![\"\"](\"https://www.epfl.ch/labs/deplanckelab/wp-content/uploads/2025/11/Fondation-Latsis-logo.png\"){kind=logo}
You can apply to participate and find all the relevant information (speakers, abstracts, program,...) on the event website: https://www.cecam.org/workshop-details/from-data-to-dynamics-machine-learning-in-statistical-mechanics-and-molecular-simulations-1487.
\r\n
\r\nRegistration is required to attend the full event, take part in the social activities and present a poster at the poster session (if any). However, the EPFL community is welcome to attend specific lectures without registration if the topic is of interest to their research. Do not hesitate to contact the CECAM Event Manager if you have any question.
\r\n
\r\nDescription
\r\nSince its introduction in the 1970s, molecular dynamics (MD) has become an indispensable computational microscope for studying complex biological systems at atomic resolution. It has enabled detailed investigations into protein folding, conformational dynamics, and ligand binding and unbinding. Over the past decade, increasing computational power has made microsecond-scale simulations routine, producing massive datasets that demand sophisticated analysis strategies [1]. Despite these advances, conventional MD simulations still face a fundamental limitation: many biologically relevant events occur over milliseconds to seconds—timescales largely inaccessible to standard MD.
\r\nTo bridge this gap, researchers increasingly turn to enhanced sampling techniques—such as metadynamics and umbrella sampling [2,3]—and coarse-grained (CG) modeling approaches [4]. These methods enable more comprehensive exploration of the system’s free energy landscape, yet their success critically depends on the selection of appropriate reaction coordinates or collective variables (CVs). CVs must capture the slowest, most functionally relevant motions to accurately reflect thermodynamic and kinetic behavior. However, identifying suitable CVs remains one of the field’s most challenging tasks, typically requiring domain expertise and iterative refinement [5, 6].
\r\nThis complexity has fueled growing interest in machine learning (ML) techniques, which are now transforming how MD simulations are analyzed, interpreted, and even conducted. ML methods have been applied to automate CV discovery, perform dimensionality reduction, build thermodynamic and kinetic models, and enhance sampling efficiency [7]. These models often employ artificial neural networks or graph neural networks to map high-dimensional molecular configurations—such as Cartesian coordinates or molecular descriptors—into low-dimensional representations suitable for analysis [8].
\r\nDepending on the structure and type of data, ML algorithms can be broadly categorized into supervised, unsupervised, and reinforcement learning paradigms [9]. Supervised learning uses labeled input-output pairs to predict properties such as molecular energies or binding affinities [10], while unsupervised learning enables the identification of latent features, such as CVs, directly from data [11].
\r\nA cornerstone of modern ML-driven simulation is the development of symmetry-aware molecular representations. The predictive power of ML models hinges on encoding physical symmetries—like rotation and translation—directly into the model. E(3)-equivariant neural networks have emerged as powerful tools for this purpose, significantly improving data efficiency and generalization in learning potential energy surfaces [12]. Ongoing research continues to explore the optimal balance between enforcing strict symmetry and retaining model flexibility.
\r\nMeanwhile, breakthroughs in structural prediction—most notably the advent of AlphaFold 3—have revolutionized how researchers obtain initial molecular configurations. AlphaFold now provides remarkably accurate models of not only proteins but also their complexes with nucleic acids, ions, and small-molecule ligands [13]. However, these are static snapshots. They cannot capture dynamic behaviors, allosteric transitions, or binding kinetics—areas where physics-based simulations remain indispensable. Initial benchmarks suggest that even state-of-the-art predictors still fall short in modeling protein dynamics and ranking ligand binding affinities, further emphasizing the role of MD [14].
\r\nTo address the dimensionality and sampling bottlenecks, unsupervised ML approaches such as time-lagged autoencoders have reframed CV identification as a data-driven task. More recently, generative models—including diffusion models and variational autoencoders—have emerged as a new frontier. These models can learn the full conformational landscape of biomolecules and enable enhanced sampling, in some cases eliminating the need for predefined CVs altogether [15].
\r\nOnce accurate structural models and CVs are established, ML can significantly improve the estimation of thermodynamic and kinetic properties. In drug discovery, for instance, predicting protein–ligand binding affinity remains a central challenge. ML potentials trained on quantum mechanical data can be combined with enhanced sampling to yield highly accurate free energy landscapes and binding kinetics—results previously unattainable due to computational limitations [16]. However, challenges in data quality, model interpretability, and transferability remain critical areas of ongoing investigation [17].
\r\nFinally, ML is driving a renaissance in CG modeling. Deep neural networks can now learn many-body CG potentials directly from all-atom simulations, capturing emergent properties and enhancing transferability [18]. These models open the door to longer, larger-scale simulations with greater physical accuracy.
