I want to do binding energy calculations: The basic idea is the obvious to compare the energy of the complex with the energy of the ligand and the apo-protein.
1. Rosetta Dock does something like this: "the interaction energy between protein and ligand only (interface_delta); this is the score difference between the components together and the components pulled apart by 500A". Regarding this one question. Are the components (especially the protein) relaxed before scoring? Otherwise I guess it would b highly unrealistic, no? Since the sidechains in the binding pocket would most likely arrange different in the absence of the ligand.
2. I tried my own approach, being:
- docking of the ligand --> choosing representative structures
- relaxing the docked structures --> choosing lowest E structure --> rescoring with RosettaScore = COMPLEX
- getting rid of the ligand and relaxing this structure --> choosing lowest E structure --> rescoring with RosettaScore = APO
- relax ligand --> choosing lowest E structure --> rescoring with RosettaScore = LIGAND
While doing this I realized, that the energy of my relaxed docked structure is very different from the original pose directly after docking (both rescored):
after relax 1393.364
original docking -465.427
The major change can be attributed to fa_rep (1946.389 as opposed to 91.803) ... which might be due to the fact that I use soft_rep in docking? But I gathered from the manual that the final minimization is done with normal rep: "## Use soft-repulsive scoring during search (but not final minimization). ## This slightly improves search. Hard-rep is used for final scoring,
## however, because it gives better discrimination."
So, why this very different scores? Am I doing something wrong in docking? Is it sound to do a relaxation (FastRelax) after docking to get a "proper" structure for energy calculations? The ligand position changes slightly during relaxation.
Thanks & cheers