 Good afternoon everyone, welcome to our number six, the provost lecture series. And as you can see, so the provost lecture series launched in October 2022. And we have gone through five faculty members who either recently retired or recently got promoted. And it's with my great pleasure to open up today's provost lecture series, will be presented by Professor Simone Pigalotti, who's recently promoted to full professor. And so again, I want to thank many people shown on the slide, and in particular also to people from the provost office. And so we have been using this as our provost office almost monthly event. And so everybody is very happy to see what the OIST professors are doing and reconfirm our commitment to support excellent research within OIST. And also many other divisions shown here, they have been very supportive either in the budget setting or from the CPR communication and posting events and also recording every lecture to be potentially posted online. So I'm just going to keep my introduction very short. And so for today Simone's lecture, Professor Izumi Fukunaga will be introducing Simone. And in addition, Izumi will be giving her own provost lecture series in April. So this is also a means to promote Izumi's provost lecture series, which will be introduced by Simone. So that's perfect. So Izumi, now the floor is yours. Well, it's my great pleasure to introduce my great friend and colleague, Professor Simone Pigolotti, on this very special occasion of him being promoted to full professor at OIST. So Simone grew up in Rome, where as a child he already had a great propensity to excel. And later he studied physics in particular statistical mechanics. And after an extensive international training, he joined OIST in 2017 to lead the biological complexity unit. And over the years, many of us have had the great pleasure to get to know all of the family members who also themselves have become really integral parts of our community. So from the start, Simone took active roles to serve OIST. For example, as a member of the faculty council, representing the colleagues with his fresh and innovative ideas. And also by serving on recruitment committees, which takes a lot of time, that brought excellent new colleagues to OIST. Simone also contributed really enormously to the intellectual environment of OIST. And these are just two examples of events that he organized. And looking at these, I really appreciate that he puts in time and energy to bring world-class science to our campus. And Simone is also someone who has a knack for turning opportunities into success. And yeah, just a few years after this 2018 meeting where Professor Luca Pellitti was one of the invited speakers, we saw a publication of this really beautiful textbook titled Stochastic Thermodynamics. Luca on this occasion has sent me a message to share with you, which explains the background to this project. He says, I found a kindred mind in Simone, both in his interest in stochastic thermodynamics and in the attempt to make use of statistical physics to understand evolutionary processes. So I took advantage of his invitation to OIST a few years back to suggest working together on a comprehensive introduction to stochastic thermodynamics. The collaboration, although carried out mostly remotely, was as challenging as it was pleasant. So I am very happy that his talent has been recognized by this well-deserved promotion. And in the book, especially in the introductory chapter, Simone and Luca explain that stochastic thermodynamics is a study of non-equilibrium or non-static properties of physical systems that span from microscopic to macroscopic scales. So to me, Simone's research creates new ways of thinking about the world that we live in, which certainly is not static. And as a result of this unique ability to analyze the complex world, Simone has been in high demand by colleagues both globally and also here on campus, as you can see from these diverse collaborations that studied everything from how genetic materials replicate within bacteria or how microorganisms multiply when there is a spatial constraint and to grander things like how things spread in the ocean. Now, what makes Simone so successful in collaborating with others? Well, Professor Amy Shen has shared with me that Simone's ability to abstract many world problems has enabled him to quickly find biophysical models to explain experimental observations. And Amy also noted that his open-minded approach and frank discussions have been very key ingredients in the success of their collaboration. And I'm sure we will get to hear more about the secrets of his success and his fascinating research from Simone himself. And I'd like to end my introduction with a haiku I wrote on this occasion, which is also etched on the gift plaque that we presented later. And it goes like this. The language of chance put to our dynamic world surprise everywhere. I look forward to your talk. Thank you. So first of all, thanks Izumi for the very nice introduction and thanks to Amy and the Provost Office for organizing this lecture series. So I have to confess that I'm not really good at preparing and presenting these kind of seminars. So in the last few weeks, after I was invited, I've been thinking a lot about what should I tell you about today. So one obvious choice would have been to just list all the research achievements in these last few years at OIST, or maybe I should tell you about the story of my