Computing free energies with PyBrella

Calculations of the rates of disassociation between small molecules and proteins have numerous applications, including assisting rapid discovery and testing of novel drugs. Free energy calculations consider the enthalpy and entropy of the full protein-ligand-water system and so have the potential to be more accurate than faster, single-point calculations. In this study, methods are explored to predict the binding affinity of various molecules to proteins by molecular dynamics and umbrella sampling. An attempt was made to determine the potential of mean force (PMF) for the molecule, which was compared to its known binding capability. Factors including simulation resources, amounts of sampling, force strength parameters, and correlation between predicted energy and actual rate constants were considered in order to evaluate the umbrella sampling methods. Limitations in the simulation environment, such as the scaling of the PMF for sampling, biases in the SMD trajectory, and variations between ligands, were also investigated in the hope of creating a more comprehensive approach for predicting the target-molecule interaction