So I've ran my fixbb rosetta design script and have been happily doing some analysis work, but it just occured to me that I had stripped all the water molecules from my input protein structure and I'm curious whether that biases the rosetta design in a negative way.
What I want to do with my project is to take a fixed-backbone structure of my choosing, optimize the sidechain identities using Rosetta, and then ultimately express it in a cell to see if the optimized sequence folds back into the original backbone conformation.
My general workflow is to run molecular dynamics of my solvated structure --> strip water from output trajectories --> perform clustering to pick out representative structural conformations --> plug these into fixbb Rosetta design to generate candidate sequences.
Should I be keeping water molecules in my structure before plugging these into Rosetta as that will be more 'representative' of the actual in vivo cellular environment? Or does it not really matter?
Curious to hear people's thoughts.
Rosetta's default scorefunctions use implicit solvation. Individual explicit waters are fine, but generally you should leave water out without a good reason.
Look for a paper from Ryan Pavolvicz and Frank DiMaio for details on their WaterBoxMover code, which is meant to place waters around a desolvated protein.
In no case should Rosetta be offered bulk solvent, that will not end well.