life, or maybe I can make some story about anecdotes of how, in what moment, I thought about the idea that helped me solve some particular scientific problem. And then I started thinking more about what do you might be interested in hearing from me today. And I saw that I realized that even people that know me relatively well at OIST sometimes when they feel particularly braved end up asking, OK, but what is it that you actually do? So this is, I think, a bit of an overarching question. So I hope I'll talk about most of these things, but I hope that by the end of my lecture you'll get a better sense of what we do in our unit. So I'd like to start my story in 2017. This is the year when we arrived at OIST. OIST at the time looked pretty much like this. There was no lab four, no lab five. It was a little bit smaller. And I was attracted by this place because of the research that we do that bridges between physics and biology. And so I was really into the interdisciplinary aspiration of OIST. And that has worked very well for us. So we established nice collaboration and fruitful collaboration with a number of people from different disciplines from biochemistry to ecology to physics and science. So this has worked very well. So if I fast forward to winter 2021, 2022, things were going pretty well. There was still we were in the middle of COVID by then. So you can see them wearing a mask outdoor. But overall, our family was happy. Science was working great. So things were pretty nice, except that at that specific time, I started receiving the first emails from the faculty affairs office saying that maybe it was started to be about time that I should think about preparing my tenure package. So that was a bit of a concern. And in particular, I started thinking, how should I package together all these projects that we have done where we have mostly followed our scientific curiosity into some kind of coherent story. So ended up with a scheme. I thought that thinking about three areas is good. Three is, they say it's the perfect number. But this really represents pretty much everything that we do. So we work in non-equilibrium thermodynamics. We are interested in a small scale biology and in particular how cells process information. And at the large scale, we look into population dynamics. So what I ended up doing for the review is just look at all the projects we did in doing these years and just make them fit into this scheme as a classification to just give them some structure. So I don't want to do this today because I don't want to bore you as I did with my reviewers. And I will instead try to talk more about why these themes make sense together and what are the commonalities within these areas. And use a few individual projects as key examples. OK, so I'll start from physics just for convenience. And I apologize that this is very elementary for some of you, but it's a general lecture. So I want all of us to be on the same page. So in a nutshell, the idea of physics is intrinsically reductionist science. So we believe that we should be able, at least in principle, to understand the properties of matter from its constitutive elements. And this is precisely the goal of statistical physics. So the idea if we have, let's say, water, we should be able to understand properties of water from the properties of, say, the water molecule. So making this jump is essentially the problem of statistical physics. And I will not review everything in statistical physics, but I just want to give you a few messages. One is that we became pretty good at doing this when the system that we are studying is a thermodynamic equilibrium. So the idea of a system of thermodynamic equilibrium is a system whose properties do not change in time and is not caused by any flow of matter, current, heat, and so on. This problem was essentially cracked at the end of the 19th century by the work of Maxwell, Boltzmann, and a number of other people, Gibbs, for example. For example, one key result in equilibrium statistical physics is the so-called Maxwell-Boltzmann distribution that dictates that if I look at all the possible configurations of a large system, the probability of finding the system in one of these configurations is given by this expression. So E of x will be the energy associated with that specific configuration. K is a constant, the Boltzmann constant, and T is the temperature. So again, if you don't like formulas, basically the idea is that if you have a system at low temperature, I would find the system with high probability in states of low energy and with low probability in states of high energy. And if temperature is high, this probability distribution is much more homogeneous. So you can think of equilibrium statistical physics as this battle between energy differences and temperature. Energy differences tend to make the system more organized and more structured, and temperature is essentially acts as a noise that makes everything much more random and uncertain. The other thing I want to stress here is the idea of universality, which is a key idea in statistical physics. So it is pretty remarkable that this result is valid for a very large class of systems. I don't need to know what my system is made of. I don't need to know whether it's a liquid or a gas or a solid. This result will be always valid. On the very weak assumptions. OK, so I told you all of this because what we are actually interested in is non-equilibrium statistical