On a related note, I'm given to understand that the TIP3 water model isn't as successful at producing the properties of liquid water as it could be, especially at longer ranges, ie
Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D. "Comparison of simple potential functions for simulating liquid water" J. Chem. Phys 1983, 79, 926-935.
Is this true of the modified model as well? Ignoring the additional computational cost for models like TIP4P and TIP5P for a moment, which water model as implemented in CHARMM best reproduces the properties of water at room temperature? As of 2003, at least, there seemed to be some question
of whether even the implementation of TIP4P was producing good properties in CHARMM. How is it now?