How to replace vina with other scoring function?

[ldx022]

Hello! It's great to hear that your model is trained based on crystal RMSD, which theoretically should perform very well on the docking power level. I'm curious if this method could be applicable to the following situation: For the same protein system (with the same protein and ligand), many different conformations have been obtained. Since sampling also included the side chains of the protein's pocket, in some systems, even some of the pocket's main chains have undergone slight changes. Could this correction method be combined with RTMScore to address the issue mentioned above?