physics. So if I take an equilibrium systems, there are many ways of bringing it out of equilibrium. One possibility would be, for example, to warm up one side of the system, for example, the bottom here, and cool down the other side. So if we do that, we can imagine that there will be a flow of heat traveling through the system from the bottom to the top. So this will manifestly violate the assumptions of thermodynamic equilibrium that I mentioned before. And this problem is way more complicated. So the dynamics that would be caused by this imbalance would be heavily system dependent. Heat flow is very, very different in solids or in liquids or in different kinds of liquids. And so there are no obvious universal principles. So the state of the art here is we know pretty much everything about equilibrium statistical physics. We know next to nothing about non-equilibrium systems. So what do we do? So one possibility to study, to build up a theory on non-equilibrium statistical physics is to start from small systems. So here, one example would be like a relatively small particle in a fluid. So let's say this particle would be less than a micron. So it would be large, respect to the molecules of the fluid, but still relatively small. So the advantage of studying this kind of system, we can also think about the non-equilibrium version of this problem would be, like for example, dragging this particle with a constant force. And this is the kind of logic that Einstein followed in his famous paper on Brownian motion. So the advantage of studying this kind of system is that they are small, they have a few degrees of freedom and so they are more tractable than large systems. The disadvantage is that the dynamics of a small system is stochastic, because this particle is not very large compared to the molecules of the surrounding fluid. Each time that there is a collision, this particle will receive a kick in a random direction and so its trajectory will be characterized by certain degree of randomness. So essentially, and thanks Izumi for introducing this, the field of stochastic thermodynamics has been a developing of this approach to non-equilibrium system. So the idea has been to formulate a consistent, complete and coherent thermodynamics for systems that are small. So the general framework is that, again, you have a small system. This system is embedded in an environment that is fixed at a certain fixed temperature. And this system is brought out of equilibrium by, for example, an external agent, can be an experimentalist that could perform work on this system or try to extract work from it. And the examples of this can be colloidal particles, as we saw before, or, for example, molecular motors. So these are molecules that consume chemical energy to perform work inside cells. So there are lots of, also, applications. And this is the book that we recently wrote in which we try to provide a pedagogical introduction to this field. But so I want to just give you a sense of the kind of results that people have been studying in stochastic thermodynamics and why this is interesting. So if you take one of these small systems, there is a way of defining an entropy. So the entropy that the system produces during its dynamics. And since the trajectory of the system is random to some extent, so will be the trajectory of the entropy production as a function of time. So on average, entropy will start from zero and go up because of the second low-othermodynamics. But because of this fluctuation, you can have that the fluctuation will bring entropy production below zero, therefore creating these apparent or transient violations of the second low-othermodynamics. So this is a phenomenon. This has been called transient violations of the second low-othermodynamics. The name is slightly misleading because the second low-othermodynamics is formulated for large systems for which there are no fluctuations. So these phenomenon are really absent in classical thermodynamics. But this is what happens. So clearly there has been a lot of interest in characterizing how these fluctuations can create these effects. So in general, the trajectory of entropy production is, as we say, this system dependent. Each system will have its own dynamics. But there are some properties that are universal. For example, if you look at the distribution of the minimum point of the entropy production along the trajectory, we found that this has a universal distribution. In fact, in particular, its average is minus 1 kB, so 1 in units of the Boltzmann constant. The maximum before the minimum also has a universal distribution. The number of times entropy production crosses this zero line also has a universal distribution and so on. So here in these curves I'm showing you lines are mathematical predictions from our theory and points are simulations from three different kinds of models of small systems. So there are some properties that are universal even out of equilibrium if you look for them. OK, so this is a bit my introduction about physics, but really what I wanted to tell you about today is about the relation between non-equilibrium physics and biological systems. So this is a long-standing question and I believe that one of the problems in biological physics is of historical nature. So physics traditionally has been built upon describing inanimate matter and it took us a lot of time before really being brave enough to think about what to do for biological