Episode 16½ - End-of-season 2 commentary

Transcript

Milosz: 0:17

Okay, let's do this. This is the end of season two of phase space invaders. But we just had eight great conversations with researchers working on exciting topics. We've had Michele Vendruscolo who works on addressing protein misfolding through drug design. We've had Ariane Nunes-Alves who tries to study the role of kinetics in drug design, as well as the effects of molecule crowding. We've had Justin Lemkul who develops polarizable force fields for nucleic acids. We've had Vlad Cojocaru Who investigates pioneering transcription factors involved in reorganization of the genome. Then we've done the conversation with Daniel Zuckerman whose main focus is on the fundamental statistical mechanics through the lens of trajectory, ensembles. We've had Syma Khalid. Who studies the bacterial envelope and the physics of permeation of drugs through the bacterial external defenses. We've talked to Paulo Souza who works on the development of general purpose coarse-grained force field Martini. And last but not least. We've had Janusz Bujnicki who talked about developing methods for the structural modeling of RNA from data and physics. This were, the scientific topics, but in every conversation, I also try to touch upon social or tangent topics that are relevant to what we do. Outside of our main research lines. But play into who we are as computational scientists. And so one topic that was kind of recurrent what's that off the responsibilities of a scientist. This idea, being that we have plenty of people who are intrinsically driven as research scientists really get to, you know, follow our own curiosity. But then as, for example, pointed out by Michele. We need to remember that we are appointed by the public on the mission, not to just selfishly follow our own personal goals, which we get to do. But also to unlock the understanding of the world for everyone out there. And hopefully to contribute to new therapies technologies and so on. So. We need to carefully approach and balance two kind of connected goals. One goal is to create research that will translate into useful knowledge for the society at some point. The other goal is to simultaneously maintain this curiosity that drives us, you know, we need to avoid burnout and to do this, we need to explore questions that might not be immediately useful, but are interesting. possibly conductive to huge breakthrough, somewhere down the line, maybe many, many years from now. But then in practice, it's really hard to justify playing with some absolutely cryptic idea that is. of no immediate interest to anyone, but as for say, 10 or 20 years, While being funded by public money. And I think this tension is close to the heart. Of many scientists who have kept their natural curiosity. Right. So how do we do that? We need to make sure that we set aside time to explore new things. We need to make sure that we're not just kind of mindlessly following the latest trends that are being pushed in within our community. For the sake of landing a citable paper or, you know A press release. But then we're also contributing to the creation of new horizons or bringing ideas from other fields. that's a creative process. That's cannot be really automated. On top of that. we need to think about the way we engage with the public, whether these are guests appearances in popular science outlets. Having our own YouTube channels or blogs, or maybe Tik TOK or Instagram pages. Eh, all of this will be needed for the future generations to be brought up in appreciation of science. Feeling that science is something worthwhile. So, you know, power to the influencer ones among us. If you have it in, you absolutely grabbed that smartphone and make that story. real, that tick-tock that content. If not, you can always be like me. I've always been terrified of talking to the public but. I dragged myself, kicking and screaming into what I'm doing now. And it's a bit less scary each time around. But you don't need to be born for that, that for sure. Then we might expand that to think about how we can also have an impact in the real world through policymaking. So bringing up the story from Janusz. Can we reach out to a political body to offer, or maybe we're asked to offer our knowledge or expertise, our, you know, our skills in interpreting the evidence. For people who want to draft new policies. That will certainly take getting out of one's comfort zone, but. It can also be a very impactful and rewarding contribution to the world of science and politics and a. If say our original goal was to change society for the better. as I imagine. It was for many the listeners. So engaging in such activities can fit into our, you know, our value systems. Right. I mean. It is very easy to feel that we have no time for all of these outreach activities and engagements and so on. It is true. We are all overloaded with work, but in the longterm, if we solely focus on the technical aspects, And purely scientific questions we risk exactly this feeling of burnout. and asking ourselves, okay. I studied the most technical details of the most specialized theory out there and done? Uh, administration documentation and so on, but in the end, what would the passionate teenager me think of that? The one who wanted to unlock. You know the mysteries of the world and maybe help people and maybe bring excitement about science and the public. That's my reminder. That we should ask ourselves these questions, at least once in a while. And, um, That kind of bleeds into the other topic that was explored by among others by, by Syma, by Michele. and by Ariane. Of our focus on therapies and drug design. So we've said many times we are now in this world where increasingly competition with people. We can start to think of, you know, not just discovering mechanisms and describing. What is happening on a molecular level. But also thinking of. How we can go in and modify something, how we can play or tinker with this mechanism. How we can use something that biology devised. to solve a problem that either biology or civilization, created. And so not only we have more. Validated molecular targets, but we also start to discover what causes diseases on the deep level. Right. We start to understand the Alzheimer's. We starting to have mechanistic. Understanding of autoimmune disease. There was recently a paper about the