An optimized contact map for GoMartini 3 enabling conformational changes in protein assemblies

Abstract

Advances in structural biology, particularly cryo-electron microscopy, have enabled high-resolution characterization of complex biomolecular assemblies. These developments emphasize the need for computational approaches capable of describing biologically relevant conformational changes over extended timescales. G[o]Martini 3 is a coarse-grained approach that demonstrates computational efficiency and versatility across several systems, from protein-binding membranes and soluble proteins to intrinsically disordered proteins, while preserving key physicochemical features. In this work, we introduce an optimized approach that integrates dynamic contact information from AA-MD simulations to refine the contact map in G[o]Martini simulations. Specifically, we define high-frequency contacts (HFC), which reduce the number of original G[o] contacts set by {approx}20-30%, thereby improving the representation of conformational states beyond the original approach. Benchmarking different contact-selection criteria revealed that including intra- and interchain HFC captures structural flexibility and domain dynamics. The method was tested on three small soluble proteins and on the SARS-CoV-2 spike protein. Overall, the optimized contact map improves sampling efficiency and expands the accessible conformational landscape relative to the original G[o]Martini 3 implementation. The full framework is available as an open-source resource for large-scale simulations of biomolecular assemblies.