systems. And it shouldn't be a surprise that the first things, the first systems that biophysics studied were things like proteins and membranes. For example, this is really the fabric of living systems. But you might think that if I want to study a protein, it doesn't really matter that this is part of a living system. This is ultimately a large molecule and I can study it with the same tools that we can use to study inanimate molecules, at least from a physics point of view. And so I will argue that there is a much larger jump in going from this type of biophysics to a biophysics of things that are really alive, they can grow, they can replicate, they can sense their environment and take decision about the state of the environment and so on. So this is thinking about physics of the system is a much more challenging task. I will also argue that this is a task that is worth thinking about. I'm not the first arguing for that. So for example, one very famous example is this very nice book by Erwin Schrodinger that already in the 40s started thinking about physics of life. This is not the original edition, but you really can't be discovered. And the line of reasoning of Schrodinger is really heavily based on non-equilibrium statistical physics. In fact, perhaps the most famous quote from this book is this life feeds on negative entropy, which is a quite profound concept, I would say is expressed in a slightly obscure way. So I think what Schrodinger wanted to say here is that in order to function cells need to create some kind of order structure inside them. The natural tendency because of temperature would be to have disordered inside cell. And so there should be some non-equilibrium mechanism that dumps the entropy corresponding to this disorder and throws it into the environment. So this is essentially what Schrodinger wanted to say here. Because of the room we are in today, I want to tell you of another quote. That I discovered only recently, in fact, just a few weeks ago. So this is a nice opinion paper in nature ten years ago called the Lies Code Script. I believe 2012 was 100 years since the birth of Alan Turing. And nature made these special opinion pieces about contributions of Turing to various fields of science. And Sidney Brenner wrote one about contributions of Turing to biology. And this is what he wrote. This is an extract from this article. So he starts by saying biological research is in a crisis. So he really didn't sugarcoat it. And this is really the first paragraph of his article. He just, he didn't mention Turing at all. He just said biological research is in a crisis. We are drowning in a CO data and thirsting from some theoretical framework with which to understand it. Although many believe that more is better, history tells us that the least is best. We need theory and a firm grasp on the nature of the objects we study to predict the rest. It's very nice. Then at some point he talks about Turing, but not in the very beginning. OK. So I started thinking about these kind of problems before joining OIST. And in particular, I looked at the non-equilibrium thermodynamics of replication of information. So this is a basic task that all cells need to do. Information in cells is contained in hetero-polymers like DNA. And there are enzymes that are responsible of reliably copying this information. So from the very first principle point of view, ideally you would like these machines to copy this information with a very high accuracy because you want faithful replication of information. You don't want to spend too much energy to do so because energy is precious in cells. And you also like to do it at high speed because these molecules are very long. And if you take too much time, that might impact your fitness. If this process was carried out at thermodynamic equilibrium or close to it, you would have very low accuracy, again, because temperature in equilibrium systems tend to create noise and prevents any kind of order structure. You would have no dissipation, so this would be good. And also you would have vanishing speed because equilibrium systems are pretty much static. So from this kind of argument, you can think that these machines clearly operate out of equilibrium. And what they really do, they have to trade off the energy they spend to improve their accuracy and improve their speed. And so there are lots of interesting questions about the trade-offs involved in this business. So how much dissipation goes into improving accuracy versus improving speed? And what are the trade-offs between accuracy and speed themselves? The other question is, how do we study all of this experimentally? Because coming here at Thais, I was also interested not in doing abstract theory, but thinking how to measure this thing in a lab. And so one thing that we started thinking about is that we can exploit statistical properties of growing population. So we know, in general, from microbiological studies, if you give plenty of nutrients to a microbial population, after an initial lag time, the population will grow exponentially. And then there will be a stationary phase in which the resources are finished and the population will stop growing. And eventually, there might be a death phase. So in the exponential phase, the number of cells grows like exponential of time, with some exponential growth rate. And so one point that I want to mention here is that if you imagine to have diverse populations where individuals carry