Abstract
Lipid membranes play crucial roles in cellular functions and offer diverse engineering applications. Studying their properties is critical but computationally demanding through atomistic simulations. In this work, we develop coarse-grained (CG) models for phosphocholine lipids, aimed at balancing computational efficiency and predictive accuracy with chemical and temperature transferability. We introduce chargeless CG beads with 2:1 or 3:1 mapping and optimize the force fields through a combination of systematic and accelerated approaches that integrate particle swarm optimization algorithm with molecular dynamics simulations. The optimization utilizes structural and elastic membrane properties obtained experimentally through X-ray and neutron scattering studies including lipid packing density, membrane thickness, and bending modulus. Validation against atomistic simulations shows that our CG models accurately reproduce the structural features of lipids including bond and angle distributions, radial distribution functions, and key bilayer properties across various system sizes and simulation time steps. A unique feature of our CG models is the bead transferability across lipids of different chain structures as well as polymeric macromolecules with similar atomic grouping. This capability facilitates future studies of more complex systems including lipid mixtures, hybrid lipid-polymer membranes, and lipid-glycomaterial complexes─thus offering an efficient platform for predicting structural and functional dynamics while mitigating the computational challenges of atomistic simulations.