potentially bigger, even then in an in genomics. So I want to say the purpose of this lecture is not to give you any answers. It's not to, it's also something that I'm only started to be interested more or less recently so I don't think I could give you any of the answers. But I would like to invite you in particular when you come from like myself from an area, which is potentially more like on the computational side or the physics side. To actually think beyond the technical details of your work, and to start thinking also about wider implications of our research and work and how can we discuss and work together in an interdisciplinary way to maybe solve some of these problems. So, most of the questions have to do, I would say, with a couple of factors some of them are autonomy, privacy, equal opportunities and responsibilities. So very big questions that need to be answered. And I'm giving you a couple of examples where these, these questions could come into into place. So the one example would be assume that you have an epigenetic tests that can show that there's an increased risk for lung cancer and this increased risk is direct consequence of the choice of smoking. Okay. So in this case, you could argue then that the health system might want to account for patients responsibility on this action and you would put the responsibility to some extent on to, and then you would might want to put the responsibility towards the patient. In another scenario, you might have a fine, you might find that a specific environmental toxins like pollution plastic components used of pesticides or hormones and food, all influence disease onset, via epigenetic, epigenetic mechanisms. And in this case, the question then would be who's liable for such exposition. So is it, is the policy makers, is the public more, is it more important, should they have their bigger responsibility to protect individuals, the society and potentially even further generations. So these kind of questions have a lot to do about, can we evaluate the consequences of lifestyle habits on health through epigenetic biomarkers. So can epigenetic biomarker or I think we can do that. So we can use epigenetic biomarkers in the future to evaluate the consequences of lifestyle habits and health. So, and that are then very important question of rights which which touch on the question where's the limit of individual responsibility of their own, or even the next generation's health. There are also other problems I think in these data sets. So the social and political structures could influence the risk of epigenetic based diseases. So in this case, low socioeconomic classes are most likely to be mostly epigenetically in a most in a more epigenetically disfavored situation. And then in that context, vulnerable populations are at higher risk of social discrimination in inequality in the universal access to epigenetic based medical care. So these are really big questions to answer for the future as this field is majoring and moving on. I want to say, maybe there's, there's already some protection in the sense of how our data sets are being used. And as scientists, I think we very often despise some of these constraints and mechanisms that slow down our research. They are really roadblocks sometimes to access data. But I think they have very important functions. So one of these things is the general data protection regulation GDPR, which was introduced by the European Union, and it really is meant to protect the way data is used. Based on the charter or the principles are kind of derived back from to the chart of the fundamental rights of the European Union. And it comes back to the question and the there and in particular article eight of that, which concerns the protection of personal data. And this is important in that sense, because it says that everyone has the right to the protection of personal data concerning him or her. And such data must be processed fairly for specific purposes on the basis of the consent of the person concerned, or some other legitimate basis laid down by law. Everyone has the right of access data, which has been collected concerning him or her and the right to have it rectified. The clients with this rule shall be subject to control by an independent authority. So, this, this is quite relevant I think to protect, not just the to protect the human. So that it's, it's, I think what we have seen from the example with the dissecting the racial biases is that we have multiple interests. And the problem is to receive the best health care that we potentially want but we also need our human rights to be protected and sometimes these are in conflict to each other and we have to somehow find ways to do that and the, the, the, we also convert economical interests in play to organize the health of the, the health of a population, for example. So, but so to go back to understand why these, these kind of frameworks are very important, despite, you know, lots of scientists sometimes hating them. They go back where they come from. And that's what I want to do here to some extent. So I think these ethical guidelines have been put forward mainly for practitioners for doctors for medical staff. At least for myself, coming from physics and medical and computer science and then moving into biology studying firstly, and plants and worms and model organisms and only lastly moving into working with human data. These were not so relevant for me until rather recently. And I think that as machine learning and genomics is becoming ever more powerful, it is really important that we understand where these guidelines where these ethical