General Parametrization Strategies for CHARMM and AMBER

The parametrization schemes for CHARMM, AMBER and many other force fields are quite similar with regard to the bonded interactions while greater differences can be found in the derivation of charges. Typically force fields are parametrized against a combination of experimental and quantumchemical target data, i.e. properties of a molecule that the force field should be able to reproduce. Examples are equilibrium geometries, vibrational frequencies or dipole moments. In comparison CHARMM relies somewhat more on experimental values while AMBER utilizes a reasonable amount of quantumchemically determined values. The equilibrium geometries for the compounds are typically obtained from X-ray crystal structures and for gas phase structures from microwave and electron diffraction data. Force constants can be determined using vibrational frequencies from infrared and Raman spectroscopy. Whenever it is needed supplemental data can be generated by quantum mechanical calculations.

The treatment of charges differs significantly among force fields. In CHARMM they are based on Mulliken charges which are scaled in a way such that the interaction of polar groups with TIP3 water reproduces the interaction energy computed ab initio. For AMBER a method called RESP (restricted electrostatic potential fitting) is used The degree to which one follows these protocols and the amount of effort one invests into the development of parameters for a novel component depends on the purpose for what they should be used. In case an organic ligand just needs to fill a binding pocket in the right conformation an abbreviated protocol might be sufficient. However parameters valid for accurate binding free energy calculations will require much more tuning and testing. PARATOOL aids generating the basic set of parameters that the user might then test against his target data such as conformational stability in simulations or vibrational frequencies.