Towards a Universal Force Field for Efficient Atomistic Monte Carlo Simulation of Protein Folding in Dihedral Space

The Computational Protein Folding Prediction (CPFP) problem is one of the main research problems in structural bioinformatics that has been studied for well over half a century and is still the subject of substantial research. The challenge in CPFP is to computationally fold 1D linear polypeptide chains into functional 3D-shaped proteins. Simulations of protein folding with full atomistic detail and explicit water molecules are limited to short polypeptides and require massive computing and financial resources. Recently, we developed Monte-Carlo simulations of proteins with implicit water molecules in dihedral space [1,2] that is computationally efficient and affordable. The model does not compromise the atomistic resolution of the simulated protein. Efficient simulations in dihedral space are based on a coarse-grained model that was designed to retains atomic-level resolution of steric repulsion of heavy atoms and backbone hydrogen bonds, but is based on a coarse-grained representation of the residue-residue interaction potential in the form of a 20x20 matrix that defines the energetic cost of pairwise contacts for the 20 naturally occurring amino-acid residues. It was previously demonstrated that a single interaction matrix is not sufficient for the calculation of energy of contact for all proteins [3]. The focus of this work is the source of the disparity of residue-residue interaction matrices for different proteins and to prepare the grounds for the development of a universal force field for the efficient atomistic Monte Carlo simulation of protein folding in dihedral space.

[1] B. Haimov and S. Srebnik, “Assessment of hydrophobicity scales for protein stability and folding using energy and RMSD criteria,” bioRxiv, 2017.

[2] B. Haimov and S. Srebnik, “Efficient Atomistic Monte Carlo Simulation of Protein Folding in Dihedral Space,” presented at the Third CCPBioSim/CCP5 Multiscale Modelling Conference, Manchester, UK, 2018.

[3] M. Vendruscolo and E. Domany, “Pairwise contact potentials are unsuitable for protein folding,” J. Chem. Phys., vol. 109, no. 24, pp. 11101–11108, 1998.