guidelines are coming from frameworks So I think, and also, I think it's also important because increasingly algorithms will support doctors in making their decisions. And while these guidelines have been developed for for doctors it is now also technicians and engineers and computer scientists were programming algorithms who have a huge impact on the health of individuals and the health of populations. So I think we need to understand where these, where these guidelines are coming from. So, historically, there's of course the Belmont report from 1979, which is from the US Department of Health Education and welfare and it puts down ethical principles and guidelines for research involving human subjects. So, even if you're only using bits and bytes, we're still using working with human data. And so I think this is something quite useful to look at. There's the declaration of Helsinki, which also puts down ethical principles for medical research involving human subjects including research on identifiable human material and data. And all of them really go back to the Nuremberg code from 1947. So this was a consequence of the doctors trials after the Nazi regime in Germany. And they, for the first time, set out 10 points on permissible medical experiments. And I do want to go a little bit, and I know that this is potentially a bit press taking and not what you have been expecting for today. But here I want to also go a bit into the history, what went wrong in the history of science as why did we have to establish these ethical guidelines. And that brings me back and that's very much motivated again by this paper by Angela, where she says that when science is viewed in isolation from the past and politics it's easier for those with bad intentions to revive dangerous and discredited ideas. Okay, so historical context, why do we need ethical guidelines, why did the Nuremberg code becoming, how did they become so important. And I think it has its origin in a really very, at the time, modern project, which we know now, now of course that it was a terrible mistake and it was eugenics. And at the time it was considered a very modern project politically progressive. It was inspired by Charles Darwin theory of natural selection. And it was studying exactly heritable traits, which were spreading through the human population. It was in fact the science of human heredity of the time. And the goal was to improve human population through selective breeding. So at the time it was a very considered a very positive beneficial project to really improve the human population as a whole and it was at the forefront of scientific discovery. And it was mainly driven by, was largely driven by two people in London at UCL, Sir Francis Galton, who was a cousin of Charles Darwin's and, and he was a biometrician, a statistician, a psychologist. And he created the statistical concept of correlation, and he attempted to study how genius and greatness is inherited from fathers to son, and one of his protégés was Carl Pearson, who established really established the discipline of statistics. And he was the chair of eugenics at UCL. So we had at the, at that time, we had two very powerful scientific new emerging fields via genetics and statistics, and they were coming together and they were proving to be very powerful but in some sense, in a much, much darker and, and, and, more than, than they had anticipated, I suppose. So they targeted a couple of traits for elimination. So the traits that they targeted were all very complex, they were subjectively defined characteristics, not very well understood at the time. They included bipolar disorders, epilepsy, alcoholism, criminality, and also something like feeble-mindedness, which is in general a moral defect, which they argued would lead to crime, laziness, financial burden on the normal. The causes of these traits were seen as purely genetic, and they completely dismissed environmental factors, including poor housing, nutrition, discrimination, and so on. So eugenics was widely endorsed in the literature at the time. It was described by the Science magazine as a work of solid merit, for example. So what I want to, what I want to understand here is that it's really not a fringe movement, but it was at the core of the establishment of science. So science in itself is not necessarily an argument for being right. It spread from the UK to the United States. There was a eugenics record office set up at Cold Spring Harbor Laboratories. Charles Davenport was the director at the time. He published the Charles Davenport Greed. I encourage you to read that. It's quite scary. And they went on to find a classification of feeble-minded persons based on beneath Simon intelligence scale, the IQ scale, and then decided which of the individuals were unfit for society, which had to be institutionalized or sterilized or both. And they also introduced a testing program on Ellis Island to identify the feeble-minded who would immigrate to the United States. So in many cases it was also connected with race sciences to be honest. So what it led to in the eugenics in the United States led to the state and forced sterilization. And in total over 60,000 individuals were legally sterilized in the US through the early 1970s and even early 1980s. So quite late that it was actually abolished. And you probably all know that there was a next level of eugenics in Nazi Germany where in 1920 already with theoretical work, these two men made a legal case for the permission to destroy life on mercy of life. That then led to huge, this was indeed put into practice during Nazi Germany. So there was a law passed in 1933 for the prevention of progeny with hereditary defects, which introduced compulsory sterilization, and even more tragically in 1939 when the war broke out, there was also a memorandum on the destruction of life, on mercy of life. And immediately patients in psychiatric hospitals were targeted for death. 