Biophysical Society Thematic Meeting | Hamburg 2022

Biophysics at the Dawn of Exascale Computers

Friday Speaker Abstracts

COMPUTER SIMULATIONS OF LIPID-PROTEIN INTERACTIONS Peter D. Tieleman ; 1 University of Calgary, Biological Sciences, Calgary, AB, Canada

Membrane proteins and membranes in general have been simulated for almost 3 decades, during which the scope of simulations has expanded by 6-8 orders of magnitude. At the most expensive end of the scale, atomistic simulations of systems containing > 100 million atoms are become feasible [1], although they pose serious scientific and philosophical challenges. A somewhat more practical approach is the use of coarse-grained simulations, where in particular the Martini model has become very popular [2], which can again be pushed to the most expensive end of the scale by simulations copies in an effort to generate enough data for machine learning analysis [3]. At a more modest scale, questions on individual proteins often use a combination of Martini and atomistic simulations, e.g. [4]. I will highlight some recent applications in the area of membrane simulations, including challenges that can be addressed with the next generation (or the next 6-8 generations) of supercomputer capabilities. [1] C. Gupta et al. The ugly, bad, and good stories of large-scale biomolecular simulations. Curr. Opin. Struct. Biol. 73, 102338, 2022[2] P.C.T. Souza et al. Martini 3: a general purpose force field for coarse-grained molecular dynamics. Nature Methods 18, 382-388, 2021[3] H. Ingolfsson et al. Machine Learning-driven Multiscale Modeling Reveals Lipid-Dependent Dynamics of RAS Signaling Proteins, Proc. Natl. Acad. Sci. 119 (1) e2113297119[4] W.E. Miranda et al. Lipid regulation of hERG1 channel function. Nature Comm. 12, 1-10, 2021

42