some different property that might be function of their age, then what you find, and this was fine, first in this very nice paper by Powell and then there's a long literature of variants of this. And we also contributed to this problem, is that the average of this rate over the growing population is the average of this distribution weighted by the exponential of minus the growth rate times the age. So this is pretty nice because this looks pretty similar to equilibrium statistical physics. So I can interpret age as energy and lambda as an inverse temperature. And so this looks really like the kind of averages we do over the Maxwell-Boltzmann distribution. And we are used to do since 200 years. So that's pretty nice because this is a manifestly non-equilibrium system. It couldn't be farther from equilibrium. But I can have expressions that transform it into something that looks like an equilibrium system. Again, this might sound a bit abstract, but let me tell you the application that we worked on. And this is a replication of DNA in bacteria. Again, let me give you some basics. So in most bacteria, most bacteria have a circular single chromosome. This will be their genome. And DNA is copied by two large protein complexes called replissums. They would bind at a specific site called the origin of replication, proceeding opposite direction. And so they will meet at the end. And at the point, you would have two copies of the DNA. So what caught our attention here is that this is something present in previous studies. If you sequence an exponentially growing population and look at the amount of DNA you find as a function of the genome position, you find this characteristic pattern with a peak at the origin of replication and a minimum at the final region where the replication finishes. And the reason is that in an exponentially growing population, there will be a lot of incomplete genomes that are undergoing synthesis. These genomes will always contain the origin, but they will have a decreasing probability of containing regions that are far from the origin of replication. So this is a problem where we really can apply directly our theory. So here we have our distribution. So this will be the genome coordinate. This will be the age of the genome since it starts being replicated. And so this will be essentially the probability that a genome of age tau will contain the position x. And so essentially, by averaging this function as I showed you before, we can predict this curve, which means we can infer the dynamics of these replissoms starting from experimental data. So I will not go into details. But again, we could apply this theory thanks to our collaborators from the Yokobayashi unit that performed experiments where we sequenced the growing populations of E. coli at different temperatures. So here the points are experimental data. The lines are our model prediction. And what we found is there is a variable speed of genome replication. So the average speed increases with temperature. This is something to be expected because temperature tends to speed up molecular reactions. But what was more surprising about this is that the velocity of these machines is not constant along the genome. But this plays regular and repeatable oscillations. So in this plot, again, in the center is the origin of replication. And these are the variation of the velocity of these machines. And you can see this wave-like pattern with variation on the order of 15%. And this is highly repeatable. These oscillations are only absent at the lowest temperature we studied. So this was quite interesting for us. And we tried to think, what does it mean? Is this a real thing? Is this an artifact of our model? And what we found is that previously people have measured the variations of the mutation rate of E. coli along with genome. And what they found is something that is very similar to what we studied. So again, here, these are velocity as a function of position. And this is a mutation rate. So if you look at how many accumulated mutations you have as a function of the position, you see a very similar pattern. So there is a very high degree of correlation between these two things. So this is quite exciting for us for a number of reasons. One is related with what I've told you at the beginning of this part about the trade-offs between speed and accuracy of this machine. So our interpretation here is that for some reason there are some areas of the genome where this machine progress at a faster speed. And as a consequence, because there should be a trade-off between speed and accuracy, when they're going faster, they make more mistakes. And so as a consequence, the observed mutation rate is also higher. So again, this is something that we plan to study more in the future. So I showed you this is an example where growth can help us saying something unexpected about a system at small scale inside cells. But of course, growth has a lot of consequences for larger systems, in particular for population of cells. So I will tell you a bit about this recent project of a student in our group, Angelica, that we also did in collaboration with the Shen Lab. So the experiment consists of looking at competition between two strains of E. coli. These two strains are tagged with a different fluorescent protein. They make them look one red and one green under the fluorescent microscope, but they are otherwise identical. So we inoculated these strains in a microchannel and