70,000 patients in total under this action T4 were initially targeted. And they did so it wasn't completed, but one of every thousand people in Germany would have been affected. So there was a huge killing of patients in particular in psychiatric hospitals going on. And this really led to the number called the Declaration of Helsinki and the Bournemond Report, and then also this is what the European chart of human rights is is referring to when you when you look at the history of that. The genetics program largely shut down for sterilization in the US did not end until the 1980s so that's not such a long, long time ago. But this is why these ethical guidelines, which might sometimes which we might be sometimes tired about or which we think sometimes is a board burden to our research are so important and they remain important today. And the topic is also relevant in the 2020s. So, I want to say very firmly that eugenics is wrong. But some people would argue eugenics is morally wrong, but it might still work. Here's a very famous example which a dog hands has said in a treat. It's one thing to deploy eugenics on idea, ideological, political moral grounds. It's quite another to conclude that it wouldn't work in practice. Of course it would. It works for cows, horses, pigs, dogs and roses. Why on earth wouldn't it work for humans. There's no ideology. So, so is it, is it just, is it just morally bad, and would it still work is that actually true. And I think again, if we are looking about genetics and and how genetics control disease and also epigenetics. This is a very important question. And that would actually work to some extent. Are there assumptions that are made here actually true other do they hold what what what assumptions does eugenics actually make can be actually to prove them scientifically not just morally. To some extent, it is actually a. I think preparing these slides is actually a legitimate question to ask whether eugenics has worked. But but I think it is important because people are raising these questions again. So, there's, there's an interesting paper that studies schizophrenia and schizophrenia was one of the diseases that was particularly attempted by by the Nazis to be eradicated through the psychiatric genocide. So it is estimated that 200 and between 220,000 and 269,000 individuals schizophrenia were either sterilized or killed by the Nazis. So that was between 73% and 100% of individuals with schizophrenia living in Germany between 1939 and 1945. So a very, very substantial amount of these patients are diagnosed people with schizophrenia were indeed affected. It turns out that in the 1970s the incidence for schizophrenia in Germany was actually higher than in the international comparison. So in this case of schizophrenia eugenics did not work. So why is that what are the biological assumptions that are underlying eugenics. So the first one that assumption is that diseases such as schizophrenia we believe to be simple Mendelian inherited diseases which are passed down from generation to generation. And we now know however that schizophrenia is a very complex disease, and that there's a large number of variants, which all have a low penetrance and all of these variants variants contribute. So what does penetrance mean, it means that there's a large proportion that the proportion of individuals with a particular genetic variant that have the disease. I'm sorry. There's a large number. So if you have a low penetrance, that means that there's a large number of individuals which have this genetic variant, but are not affected by the disease. Okay, so you have even in the healthy population you have, you find the same genetic variants. And that's that's described by the term penetrance. So only two to 7% of the individuals that are carrying high risk copy number variations actually have the disease. So it's very difficult. So there's a complex interplay of a large number of variants, which all contribute, and it's not inherited in a simple Mendelian way. If individual variants yield very small effects, perhaps linear combinations could explain the effects or maybe more complex combinations could explain the effects. And what has become very fashionable is to compute polygenic risk scores and to some extent these have been very successful and very, very. But they also are sometimes massively oversold. So their predictive power is still very low. And that is shown, for example, in this example, which is, which is explaining an outbreak of polygenic scores for coronary artery disease, slightly different disease. What it shows is that the polygenic risk scores for the healthy versus the disease population really overlap massively, and a cut off the lead to missing out to a high number of false positives and a high number of false negatives. So it's still not a good tool for predicting disease based on even a large number of genetic variants. In many cases. Also, they have completely dismissed. So the assumption was that schizophrenia was purely genetic, purely genetic disease. Mental factors were dismissed completely so poor housing nutrition discrimination and so on all can contribute to discrimination. And again, we now know that that epigenetic factors can be mediators between environmental factors and the and the disease onset. So even high risk variants