[gmx-users] dihedral MM scan

Justin Lemkul jalemkul at vt.edu
Thu Jan 26 13:54:16 CET 2017

On 1/26/17 12:30 AM, Mohsen Ramezanpour wrote:
> Regarding the CGenFF article by Vanommeslaeghe et al. (2009):
> It is mentioned that:
> "... are correlated, as are their parameters. In such cases, the individual
> dihedrals are still scanned at the QM level with all other degrees of
> freedom, including other dihedrals that have to be fit allowed to relax.
> However, when scanning the corresponding MM surfaces for a given dihedral *it
> is necessary* that the remaining dihedrals associated with parameters that
> are being fit *be restrained in the QM optimized structures from the QM PES*.
> The use of these additional restraints assures that the energy
> contributions from those dihedrals are taken into account when performing
> the fitting" (Vanommeslaeghe et al. (2009))
> When I did the MM scanning, I ignored the other associated dihedrals (some
> of them were highly correlated) which are in their local minima in QM
> scanning. However, I did an EM step with a tolerance of 50 and I assumed
> that those associated dihedrals will go into their local minima which
> should be similar with QM case.
> If I understood correctly from this article, I should use exactly the QM
> optimized structure for each specific angle, and apply restraints to all of
> them (except the one that I am trying to scan) during zero-step MD energy
> calculation. That being said, taking the structure from that and
> restraining, I do not need the EM step by MM force field any more. Is this
> right?

No, it's not.  The problem that is being discussed above is that if you have two 
dihedrals that use the same parameters, when trying to do the fitting, if you're 
messing with the dihedral parameters, it's affecting two dihedrals, therefore 
throwing off the conformational energy doubly.  So you wind up fitting the wrong 
thing.  In your case, since you have two equivalent dihedrals, you need to be 
fixing one of them at exactly its value in the QM-optimized structure, then 
perform the scan of the other dihedral.  That way, there is a constant energy 
contribution from the fixed dihedral.  Remember, you're doing a *relative* 
potential energy scan and fitting to it.

The single-point energy requires no additional restraints or manipulations; you 
just need mdrun to tell you "the energy of this configuration of atoms is X." 
That's it.  Restraints are only applied during the energy minimization and 
should be turned off completely during the single point evaluation to avoid 
spurious contributions from the restraint term(s).



Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441


More information about the gromacs.org_gmx-users mailing list