we looked at their competition dynamics. So this is the movie I'm showing you now. So in the beginning, these two strains are well mixed. They can reproduce and reproduction make them expel from the channel. And what you see is that in a few generations, we observe this very regular pattern of alternating lane of green and red bacteria in the channel. And at some point, at even longer time, this pattern gets disrupted. So the idea of this experiment came from theory. So we actually studied a model of how these dynamics could go about. So in the model, individuals are placed for simplicity on a lattice. So we have lines of n cells and you have m lines. And each cell can become two different types of strains. They're here represented by different colors. And every time a cell reproduces, it will place a daughter cell in one of the neighboring positions and push an entire lane of cells leading to an expulsion of another cell from the channel. So if you simulate this model starting from this infinitely diverse initial state, you will see that after a few generations, the model predicts formation of stripes. In this case, only two stripes. And at a very long time, there will be competition between these stripes. And finally, one species will dominate. So we try to analyze it mathematically, these simulations. And essentially, the idea is that we could solve exactly the case where the channel is so narrow that there is a single lane. In this case, our mathematical theory tells us that the fixation process, the process by which one of the types dominates the channel, is extremely fast. So this is a very fast process. And so the idea was, well, probably this means that if you have many channels, then since this dynamics is fast, in a short time, each of the lanes will be dominated by one species. And then it will start to have competition among these lanes that we can also, at this point, describe because each lane counts as an individual unit. And this will lead to a final fixation inside the channel. So again, because we wanted to test this, so Amy and Paul in the Shen unit constructed these microfluidic devices with the channels of different widths, hosting from one to four lanes of bacteria. And by tracking the bacterial dynamics, we were able to test our predictions. And so essentially, this is the number of types that we observed in this first regime of fast dynamics. And this is the dynamics in the second regime. And in this plot, the lines are our mathematical, exact, approximated, but mathematical results. The triangles are numerical simulations. And the circles are the experiments. So not only we can qualitatively describe these competition dynamics, but we can be very quantitative about what happens and the dynamics in these systems. OK, so let me finish. I will conclude. So I hope I convince you that there are many interesting open problems in biological physics. And we have a hope to find interesting general principles and simplicity in biology if we look for these principles. And I'd like to conclude by acknowledging lots of people. I think here I would have needed the rolling credits, like in the movies. First of all, my unit, without them, this would not have been possible. So this is a relatively old photo. There should have been a surprise photo by Scotty, but I don't know what happened with that. So Deepak and Floriana have been working on the Reprisome project that I showed you before, and Anselika worked on this microchannel competition project. There are many collaborators. I've worked mostly with external collaborators for this stochastic thermodynamic work. Samuel and Yohei for the Yokobayashi unit, and Charles from the Lascombe unit for the Reprisome project, Amy and Paul for the experiments, bacteria and microchannels. And all my other collaborators at OIST, Buugagnan unit, Kuzumi unit, Laudino writer. And probably I forgot somebody I apologize in advance. And thanks for your attention. Thank you so much for this fascinating talk. I'd like to take questions from the audience, if you have any. G, yep. Thank you so much for the talk. I have a very specific question regarding the fluid from the Reprisome project that we can see. It's basically the waves that it derived is quite symmetrical. And you were also mentioning that there could be underlying biological, underlying principle for all the waves or all the dynamics of the fluidics in terms of population growth. And I also read the E-Life paper real quick to see the underlying logic of it. So do you know about sonification? So we use sonicators as biologists to really mix up and to basically to make our solutions or make some organic solvents evenly distributed. And it is very commonly used, for example, if we want to make liposomes. And it is a very difficult technology for separating liposomes because even when you were shaking and sonicating those fats and trying to make them into balls into this solution, you make big balls and you make small balls. And it's very difficult to make them evenly distributed. Do you have any insights for that kind of like, because when we talk about those kind of underlying fluidics and waves and then in terms of biological, and now think about actual technical, those waves and distributions like the sonicators, would you have any insights to offer for that kind of, for our biologists when we do experiments to make our solutions and make those organic and fatty solutions to make that very evenly distributed