carried by many most never develops schizophrenia. There's an important interplay between predisposition predisposing genes and environmental exposure. So, and the epigenetic regulation of the genome may mediate dynamic gene environment interactions. None of that has been taken into account in that context. A very important example is also the Dutch hunger winter in 1944 1945, which led to a markedly increase in schizophrenia in Holland. Now, again, causality is very difficult to establish, but it's certainly a purely genetic approach is doomed to failure. And, of course, it's also the third reason the third assumption that underlay eugenics was that trades that it, such as epilepsy but also intelligence levels manic depression people mindedness and criminality are subjectively defined or subjectively defined. They are really difficult to measure. And this problem with epigenetics has been identified very early on so Thomas and Morgan in the 1930s already said, the main difficulty is one of definition accurate work on heredity can only be obtained when the diagnosis of the elements the trades is known. And a lot of psychiatry diseases in particular it's really the still its spectrums of diseases with very much overlapping features etc. I want to repeat here that there's no, no conditioning no moral conditioning it's just eugenics is wrong it's scientifically wrong it's morally wrong. But it serves as a warning how science in the best of the in the best of interest, even if you assume that the, the purpose of this scientific exercise was to improve the human population led into catastrophe. And I think we have to keep that in mind as we are moving along studying researching and trying to find cures for diseases or, or interventions. So, if you're going back to the epigenomics. These are a lot of questions that still need to be answered. And we need to have not just a dialogue. So the interdisciplinary research doesn't have to be limited to biologists and or biologists or life scientists, medical practitioners practitioners and data analysts or machine learning and practitioners, but it has much bigger consequences for both the individual and the society and we have to discuss those in a much bigger context. The reason for that is of course because it somehow defines who am I in the world what are my connections. How did I did I get to the place that I that I am in and so on so we are the subject, we are our own research subject, and in this context we cannot assume that we are objective. The take home messages from this so I will leave a lot of time here for for discussion, and so on. The core messages which I cannot stretch a stress. Strong enough is that we have to be really aware of our biases. We have to understand where we are coming from and we have to try to understand what are value judgments and what what are really facts. It also shows that diversity in research and teaching is not just nice to have, but it does promote better science science has gone wrong in the past. And, and I think we can by hearing many voices. We can improve on that. I'm very, very careful about assumptions on which we build our own research questions on and you have to make sure that they hold. I do think that we cannot just know our place. So a lot of the past mistakes have been propagated through the leaders through the sort leaders in the fields. We have to question not just our own biases but we have to questions question our seniors we do stand on on on massive achievements on the generations that come before us but we also have to question what came before us. I do think that as researchers and in particular in this field of machine learning applied to biomedical data, we have an extra responsibility to think about the impact, and the ends of our work in both in the positive sense and the negative sense we have to think about worst case scenarios as well in order to to prevent them. So I think that we have a very big responsibility to make our science understandable, and to explain consequences to the public. And here are some of the references that I was quoting in, in, in the lecture, and with that, I, I would stop. Thank you. Thank you, Gabriella. This is really super interesting lecture. So we, we are ready to take questions from the audience. If. Yes, I'm going to hand out the mic. And thank you very much for this lecture it was really very interesting and educational. So, I just, well, it's not really a question in the sense that what is a question but will just complicated I know, but I've seen some studies sometimes on epigenomics trying to, for example, find epigenetic. Yeah epigenetic changes in DNA for, for, for example, predicting if someone would be LGBTQ for example, or also trying to see if a certain type of race is has a specific type of gene or epigenetic thing also that would make them more intelligent. So I know that maybe the intention of the scientific of the scientific that do this is not wrong, but I also think that it may be dangerous in the sense that there are that are running now that discrimination against the race and sexuality and gender still exist. So I just wanted to ask, what do you think about this kind of studies. I think this is very, very good questions. I think that science. So, there's a professor for Dr. Rutherford he's a geneticist, and he's also moved into broadcasting a little bit and that's very good programs on the BBC and he once said, science is no ally, sorry, to race science or racism, for example, and I can only, I can only agree with that. I feel that we can use science to actually debunk quite a lot of myths that are out there, and