We are docking small molecule ligands into non-enzyme proteins to get an idea of ligand distribution/convergence within the binding pocket, using the LigInterfaceEnergy mover. While the results are promising, ligands with the lowest LigInterface energy are sometimes outliers within the ligand distribution. There is no native structure to compare to other than the input pose.
I am curious if there are specific factors (H-bonds, hydrophobic interactions, ligand orientation, solvent accessibility, etc) that weigh more in determining the LigInterface energy score, so that I can try to find a trend within these structures. We're not including the enzdes style cst_energy, as we do not use specific interface constraints for the dockings (I'm assuming that's where those come from).
EDIT: Perhaps to simplify the question (or maybe this poses a different one): What are the differences, if any, between the LigInterfaceEnergy and interface_delta_x metrics?