BPS2026 – Challenges of cell-scale molecular modeling and simulation

Ambitious computational efforts usher in an era of cell-scale molecular modeling, seeking to integrate new experimental data into models of ever-increasing completeness and complexity. New experimental insights inform the modeling about complex molecular assemblies that constitute cellular structure thanks to advanced techniques like single particle cryo-EM. Aligning with the goal of cell-scale modeling, we have implemented robust methodologies for the construction of cell-scale membrane compartments composed of lipids and embedded proteins that overcome a number of key challenges. For one, the wide range of membrane protein size and shape that make automated assembly into cell-scale structures difficult. To overcome this, we approximate the shape of desired proteins with ultra-coarse grained (UCG) representations generated by hierarchical clustering. The UCG representation is then used in rapid, cost-efficient molecular dynamics (MD) simulations to randomize the placement of the proteins along the desired membrane surface. As simulation timescales of these massive systems have extended by virtue of advances in computing hardware, new challenges have emerged, specifically associated with the long-term stability of these algorithmically generated membrane structures. To ensure the structural integrity of our models during simulation, we restrain the lipid-protein membrane to its target shape using “grid-forces.” This approach prevents transient delamination of the two leaflets of the lipid bilayer that may be encountered at simulation startup. Additionally, restraint of the membrane structure prevents membrane rupturing upon the initial formation of interior vacuum bubbles, which are subsequently re-filled to ensure optimal solvent packing. Lastly, restraint of the membrane with grid-forces fixes the structure of the membrane to its target, preventing drift from the target structure if its preservation is desired. These protocols mark a crucial step as MD simulations seek to animate larger cell-scale structures of great complexity.