really try to understand it and and explain it. So, so what was quite interesting for me is so this is an area that I have not studied at all but I. So, we have looked at. So, so the first is the first question would be, if there are epigenomic markers that predict, for example, lgbqt plus etc. I do think those markers potentially exist, and that they are actually quite relevant, and but relevant in the sense that we can explain that these are real, real, real issues that they are not made up. So what we see, for example, and that's very far away from that in practice but many years ago we've been studying. Medulation changes in the mouse brain postnatally and it was really and we changed the modulation there. And what was really fascinating was that a lot of genes were affected that are in the literature to be described to be sex determining genes and that would be expressed, actually rather in sperm or in ovaries and so on. So that postnatally a lot of these genes are actually expressed in a specific way in the brain and I think that we have a gender representation, a gender representation, an identity that establishes in the brain and there's a lot of things can go wrong in biology always. And it doesn't have to, and sometimes this could be in, so in a condition it could be that what is that there's a mismatch to what is expressed in the brain and for whatever reasons and these could be contributed by epigenomic factors. And I very much understand that this would cause huge problems for the people that are affected, and that is invisible. And it is not a matter it's certainly not a matter of choice, you cannot choose to be someone, and you can't be reeducated to be someone that you are not. And if we understand the basis of that that could actually help to promote understanding in the society, I think. So I think there's a lot of, I think epigenomics there plays a huge importance I think also what we are seeing is that drama, and a particular early childhood trauma is is kind of recorded with epigenomic patterns also largely in the brain. And if we understand that and, and this can be caused by discrimination by all kinds of experiences. And if we understand that in a better way I think we could also help to overcome these traumas. And to have better ways in, in, in reversing these effects. So yeah, so I think that understanding biological causes is a good thing explaining explaining biological causes is a good thing. I think that it's very important how you are framing question your research question. I think it is also quite important to understand, for example, why it might be a decent research question to understand why certain children are struggling to learn and others are having it having an easier time. Combining that with a different question which is differences in races or differences in gender or sex is I think putting on mind model on to a biological question. There's no reason to believe in the first place that because your skin pigmentation is different. That should have an effect on your cognitive abilities. These are completely if you're thinking about it. They're completely independent of each other. So if you're, if you're trying to find associations between intelligence and skin pigmentation, then I think you have almost an agenda. It makes sense to study how skin pigmentation is inherited and it's, it's very complex and fascinated. It's also interesting to understand how intelligence and cognitive abilities are working. Putting these two together has, for me, no by no scientific merit and, and potentially comes with a value judgment. I hope this answers your question. Yeah, thank you for your insight. So thank you. Yes, I would like to add that really this is a very powerful message that is important for us all as scientists, but especially for you are early researchers, early stage researchers. We have the responsibility in avoiding that science is weaponized. And to perpetuate inequalities. It should be the other way around. Okay. So let's see if there are other questions in the audience, please. In the previous lectures we talked about, like red syndrome, and there are illnesses and syndromes that are quite clearly something that we know we want to search a cure for, but there are other disorders. We are especially dealing with neurodivergencies, but with a lot of illnesses that we don't really know what where to go to the line where to call something a disorder and illness, or just something that creates challenges, but that might not be you only to the disorder to the way society works. So how can we know when to stop when to stop searching for a cure for something because it actually doesn't need one. I think that is a very, very difficult question to ask, which I find. I think that's a very, very difficult question to ask it's also a very I have to say. I mean, there are very different perspectives on it. I know one of the families who have a boy with MSP to over expressions and drone who sadly died just before COVID in 2019 and I followed their story a little bit and how they, you know, how they were, you know, were basically living in the hospital with that boy and they were desperate for a cure. And to some extent, there were some scientists who were feeding into that, into that hope, and telling them, Okay, we need one more. We need another 100, another million dollars before we can start with clinical trials we have something that works in mice maybe we can do that for find a cure in this case. And so this mother in addition to caring for her child would also do a lot of she was very much involved in fundraising and charity and charities for that