for the underlying biological waves? Thanks. I'm not sure I understood the entire question. So one thing I can say is, in the model, we assumed that the waves are symmetric. I'm not sure how much you can see on the figure, but the experimental curves look pretty symmetric as well. Maybe there are some points where there's a slight lack of symmetry, but we didn't want to introduce too many parameters. We believe that this is real. This is not due to the experimental method used. So we use sonication as well to break the DNA. Of course, it would be nice to try different things. But since the population is not synchronized, it would be unlikely that this depends on anything of the experimental protocol. But of course, there are lots of things that we could check. And maybe we can talk separately. No, no. I was just using this as an example, because basically this is growth. You were unzipping the double-stranded DNA and rapidly making them. But for sonication and also raves in the brain, sometimes it's like you do this kind of waves, but then for biological, for those fluids, it's like that kind of spatial distribution. And then the product you make is relatively, I was just wondering in terms of mathematics and then you kind of physics, if there's any insights in terms of the equations for the kind of sonications. I'm not sure, but maybe we can talk separately. OK, yes, sure. Thank you. I'll rest. Thank you. Thank you, Simone, for this interesting talk. I have a simple physics question. In the beginning, when you talked about negative entropy, this is the first time I heard about that, normally entropy is a measure of disorder. And bringing entropy down means making order. And I can't imagine something negative order. So am I misunderstanding the concept of entropy? Well, I mean, I didn't spend too much time formally defining everything, so it's justified. So essentially, what we call entropy has two parts. One is the change in system entropy. So there would be an entropy of the system in the beginning, and this would have a variation. And the second is essentially the heat released in the thermal bath divided the temperature. So it would be what one called the reservoir entropy in thermodynamics book or general environment entropy. So this would be the sum of the two. So it's the total entropy produced in the process. And it's the sum that in the second law details that the sum must be positive. But you can have fluctuations affecting both. So essentially, the idea is how can I imagine that you have a simple case is you have a particle, like a colloidal particle, and you are dragging it with a certain force, right? So if the force is large, that essentially, and we are in a viscous medium, all the work I'm doing is dissipated into the fluid. But if the force is not so large and the particle is small, it is possible that the particle will move in this direction instead of this direction. So if this happens, I am extracting work from the particle because the particle is pulling me, right? So if this happens, since the particle has no internal entropy, all the entropy is the entropy released in a thermal bath. Here, effectively, I'm sucking entropy from the thermal bath and transforming it into work. Of course, this can happen only because of fluctuations. On average, the particle will move in the direction in which I'm pulling it. But even for this simple system, the particle will only move to the right on average, right? Would that make sense? OK. That means on average, you don't have negative entropy. On average, you don't. If you average out, these things disappear. Thanks. Questions? Because if not, I have one question for Simone. So actually, it's relating to the speed accuracy trade-off that you showed in the replosome. Do you think that there is some advantage to having such a system? In other words, are these high mutating parts there for some reason? Is there some advantage for these genetic materials to have mutations? And very close to the origin, for example. That's a great question. It's a very good question. I am not sure if these variations are so large to be selected for. So I would imagine that you would want to place, say, for example, highly conserved regions. This doesn't. Oh, this one, sorry. Take the wrong one. Highly conserved genes close to the minimum of this. You don't want to have too many mutations. We are planning to look at the E. coli genome and see whether there's some correlation between position of important essential genes and this pattern. But we really didn't study that. That's a good idea. Anything else? OK, then we'll continue the tradition. We have a gift plaque that the provost of his kindly prepared. So it looks like this. And it has a haiku on it. So congratulations, Simone. And we have another surprise that his lab members have also prepared a little something for Simone. So would you like to come to the front, Simone's lab members? Every occasion needs cake. Thanks. I don't know what it is. No, you have nothing to say. Wow. Thanks. So this plaque is the result of a vote. To be fair, the vote was taken within only our group. So OK. Thank you. But still a vote, so. That's great. It's from show. Oh, yeah. Thank you. Should we do candles? Maybe we can go outside. No, we'll go outside. We'll do it. OK, thank you. No food in the auditorium. OK. Thank you very much. Right. Great. Well, this concludes this provost lecture. But there will be a refreshment outside. So please.