Open Arthurcarre opened 1 month ago
Hi, Arthur,
Let me try and summarize to check I understood the issue:
Docking (ligand + water) -> protein
does not give the same score as docking ligand -> (water + protein)
The difference lies in the intramolecular section (SCORE.INTRA
) of the scoring function, the intermolecular contribution (SCORE.INTER
) has the same value in both cases.
I don't think I've ever docked ligands with water before, I'll need to do a couple checks in the code and with the example system you provided. Are you performing a tethered docking?
Regards
Hi Guillermo,
There seems to be a misunderstanding.
When I dock a ligand, I get a docking score
corresponding to the (ligand)
docking pose. When I rescore the (ligand)
docking pose, I get rescoring score
= docking score
.
When I dock a ligand with a water molecule (ligand + water)
, I get a docking score
corresponding to the docking pose (ligand + water)
. When I rescore the docking pose (ligand + water)
, I get rescoring score
≠ docking score
.
Does that make things clearer ?
Bests, Arthur
Description When docking a ligand with HOH (water molecules) using either the standard scoring function or the solvation scoring function, I obtain an SDF file containing the docked ligand, the HOH molecules, and a score (in my case the score was -20.9355).
If I remove the HOH molecules from the SDF pose try to rescore the SDF pose with the same scoring function providing the HOH molecules docked in PDB format in input to the rbdock function, I expect to get exactly the same score. However, I end up with a rescoring pose in the output SDF containing the ligand + HOH and a score of -25.0127.
If I repeat the same experiment of rescoring the SDF pose but without the HOH molecules docked PDB file in input, the score is -19.8285.
The point I want to highlight here is that for both scoring function, if you score a ligand with docked water molecules, the score obtained is different from the docking score.
I investigated this further and found that the difference lies in SCORE.INTRA, which "represents the relative energy of the ligand conformation" according to the documentation. The pose at the end of docking is penalized by SCORE.INTRA, but the rescoring pose is not. SCORE.INTER, which "is the main term of interest as it represents the protein-ligand (or RNA-ligand) interaction score," remains the same in both the docking and rescoring outputs.
I tried to verify if this point holds outside of docking with HOH, and it does not; the scores are the same.
Steps to Reproduce 1) Dock a ligand + One or several HOH molecules. 2) Delete the HOH molecules docked in the SDF output pose and write these HOH molecules docked into a PDB file in rbdock expected format (Illustration below). 3) Rescore the SDF pose with the same scoring function, providing the HOH molecules in PDB format. 4) Note the score difference between the docking and rescoring outputs. 5) Repeat the rescoring without providing the HOH PDB input. 6) Compare the scores in all scenarios.
Expected Behavior The score obtained from rescoring a pose with the same scoring function and identical conditions should match the original docking score.
Actual Behavior The score obtained from rescoring a pose with HOH molecules docked provided separately does not match the original docking score. The difference is due to the SCORE.INTRA term, which penalizes the docking pose but not the rescoring pose.
Additional Context This discrepancy is only observed when docking ligand and HOH molecules. When docking only ligand, the scores are consistent between docking and rescoring.
Example of SDF output pose of
rbdock
command docking water molecules.I extract the water molecules from the output SDF and write them into PDB file.
The resulting SDF pose without water molecules docked.