disease and so on. And I think it gave her it gave her hope and not just for her boy but for for the life of her boy was not in vain and that was a very personal story. I heard some of the scientists and doctors say, maybe, you know, this these children we are doing so much at the beginning of their lives to save them. And their life is really a life of suffering. And where do we, where is the care, where is it more dignified to to stop and let someone die rather than moving on. And it's, it's, these are questions that I wouldn't dare to answer I think they also it, but it's it's questions that we have to confront in some some sense I suppose. In other senses I completely agree I think we have to be more. Instead of, we have to create also more opportunities for people that have some kind of challenging health conditions to not to eradicate the trade or the disease but to make it easier for them to live in the society and to remove the hurdles. Very difficult questions. Thank you. I guess, to some extent, what I also find very interesting is that, for example, when you are looking at life expectancy for that for example in the UK. It is, it is quite shocking that life expectancy has started you. So, at the other end of. So not the beginning of life and being born with challenging conditions but more towards what's the life expectancy of the population and it turns out that the life expectancy is, is the rise of life expectancy has is starting to decrease and it's starting and that happened before this trend started before covered actually and it turns out that again we can separate different socio economic parts of the population and the impact it creates in life expectancy happens much stronger in in the more vulnerable groups of our society. And I think there are some, some, some interesting questions here so we are trying to improve we have improved on cancer we have improved on all kinds of diseases we have it was a huge success in the last century and scientific breakthrough, but it hasn't translated really to benefit everybody. And a lot of other things like public health invention interventions for housing education food, which is, which is, which would give equal opportunities might be actually more efficient than personalized and precision medicine. So, thank you for the talk. It was very good and intense. I have a question two questions actually the first one is, so if something could be done, it's not necessary to be to do it right. So I was wondering if in the field of epigenetics there are examples of such things that like that. The second question is, well, how a lot did the scientists actually learn from what happened in the past, at least in epigenetics field, or it's very, or it's how, what's the alarming like, is it alarming this times for us regarding field of epigenomics at least. What do you think. So I don't want to put eugenics and epigenomics I think I've mentioned them in the same lecture, but I think it does go beyond. It's not, not specific to epigenomics I do think that we see very specific challenges at the moment because we have again. Similar to the, so statistics and genetics came together and, and they were very powerful, but they went into the wrong direction, or into a very traumatic, traumatic direction. We are at a similar time where we have incredible powerful genetics and sequencing technologies and we have at the, at the same time, these huge improvement in machine learning and how to analyze these data sets. We do see that some people such as for example that you can predict the intelligence from from your germline and, and that you can make, you know, your whole life fate is predictable. And I find that very difficult to swallow. I think, I think it's not just science I mean it's all, all us. I think we, we, we, I think in many, in many cases, we are telling people to trust science. We did that during covert and we did that for climate change and I think that is right. It's important that we trust science but at the same time, we have to be worthy of that. I mean, as scientists, we have to be super self critical and we have to be understanding of the risks that we are that's that science comes with. And sometimes it's like, yeah, yeah, yeah, you know, producing a car and sending out a car but it has only accelerators and no break system. And I think when we are building these cars we have to think about the break system and that's that's what I'm trying to do here as well. So, I think, but it's not just for for scientists right it's not. I think there were politicians who were very much involved in eugenics there were kind of it was in the community levels there were doctors. There were medical stuff. It was a lot of people bought into eugenics at the time, and I think the response of it was the number code it was the Helsinki Helsinki Helsinki declaration it was there. And as a society we have learned from it and it's important that we do not unlearn these lectures it's important that we understand that as scientists that mechanisms that slow us down like GDPR, which is based on the European chart of human rights, that these have an important function, and that we are not trying to remove the safeguards again. I don't see any other questions here in the audience and the chat. Either. So I think that with that, we can conclude thank you, thank you really so much Gabriela for the beautiful series of lectures and very welcome. They truly hope to have you here in person sometime soon. I would enjoy that. Take care and enjoy the rest of your summer school. Okay, thank you. Bye bye. Thanks for the organization as well. Sure, of course. Bye.