Hi developers,
A nice feature that can be integrated is rescoring under dynamic conditions through MMPBSA/MMGBSA. One approach I can envision is to rescore only the top subset of best poses after the Consensus protocol. Since MMPBSA (which theoretically is more accurate than other physics-based scoring functions) should not have the same weight of the other scoring functions.
Due to the fact that this scoring function is quite expesive in terms of computational power; it could be possible to imagine a semi-automatic pipeline in which the user will manually select the poses that makes sense and then with pass them into the MMPBSA final rescoring.
It is my pleasure to share with you my code. Although it is poorly documented and not yet production-ready, it performs this rescoring in an automated fashion, using OpenFF for ligand parameterization and various AmberFF for the protein.
Hi developers,
A nice feature that can be integrated is rescoring under dynamic conditions through MMPBSA/MMGBSA. One approach I can envision is to rescore only the top subset of best poses after the Consensus protocol. Since MMPBSA (which theoretically is more accurate than other physics-based scoring functions) should not have the same weight of the other scoring functions.
Due to the fact that this scoring function is quite expesive in terms of computational power; it could be possible to imagine a semi-automatic pipeline in which the user will manually select the poses that makes sense and then with pass them into the MMPBSA final rescoring.
It is my pleasure to share with you my code. Although it is poorly documented and not yet production-ready, it performs this rescoring in an automated fashion, using OpenFF for ligand parameterization and various AmberFF for the protein.