\r\nIn this rapidly evolving context, it becomes imperative to critically assess both the promise and limitations of ML in biomolecular simulation. The excitement surrounding these developments must be tempered by careful validation and benchmarking. This workshop thus serves as a timely opportunity—especially for early-career researchers—to explore these cutting-edge methods, engage in constructive dialogue, and chart new directions in the application of machine learning to molecular dynamics and drug discovery.
\r\n
\r\nReferences
\r\n
\r\n[1] J. Behler, M. Parrinello, Phys. Rev. Lett., 98, 146401 (2007)
\r\n[2] P. Sahrmann, G. Voth, Current Opinion in Structural Biology, 90, 102972 (2025)
\r\n[3] K. Kříž, L. Schmidt, A. Andersson, M. Walz, D. van der Spoel, J. Chem. Inf. Model., 63, 412-431 (2023)
\r\n[4] K. Ahmad, A. Rizzi, R. Capelli, D. Mandelli, W. Lyu, P. Carloni, Front. Mol. Biosci., 9, (2022)
\r\n[5] S. Mehdi, Z. Smith, L. Herron, Z. Zou, P. Tiwary, Annual Review of Physical Chemistry, 75, 347-370 (2024)
\r\n[6] H. Zheng, H. Lin, A. Alade, J. Chen, E. Monroy, M. Zhang, J. Wang, AlphaFold3 in Drug Discovery: A Comprehensive Assessment of Capabilities, Limitations, and Applications, 2025
\r\n[7] J. Abramson, J. Adler, J. Dunger, R. Evans, T. Green, A. Pritzel, O. Ronneberger, L. Willmore, A. Ballard, J. Bambrick, S. Bodenstein, D. Evans, C. Hung, M. O’Neill, D. Reiman, K. Tunyasuvunakool, Z. Wu, A. Žemgulytė, E. Arvaniti, C. Beattie, O. Bertolli, A. Bridgland, A. Cherepanov, M. Congreve, A. Cowen-Rivers, A. Cowie, M. Figurnov, F. Fuchs, H. Gladman, R. Jain, Y. Khan, C. Low, K. Perlin, A. Potapenko, P. Savy, S. Singh, A. Stecula, A. Thillaisundaram, C. Tong, S. Yakneen, E. Zhong, M. Zielinski, A. Žídek, V. Bapst, P. Kohli, M. Jaderberg, D. Hassabis, J. Jumper, Nature, 630, 493-500 (2024)
\r\n[8] Fabian B. Fuchs, Daniel E. Worrall, Volker Fischer, Max Welling, NIPS'20: Proceedings of the 34th International Conference on Neural Information Processing Systems, Article No.: 166, Pages 1970 - 1981 (2020)
\r\n[9] H. Sidky, W. Chen, A. Ferguson, Molecular Physics, 118, (2020)
\r\n[10] Y. Wang, J. Lamim Ribeiro, P. Tiwary, Current Opinion in Structural Biology, 61, 139-145 (2020)
\r\n[11] K. Butler, D. Davies, H. Cartwright, O. Isayev, A. Walsh, Nature, 559, 547-555 (2018)
\r\n[12] F. Noé, A. Tkatchenko, K. Müller, C. Clementi, Annu. Rev. Phys. Chem., 71, 361-390 (2020)
\r\n[13] S. Kaptan, I. Vattulainen, Advances in Physics: X, 7, (2022)
\r\n[14] V. Limongelli, WIREs. Comput. Mol. Sci., 10, (2020)
\r\n[15] A. Glielmo, B. Husic, A. Rodriguez, C. Clementi, F. Noé, A. Laio, Chem. Rev., 121, 9722-9758 (2021)
\r\n[16] A. Pak, G. Voth, Current Opinion in Structural Biology, 52, 119-126 (2018)
\r\n[17] M. Lelimousin, V. Limongelli, M. Sansom, J. Am. Chem. Soc., 138, 10611-10622 (2016)
\r\n[18] C. Abrams, G. Bussi, Entropy, 16, 163-199 (2013)\r\n
Instrumented asset pricing models use large sets of firm characteristics to identify latent sources of systematic risk. As the number of available characteristics continues to grow, a natural question is whether additional instruments provide genuinely new information or instead repackage existing information in higher-dimensional form. This thesis-based work studies that question from a geometric perspective, building on Instrumented Principal Component Analysis and the characteristic-based asset pricing literature.
\r\n
\r\nUsing a broad panel of equity characteristics, I document that the effective dimensionality of the characteristic space is substantially lower than the raw instrument count suggests, and that the resulting long-run geometry is highly anisotropic. In such a space, conventional Euclidean notions of orthogonality can be misleading, allowing latent and observable factor directions to overlap. Motivated by this observation, I develop a no-arbitrage geometric extension of IPCA that incorporates empirical characteristic geometry through a QZ-based metric and an economically disciplined identification structure. Preliminary evidence suggests that accounting for this geometry is important for factor identification and is associated with improved cross-sectional pricing relative to standard Euclidean IPCA specifications.
\r\n
European Framework Programs for Research and Innovation.
