(360an) Development and Validation of Non-Bonded Interaction Parameters between Coarse-Grained Amino Acid Models and Water

The interactions between amino acids and water are crucial in determining the behavior of many biological molecules as it may play an important role in protein folding, affecting their function. Thus, understanding these interactions at the molecular level is important for us to be able to control their function. Here, we present a set of non-bonded force-field (FF) parameters between 20 coarse-grained (CG) amino acid models and a one-site water model previously developed in our group. These non-bonded interactions between amino acid and water, which are represented by 12-6 Lennard-Jones potential, were obtained by employing particle swarm optimization (PSO) algorithm integrated with an artificial neural network (ANN). At first, we conducted all-atom (AA) MD simulations of the amino acid dipeptides, represented by CHARMM FF, in a TIP3P water box and calculated their Gibbs hydration free energies. Then, the non-bonded FF parameters between CG amino acid beads and the one-site water model were optimized to reproduce the hydration free energy of amino acids from AA MD simulations. The optimized parameters were then validated by utilizing them in CG simulations of the analogues of the amino acid side chains. The results show that the newly developed FF parameters could accurately capture the experimental hydration free energies of the analogues. This also indicates the transferability of the developed models for other molecules in CG MD simulations with water. Additionally, to further validate the accuracy of these parameters, we used the amino acid CG models in MD simulations of peptide amphiphiles to study their self-assembly behavior as well as to study polypeptides in solvents.