should I use the crystal structure of the protein directly obtain from RCSB for docking or i should do a molecular dynamics simulation to obtain an average structure and then do the docking with the average structure of the protein.
i would start with the crystal structure; if i could not obtain any reasonable protein-ligand complex, i would dock with representative snapshots (not with average structure) obtained from molecular dynamics simulation.
I was debating this topic early this week; it is not that straightforward as it sounds. It is widely accepted that you don`t need to do anything with your protein before docking your structure to a ligand. However, some researchers argue that if your protein was obtained from crystallography, a Molecular Dynamics step might be useful since, in contrast to NMR, crystals are obtained under high salt concentrations and are farther from a biological reality than those obtained from NMR. In this case, a 10-20ns MD would be enough. Again, this is an ongoing debate; there is no established right-or-wrong at this point. I think that creating a careful ligand topology before a MD (if that is your case) would be way more crucial step than what to do with the initial protein structure.
However, depending on the nature of your ligand, it might be useful for you to check the protonation state of your initial protein structure. This might profoundly impact your docking.
start with crystal structure, that is more important and will result in more accurate results. For docking, in addition to autodock vina use Molinspiration also for 3D structure comfirmations.
Yes, you may use the crystal structure of the protein directly from RCSB for docking purposes. For docking: AutoDock suit, Schrödinger suit may execute. Similarly, for simulation GROMACS tools with desired force fields.