Dresden 2026 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 15: Computational Biophysics III

BP 15.11: Vortrag

Mittwoch, 11. März 2026, 12:30–12:45, BAR/SCHÖ

AI-Guided Transition Path Sampling of Lipid Flip-Flop and Membrane Nanoporation — •Matthias Post and Gerhard Hummer — Max Planck Institute of Biophysics, Frankfurt, Germany

We studied lipid translocation (or "flip-flop") between leaflets of a bio-membrane and the possible involvement of water nanopores via molecular dynamics simulation.[1] We used transition path sampling[2] within the AIMMD[3] framework to efficiently sample unbiased lipid flip-flops and pore formation. A neuronal network model was trained to predict the "committor", that is the probability of a successful transition from a given microscopic conformational state. While coarse-grained DMPC lipids flip via tunneling through the intact membrane, atomistic (CHARMM36) simulations reveal that flip-flop requires the formation of metastable water pores. DSPC bilayers and plasma membrane mimetics[4] facilitate flip-flop via transient water threads or nanodroplets to cross a locally thinned membrane. Deep neural networks map the transition mechanism from a high-dimensional ( 660d) feature space to a nearly linear committor model, consistent with Cover’s theorem[5] and the concept of dominant reaction tubes.[6]

[1] Post, Hummer, arXiv.2502.11894 (2025) [2] Bolhuis et al., Annu. Rev. Phys. Chem. 53, 291-318 (2002) [3] Jung et al., Nat. Comput. Sci. 3, 334-345 (2023) [4] Pogozheva et al., J. Chem. Inf. Model. 62, 1036-1051 (2022) [5] Cover, IEEE Trans.Electron. Comput. EC-14, 326-334 (1965) [6] E, Vanden-Eijnden, J. Stat. Phys. 123, 503-523 (2006).

Keywords: Molecular Dynamics Simulation; Transition Path Sampling; Membranes; Committor; Lipid Translocation