Computing energies of binding

Dear CCLers,
 I am currently trying to evaluate energies of interaction between a group of
 peptides and it's receptor. I am particularly interested in computing these
 energies with solvated compounds. I was thinking to build a thermodynamic
 cycle. The in vacuo step seems clear enough to me, but I was wondering how I
 should consider the problem of solvation. It seems to me that my
 calculations would be inconsistent if I'm not careful.
 Assuming I need N molecules of water to solvate my ligand and M molecules to
 solvate my Receptor. Upon, solvating my in vacuo complex, should I use N+M
 molecules of water so that my system size is consistent ?
 I can see another way to consider the cycle. I could start with a big box
 filled with water, add either ligand, receptor or complex depending on the
 calculation desired and eliminate overlapping water molecules. But then the
 size wouldn't be consistent anymore.
 I am also wondering how one can define a system as properly solvated. I
 would like to keep computational expenses low and ideally I'd like to use a
 layer of a few angstroms on each compounds and restrain the solvent
 molecules from moving away from the system using a small distance dependant
 restrain (from the centre of mass).
 Best regards,
 Julien MICHEL