Automatic Optimization of Lipid Models in the Martini Force Field Using SwarmCG

Abstract

After two decades of continued development of the Martini coarse-grained force field (CG FF), further refinment of the already rather accurate Martini lipid models has become a demanding task that could benefit from integrative data-driven methods. Automatic approaches are increasingly used in the development of accurate molecular models, but they typically make use of specifically designed interaction potentials that transfer poorly to molecular systems or conditions different than those used for model calibration. As a proof of concept, here, we employ SwarmCG, an automatic multiobjective optimization approach facilitating the development of lipid force fields, to refine specifically the bonded interaction parameters in building blocks of lipid models within the framework of the general Martini CG FF. As targets of the optimization procedure, we employ both experimental observables (top-down references: area per lipid and bilayer thickness) and all-atom molecular dynamics simulations (bottom-up reference), which respectively inform on the supra-molecular structure of the lipid bilayer systems and on their submolecular dynamics. In our training sets, we simulate at different temperatures in the liquid and gel phases up to 11 homogeneous lamellar bilayers composed of phosphatidylcholine lipids spanning various tail lengths and degrees of (un)saturation. We explore different CG representations of the molecules and evaluate improvements a posteriori using additional simulation temperatures and a portion of the phase diagram of a DOPC/DPPC mixture. Successfully optimizing up to ∼80 model parameters within still limited computational budgets, we show that this protocol allows the obtainment of improved transferable Martini lipid models. In particular, the results of this study demonstrate how a fine-tuning of the representation and parameters of the models may improve their accuracy and how automatic approaches, such as SwarmCG, may be very useful to this end.

Martini 3 Force Field Parameters for Protein Lipidation Post-Translational Modifications

“…For the longer palmitoyl lipid tail, we modeled carbons 1,2,3 with an “SC1” bead, to avoid creating a nonregular-sized terminal bead for the lipid tail, which could possibly lead to excessive interdigitation between the leaflets of a bilayer. This choice is also in line with the recent mapping schemes tested by Empereur-mot and co-workers …”

Extension of the SUGRES-1P Coarse-Grained Model of Polysaccharides to Heparin

“…While this approach proved successful in this study, due to the interdependence of the parameters of the force field, more advanced optimization techniques may be applied in future developments of the SUGRES-1P force field. Recent research showed the successful application of a range of sophisticated optimization approaches, including a variety of machine-learning algorithms. − While such complex parametrization techniques are without a doubt an interesting approach in possible future developments of the SUGRES-1P force field, for the aim of proof of concept of this study, we have restricted ourselves to the empirical approach. For this purpose, we have determined 10 combinations of energy term weights and κ i for each of the three tested HP lengths that resulted in the best agreement with experimental EED and R g values from refs and (detailed in Tables S2–S4).…”

Lessons Learned from Multiobjective Automatic Optimizations of Classical Three-Site Rigid Water Models Using Microscopic and Macroscopic Target Experimental Observables

“…The optimization work conducted herein builds on a multireference particle swarm optimization software that we developed recently: Swarm-CG . , In particular, Swarm-CG has been developed to optimize bonded and nonbonded parameters in molecular models to fit experimental results (top-down references) and the behavior seen in all-atom MD trajectories (bottom-up references). Swarm-CG has been successfully tested to optimize a variety of molecular systems (e.g., lipid models). In this paper, Swarm-CG has been adapted for this specific case study (a dedicated variant can be found at: ).…”

“…Recently, we have observed similar intrinsic limitations also in the optimization of, e.g., coarse-grained models of a variety of other molecular systems. 7,10,27 This observation serves as a valuable lesson for developing models of all kinds and not just in the context of water simulations. Such inherent limitations and these challenges encountered in optimizing approximated models demonstrate the importance of considering the complexity of the system being studied and the type of information lost with approximated molecular models.…”