Biophysical Society Bulletin | July-August 2022

Thematic Meeting

Biophysics at the Dawn for Exascale Computers Hamburg, Germany | May 16–20, 2022

Molecular biophysics over the next decade will be dominated by three technologies: electron microscopy and tomography, X-ray lasers, and machine learning. Taking us a step closer toward capturing biomolecular assemblies in action, these technologies, together with molecular simulations, are delivering not only static structures, but movies of cellular functions. One common denominator to this remarkable progress is the advent of the graphics processing unit (GPU) and computer-intensive resources over the past decade. Already leveraging parallel capabilities, areas of diffraction data and single-particle image processing, hybrid modeling, molecular dynamics and free-energy simulations, and drug design and discovery are frontrunners in leveraging the prowess of exascale computing. Fortuitously overlapping with the inception of the exascale era, this meeting started preparing us to advance the development and implementation of computational algorithms toward the best use of exascale computing resources. We brought togeth er experimentalists and theoreticians working in the broad areas of protein folding and assembly, dissection of allosteric pathways, macromolecular interactions, and bottom-up structure of cells, wherein large-scale computing is expected to bring forth major discoveries. Through the talks and extensive discussions, molecular and cellular biologists, chemists, physicists, mathematicians, and computer scientists worked their way to find common ground about sharing innovations and debating future needs so that, as a community, we can move forward to best adapt ourselves to these world-class resources. We had 115 attendees, including 26 invited talks,16 short talks selected from abstracts, and 59 poster presentations. The most remarkable part was one-on-one interactions

between the expert speakers and the younger researchers, which brought forth new considerations on how to train the next generation at the interface of supercomputing, biology, and physics. A common thought that emanated was that we are taking big strides toward marrying computer algorithms with biological complexity. Not surprisingly, several cell-scale models were presented and discussed. These included photosynthetic organelles, water-droplet-bound viruses, crowded cellular environments, and Terasaki ramps in endoplasmic reticulum. Concomitantly, we learned about the power of kinetic models in handling soluble, drug-bound, and membrane-embedded biomolecular complexes. Going beyond proteins, RNA simula tions have matured into a hotbed for development of multi scale methods. Presentations showcased unforeseen progress in the development of more accurate and portable force fields, chromatin folding, and simulations of nucleosome assembly. A number of experiments highlighted the power of single- molecule imaging of disordered systems, extremely high-resolution electron microscopy leveraging new microscopes, and live cell measurements. More importantly, a clear theme emerged on how to incorporate these exper imental data with molecular modeling, which, as a number of speakers stressed, allows the determination of molecular dynamics directly from diffraction or imaging datasets. To this end, machine learning was considered a useful analysis tool, but work is needed to establish its merits in simultaneously handling complexity at different spatiotemporal scales. This need will propel the creation, implementation, and scaling of new hybrid multi-resolution methodologies, for example in transition state sampling, studying molecular recognition in crowded and confined environments, and reaching timescales required to monitor cellular interactions.

July-August 2022

18

T H E N E W S L E T T E R O F T H E B I O P H Y S I C A L S O C I E T Y