\r\n
\r\nIs your laboratory involved in one or several European projects, and is one of your responsibilities to ensure their effective administrative, financial, and human‑resources management? If so, it is important to take the time to become familiar with the structure of these projects and with the specific management rules that apply to them.
\r\n
\r\nThis training course will guide you through the contractual documents and their relevant content for project management. Administrative, financial, and HR management rules will be presented and discussed using practical case studies.
\r\nThe training will also help participants become familiar with the European portal and the SEFRI reporting platform.
\r\n
\r\nIt is recommended to attend this training as early as possible, ideally at the start of the project.
A hands-on introduction to the key concepts in image analysis for your everyday research!
\r\n
\r\nOpen to PhD students from all doctoral programs at EPFL, Swiss academic institutions and ETH domain.
\r\n
\r\nJune 8 to 12 2026
\r\nPalace de Caux, Montreux
\r\nPre-summer school Workshop will take place on June 4, 2026 (afternoon)
\r\n
\r\nMore info
\r\nApplication Deadline: March 1, 2026
\r\n
\r\nAre you a PhD student at EPFL, at another swiss academic institution or within the ETH domaine who regularly faces questions regarding the analysis of your images ? Do you want to learn more about practical concepts and tools to help you in this endeavor? Then our summer school in image analysis is for you! Throughout the week, a series of lectures will provide you with the essential concepts in image analysis – from the nature of digital images through the physics of image acquisition to the basics of deep learning, and more. In addition to these lectures, you will learn to use some popular image-analysis software during practical sessions.
The EPFL Latsis Symposium 2026: “Decoding the Cell: Modeling, Predicting, and Engineering Cellular States” will be held on October 29-30, 2026, at the Olympic Museum in Lausanne.
\r\n
\r\nThis international gathering will bring together leading scientists in single-cell analysis, computational modeling, and cellular engineering to explore how recent breakthroughs in multi-omics technologies, predictive algorithms, and synthetic biology are reshaping our understanding of cellular function.
\r\n
\r\nThrough interdisciplinary talks and discussions, the symposium will spotlight advances in single-cell multi-modal data integration, predictive modeling of cell identity and behavior, and the engineering of synthetic cell states. By connecting researchers across experimental and computational domains, the event aims to establish new conceptual and technological frameworks for modeling and controlling cellular systems.
\r\n
\r\nHosted by EPFL, the symposium will foster scientific exchange, spark new collaborations, and accelerate progress toward next-generation cell-based therapies, disease models, and synthetic biological innovations.
\r\n
\r\nJoin us in Lausanne to connect with the global community shaping the future of cell understanding and engineering.
\r\n
\r\nSPEAKERS:
\r\n• Gray Camp – Research Group Leader at the Roche Institute for Translational Bioengineering (ITB) in Basel
\r\n• Barbara Engelhardt – Senior Investigator, Gladstone Institutes & Professor, Department of Biomedical Data Science at Stanford University
\r\n• Jeremy Gunawardena – Professor, Department of Medicine and Life Sciences (MELIS), Pompeu Fabra University, Barcelona
\r\n• Muzlifah Haniffa – Head of the Cellular Genomics Programme and Deputy Director of the Wellcome Sanger Institute and Professor of Clinical Dermatology at the University of Cambridge
\r\n• Anshul Kundaje – Associate Professor of Genetics and Computer Science at Stanford University
\r\n• Prisca Liberali – Professor at the Department of Biosystems Science and Engineering, ETHZ & Senior Group Leader at the Friedrich Miescher Institute for Biomedical Research
\r\n• Ewa Paluch – Professor of Anatomy in the Department of Physiology, Development and Neuroscience and Fellow of Trinity College at the University of Cambridge
\r\n• Steve Quake – Professor of Bioengineering and Applied Physics, Stanford University
\r\n• Sussane Rafelski – Sr. Director, Quantitative Biology at the Allen Institute for Cell Science
\r\n• Kevin Tsia – Professor in the Department of Electrical and Electronic Engineering and the Program Director of the Biomedical Engineering Program at the University of Hong Kong
\r\n• Bo Wang – Chief AI Scientist, Vector Institute & Assistant Professor, University of Toronto
\r\n
\r\nGenerously supported by\r\n
A summer school taking place this September will bring together EPFL and ETH Zurich PhD students to explore how uncertainty can become a productive resource in design and fabrication through a hands-on construction robotics challenge.
\r\n
\r\nThe program combines lectures in architecture, robotics, and work psychology with the full-scale construction of a timber structure from irregular timber stock, using an ABB industrial robot in a human-robot fabrication workflow.
\r\n
\r\nNo previous experience in construction robotics is required, only curiosity.
\r\n
\r\nMore details, including the program, dates, and registration information, are available on the website: https://cap-uncertainty.epfl.ch/
\r\n
\r\nFor questions, please contact: cap-uncertainty@epfl.ch