whole pathway involved in. Inflammatory bowel disease from a genetic variant, often enhancer all the way to gene products. Regulating the production of inflammatory pathways. There's going to be new pathways, new cues. About how ALS lateral sclerosis or Ms. Multiple sclerosis. Is brought about as well. It seems that almost every month, there is a new discovery of the molecular pathway that contributes to a major disease. So this opens up a lot of space for us, computational scientists thinking about how to address those diseases. Right. We are accumulating more subtle explanations of failure, modes of biology, but also failure modes of therapy. So we understand molecular assemblies, molecular barriers, metabolism better. As we talked with Syma, for example, we know what bacteria do to prevent the antibiotics from each bridging their targets. we get a better view. Uh, that was applying to in the conversation. with Paulo of heterogeneous. A lipid membranes or cellular compartments That affect drug uptake and delivery. Ariane talked about how we can think of molecule crowding, to avoid off target effects and rethink drug efficacy. And so also with the proliferation of data of databases. We can increasingly start to discover connections between different processes of thing that we require a different approach to teaching and self-teaching as well. Everyone needs to know that you can now go to Uniprot and find 30 different aspects of a protein's function. From mutational data to post-translational modifications, to structural complexes with other proteins, or, you know, That interaction graphs with these and a bit of imagination, you can go to alpha fault. Just start predicting. Complex geometries and see if that matches some co-evolutionary patterns or, Loss of function mutations. And, um, we can start churning out hypotheses for downstream experimental validation, which also means we need to learn modes of experimental validation. That's all. Complicated. That's all a lot of work, but, That's what will transform the field in the next years? I believe. And of course, there's still a shortage of data to train our predictive models on and that's a major issue. So we would like to solve many things with AI, with those automatic molecule design or network prediction systems, whatever. But the major problem that those systems face is it turns out lack of high-quality data. Whether it's for generative models or aggressor or so classifier. So you name it. So in medicine, chemistry, for example, the question of generating molecules that are chemically meaningful and stable is kind of solved. That's perfectly bad, but to a large extent, But to know what molecules do and how they behave in this complex environments, how they bind to, you know, off target sites. How they get metabolized or absorbed. And so on this question will still rely on new sources of data. So he would have this dual, let's say predicament of both having an enormous power that we have to learn to use and deal with. That we didn't have before. But we also need to work on producing better data or extracting more meaningfully formation from the data that we have, to help us get there. And that's going to take a lot of, again, imagination and trial and error. The next question that showed up a lot was at a fostering them. Technical and scientific skills within the community. That's something we talked about with, with Justin and Daniel and, uh, something that comes up a lot is this dichotomy between on the one hand. Simplicity that can kind of drive up usability. That was also brought up by Paulo by the way. But you can drive up usability through simplicity of the tool because people will tend to use your software more. If it's really, really easy to adopt. But then on the other hand, there are many decisions that have to be made that are tailored to a specific research question that we, you know, we see. advising on online forums. So this is the depth of the knowledge of the user. Knowing the theory behind the tool. And it's not really immediately obvious how to balance those two. Right. So the knowledge of the user and the usability or the simplicity of the tool. So in other words, how to make things easy to use, but not to opponent where average user is left blind to. Important choices or assumptions, or really cannot interpret the output of the model. And that's why we can't always solve. Someone's. Technical problem just by giving them this case by case advice. That's why. We need to think of foundational training. I believe so. That's where things like textbooks or lectures come in. And so as both, I think Justin and Daniel were pointing out these two modes of interaction need to reinforce each other. It's not really. Obvious how to communicate the complexities of our field. People who are just, you know, uh, undergrad students are first time users of certain software. But we need to care about conveying the key. Physical foundation. So. Whether it is the trajectory or free energy centric vision, you know, of. Pure statistical mechanics or it's Markova and dynamics, all those things. It would probably be beneficial to have a core curriculum for molecular biophysicists and I'm sure pieces of this exists somewhere, but maybe we should make a community-wide effort to emphasize what people need to learn. In order to start working in computational biology. Because it will be a lot of different angles, right? You can have a perfect understanding of the biology, cell biology, molecular biology, and so on. What is DNA? Well, this RNA. But not know how those things translate into models and physics and so on. That's one failure mode. And another one is that maybe you're a computer scientist and you know, that the models are there and the model is the great things, you know, like, okay. You're a harmonic approximation from and on harmonious or nonlinear or whatever. But you don't really know why terminal residues are special or things like protonation stays because these are some things that's, you know, chemists care about. So perhaps could we please have these. Central resource for the understanding of computational biophysics from a biology standpoint, from a chemistry standpoint, physics, maths programming tools. And so on, of course, the tools. You know, evolve every month now, but the foundations, the basics are going to be the same for decades to come. I believe. And so it's