Is Docking Prepack simply a specialized version of Relax for dealing with complexes? (separate complex, minimize sidechains, reassemble)
If I am docking proteins A and B where each has already been individually relaxed, is there any need for running Docking Prepack on the PDB file that contains both relaxed A and B, already oriented correctly for local docking?
If the monomers are pre-relaxed, you can skip prepack.
It's a problem of false attractors in the conformational space. To improve speed, docking only repacks residues near the protein-protein interface. If one monomer has a bad local clash caused by a bad rotamer, docking is encouraged to find interfaces near that bad rotamer, so that it has freedom to repack it, causing a large improvement in that monomer's energy and a Monte Carlo accept of an otherwise wrong structure. Prepacking removes these issues so that hopefully the only energy well in the landscape is the real docked structure.
I believe prepack is set up to flex only the degrees of freedom that docking itself also flexes - I think it samples far less than relax does.