If you’re using the ligand_docking application, then the ligand (particularly the nbr_atom) should *start* at the position provided by the -start_from option. During docking, there’s a possibility that the ligand will move substantially from that position, particularly if it’s a bad one. (The ligand “shoots out” of the pocket due to clashes, for example.)
Are you consistently getting the ligand to be 50+ Ang away from where you want it to be, or is this movement just one structure? (Ligand docking runs are stochastic, and you will occasionally get ligands which fly out of the pocket. Those should be easy to discard based on their poor energies, though.)
Also, is there anything particular about the location where the ligand does end up? Is it in contact with another part of the protein? Is it similar to the input coordinates for the ligand? If so, that might indicate that the -start_from option isn’t being correctly recognized. (What is your full commandline and options file?)
.
