does anybody have experience or ideas on how to involve waters in a packing/design process?
I want to repack a protein but want to make sure to include some water filled cavities. Is there a way to define waters or a sort of placeholder, though explicite waters would be better, so that the design-packer can generate a suitable surrounding.
To which Rosetta paper are you referring? I think there have been two attempts to work explicit waters into Rosetta (a hydrated rotamers attempt and another one maybe) but I'm not sure that either were complete or published.
99% of the work done with Rosetta uses only implicit solvation.
I guess the best thing do to would be to use the ligand design modes with an explicit water as a ligand. TP3 and TP5 water are defined at rosetta-3.3/rosetta_database/chemical/residue_type_sets/fa_standard/residue_types/water. The TP5 orbitals are probably ignored, though.
Including waters in packing as independent residues (rather than a hydrated rotamer approach) is tricky due to hydrogen bonding/orientation issues. It would possibly work if you started out with the appropriate hydrogen geometry, but if you were anticipating packing to find/rearrange the hydrogens on the water into an appropriate geometry, there isn't anything that's currently set up to work. Also while we have the TP3 and TP5 residue topologies in the database, practically nobody uses them for anything, and there's random technical issues/bugs with both in stock Rosetta (which may or may not affect you, depending on how you use them).
So if know the final geometry of the water (including hydrogen positions), you can place a TP3/TP5 residue in that fixed location, which might work after a bit of fiddling, but if you're hoping to have Rosetta optimize hydrogen positioning, you're on your own. (Unless you want to dust off the Jiang et al. "solvated rotamer" approach.)
Thanks for your help! I tried with TP3 waters and so far it looks like the design packer realized the waters (space and electrostatic wise). A follow up question: I suppose there is no way to give the waters a restricted amount of possible movement, like poses for a ligand... without explicitly treating them like a ligand. Regarding this: is it possible to use more ligands in Rosetta anyway?
I assume you can use waters as ligands for the ligand docking executable, which will allow them to move around? What do you mean by "more ligands"?