Currently I need to parametrize some simple F and Cl containing solvents to use them in Gromacs. The problem is that common force fields do not contain atom-types I need, so while understanding general concepts (I have a C++ prototype to automatically generate solvent molecule PES via quantum chemical calculation (my prototype calls ORCA to do single points) and fit parameters of OPLS equations to ORCA results), some minor things are unclear for me.
1. Bonds and Angles: whether one usually just change bond length without geometry optimization, or it is relaxed scan along bond (angle)
2. Same question about dihedrals (my most serious concern): say, I have terminal -CHal3 group, so do I need to variate only C-C-C-Hal dihedral (without geometry relaxation) or I should relax geometry (almost the same as rotation of the whole -CHal3 group)
3. 1-4 interactions are subject to non-bonded interactions, am I right, that LJ parameters are also produced from this single molecule fit? Same question about charges (initially they are from AIM analysis).
4. Further I want to fit liquid densities, vaporization heat et.c., what parameters should I fit (LJ and charges?)?
Hello Oleg Gromov,
For this kind of parametrization I would consider the Joyce code: (http://www.pi.iccom.cnr.it/joyce)
It was developed for precise GROMACS parametrizations. Bonds and angles equilibrium values are taken directly from the optimized QM geometry. Bonds and angles force constants are extracted from the diagonalization of the Hessian matrix. Dihedrals potentials are fit from a relaxed scan (or can be described by a harmonic potential if it is a rigid dihedral). For intramolecular LJ and charges parameters I would consider them zero in a first step. And finaly, I would use a second package to fit intermolecular parameters, picky (http://www.pi.iccom.cnr.it/picky).
I don't know if it is a problem for you, but if you opt to use this procedure, the QM calculations must be performed with GAUSSIAN.
I hope it helped. Good luck!
Thanks, Ingrid!
That seems to be exactly what I was going to do (and partially did) on my own. Unfortunately it needs Gaussian calculation while I'm limited with free packages like Orca or Gamess.
In force field parametrization of the solvent the derived from QM force field of molecule is not everything. You need to decide what properties you should reproduce: density, energy of vaporization, temperature of boiling. It is problem. Ab initio calculations is the simpliest part of your problem. We have done several solution topologies in our lab and we now some protocols. You can access us directly if you want.
Thanks, Alex!
Your labs topology generator is a great thing, I will play around its results (as I don't really want to do new atom type parametrization in part of its consistency with the whole force field). Though, I've noticed (maybe erroneously) that it draws Ryckaert-Bellemans dihedral potential in kcal/mol, while Gromacs (at least manual says so) wants kJ/mol.
Sincerely
Oleg
- Badges
- Science topic