never going to be a single book or like a single lecture or a That someone will go through in that state. But I wonder if there is a standard set of resources out there that let's say, if someone goes through them within a year or two, they will be at least able to recognize most problems or issues within the field. I think it would be amazing to work, to compile. Such a curriculum and, um, you know, tell the young people, look, here are the basics we know you're lost. We know you're just trying to figure out what this thing is about. But, before you move into advanced, enhanced sampling with AI, or, you know, a new docking method, make sure you know, what the field has been up to in the last five decades about the foundations we have. You know, we don't need to reinvent the wheel. Also make sure that when you do things. You know, What's the goal off of what you're doing. Right? So as Justin pointed out, I believe there is this problem where people outside of the field BIS outside of the field thing that they can tell their student, oh, you know, you just go do a tutorial or run a simulation. Interpreting the results. And you now have a section in your manuscript. It will be okay. It's just molecular modeling. And, uh, we need to be aware that while the technological part can be automated. The part and that includes preparation of the setup. Almost never can be automated. So you're going to have a perfectly valid simulation on a technical level. We can have all the technical checklists and quality filters and so on. Uh, if we don't train people to interpret the setup and the results of simulations on a statistical level or on a physical level or on a biochemical level. They will never extract the valley, its conclusions from a perfectly well rounded simulation. So the whole point is this, I think, um, Even if we make things automateable. It just doesn't mean that the science will be automated. And that's a major concern to me, and that's why we need more real collaboration. That's why we should definitely try to do interdisciplinary science, but the way to do interdisciplinary science. And this is not to say it's not being done. Like this is just to emphasize this is the right way. It's not for every individual to do everything, but rather for people to engage in those productive discussions and rely on each other's expertise in solving. those big questions. So that's, uh, you know, when you need. To do this article. I think you can, you can do it yourself with the help of a friend who knows what you're doing and will give you warnings about all the potential problems you might come across or what is actually doable. What is. 20 years from now? And then to. To wrap up the last interesting topic. We have the question of setting our goals in life. So Simon alluded to this question of, of happiness as a point in life where we can follow things that are. Meaningful to us. So, you know, we give things to meaning. We have agency. We get joy from engaging with the scientific questions, and we also have the tools to deal with different aspects of life. Whether it's scientific or private. So, this is a very holistic question. In the sense that we need to have a sense of where we want to go. We need to have a sense of professional proficiency or success that gives us an actual power to do things. Because we cannot really follow our values. If we don't have a say about what we study or what we work on in our professional lives. And then we need to have a way to deal with all the typical things that happen to people, whether it's health or family matters or rejections in professional life or. collapsing life plans or handling professional. And non-professional relationships. This is a big, big conversation. Again, everyone has to have internally to figure out how life in the sciences allows you to reach a deeper sense of happiness. Right? Not just a simple, state of being content with yourself in a moment. But something that is in agreement with our deeper values. Giving us a space to grow. And resilience. And this perspective that we grow together with the people who are growing around us, I think it can help to think. In this terms, because we, scientists are increasingly, you are our interdependency. To come up with say the next discovery at the same time. Many people are driven to academia more by vocation and less by a financial prospects. So we can use the fact that people are looking forward to growth and arrange a growth and kindness centric environment. Again, having this conversation with yourself and people around you. My really benefits. Once clarity and one's path towards this fulfillment. And this seems to be another multi-scale problem in the sense that. We can find meaning on so many levels, you can find meaning in your immediate professional circle, in your lap. Or the interactions with people you mentor or with your mentors, you can find meaning in providing resources for the community, writing perspectives, books, educational materials, you know, running podcasts, you name it. Perhaps. As we discussed with Vlad it's the feeling that you're bringing your expertise to your home country. Contributing to a better future for the people you grew up among. All you can find your happiness in being a public intellectual and trying to reach a broad audience or neither are these. Maybe you find a point you really care about somewhere else. But the point would be to avoid escaping that discussion and having it in such a way that brings clarity into your lives. Sometimes stepping back and reflecting on those questions. Even though we should embrace serendipity and kind of try to extract positive outcomes out of everything. Life Throws at us. Okay. I'm ending on a somewhat philosophical note, but that's among the explicit goals of this podcast. I'm happy. Some of you take by the way, take the opportunity to reach out to me, to give you a feedback. Besides still early days of the podcast. So I take any suggestions fairly seriously. Eh, I'm also feeling more relaxed in my role as a host. So I hope this eventually translates into simply better conversations. It's been an amazing experience so far, and as always I'm grateful to my guests for the time and thoughts they put into the recordings and the audience for curiosity. I'll talk to you soon in season three.

19:16

Thank you for listening. See you in the next episode of Phase Space Invaders.