Ligands should be properly protonated. FRED accepts the input protonation state of the ligands, except in the cases when the protonation state is not fully specified by the input file (such as with PDB and MOL2 files with implicit hydrogens). In these cases FRED will use the OEChem function OEReadMolecule to assign protonation states according to the OpenEye valence model. Within the OpenEye toolset, Quacpac will enumerate protonation states or FILTER can set a single protonation state.
Charges are only needed by FRED when using the Zapbind scoring function. By default FRED will accept input charges on the ligands. However, if the -assign_ligand_charges flag is set to true FRED will assign AM1BCC charges to the ligands [5] [6]. The AM1BCC calculation has a minimal, but non-zero, cost (approximately 1-5% of the total docking time), so users should set charges on their ligand database once rather than recalculating the charges each time FRED is run.
Conformers of the input ligand database must be generated prior to running FRED. Conformers of a given molecule should appear in sequence in the input file. It is not necessary to name conformers of the same molecule in any special fashion, FRED's conformer perception matches the chemistry, not the title.
To ensure that FRED can reproduce the binding pose of a ligand accurately each molecule should have at least one conformer in its ensemble that is within 1.0 Angstrom RMSD of the bound structure. Molecules that have a conformer between 1.0 and 1.5 Angstrom RMSD can be docked with moderate success.