An Accurate Free Energy Method for Solvation of Organic Compounds and Binding to Proteins
Here, we introduce a new strategy to estimate free energies using single end-state molecular dynamics simulation trajectories. The method is adopted from ANI-1ccx neural network potentials (Machine Learning) for the Atomic Simulation Environment (ASE) and predicts the single point energies at the accuracy of CCSD(T)/CBS level for the entire configurational space that is sampled by Molecular Dynamics (MD) simulations. Our preliminary results show that the method can be as accurate as Bennet-Acceptance-Ration (BAR) with much reduced computational cost. Not only does it enable to calculate solvation free energies of small organic compounds, but it is also possible to predict absolute and relative binding free energies in ligand-protein complex systems. Rapid calculation also enables to screen small organic molecules from databases as potent inhibitors to any drug targets.