I got some useful parameter in PCFF form. I need to convert them in OPLSAA. How do I do that? As I see the functional form of these two for bond and angle is same. What about lenard jones? PCFF use 9-6 lenard jones, how to get 12-6 form in oplsaa?
Thanks in advance.
Force fields are parametrized to be self-consistent entities. You can't just convert one force field into another, especially if the elements of the functional form are different.
Of course, strictly speaking parameters between the force fields are not portable. However, force fields are rather rough things and a small adjustment may be sufficient to obtain reasonable results. OPLS parametrized by density and heat of vaporization of liquids. If we take the data from a different force field and adjust the density and heat, then we may get force field compatible with OPLS.
While in principle one can derive a new parameter set from an existing one, there are cascading problems with this approach. If PCFF uses a 9-6 LJ form, you have to re-do the LJ parameters to use a 12-6 form. Since intermolecular (and some intramolecular) interactions are a balance between vdW and electrostatics, re-deriving the LJ necessitates re-calculation of all charges. Then you may have to iterate back to the LJ and make empirical adjustments if necessary to reproduce whatever the target data are. If you've changed the LJ and the charges, you need to re-do the dihedrals (based on QM energy scans). It quickly becomes a nightmare that can take months or years to do. It is better to find a force field that reasonably covers the parameter space of interest and use it, making small extensions for new compounds as needed, within the existing framework and parameter set.
Repulsive part of VdW potential is just introduced to reproduce short-range repulsion, and because of that can be defined nearly arbitrary (by exp(r), r9, r12). One can try to extract attractive dipole-dipole "long-range" part from 6-9 and directly transfer it to 6-12 potential. The repulsive part in that case can be chosen to keep the same position of energy minima (sigma) and its depth (epsilon) as in 6-9 version. An alternative solution could be to build potential energy curve for 6-9 in Origin and find corresponding parameters of 6-12 by fitting. Of course people usually write, that force-fields are not transferable. But we should take into account, that all these FF are derived from QM. VdW interactions in "cheap" QM realizations are poorely reproduced, the concept of partial charges is questionable in itself, and charges can be very different depending on method of calculations. Thus, original FF normally requires modifications for really accurate simulations, anyway. By my experience, density is more sensitive to bond lengths, vaporization energy mostly depends on partial charges. VdW parameters are of secondary importance, because they vary not so strong from one compound to another. And, in principle, why not to use VdW parameters of OPLS?
- Badges
- Science topic