I simulated the dihedral potential (0 - 360 degrees) of a molecule with Gaussian (B3LYP) to use the parameters as input for my OPLS Force Field calculations.
As C2 (of -C1-C2-C3-C4-) in my molecule is chiral, the potential energy curve is not symmetric in respect to its minimum. This seems to be a problem when trying to fit a consinus function, which is usual in OPLS-FF, to the simulated data.
I guess this is a known issue in peptide and protein simulations but I've got no peptide-chemists around. I hope somebody here can convince me that this is not a problem but only a challenge.
I can not help you with the fitting stuff, but at least recommend to,do your DFT calculations with dispersion correction, like Grimme D3. These are now available in most QM codes and will help to get correct barriers. In principle, you should calculate energies for your critical points with CCSD(T) for at least the fragment with the direct neighborhood to your dihedral to be sure about DFT results.
Around 2000 I published 2 papers about barrier in eg bipyridyls etc which are strongly unsymmetrical.
btw which opls implementation are you referring to? The one in the Schrodinger software? Are the parameters there listed ain some radable format? Are they independent in a way that you can define a dihedrL without compromising definitions for angles and distances?
If I had to guess I would say the problem is due to the way the torsion potential is formulated. It is generally a sum of Fourier terms. If only the odd wave-number terms are included, then the potential is always symmetrical about the minimum. Changing this would require modifying the code to add even-numbered terms.
Hi Thomas,
yesterday, when I read your question, I tried to find out if it is possible to describe chiral potentials by symmetric cosine function generally. The result (AMBER, OPLS, CHARMM, MMFF94, and there are a lot of others force fields...) is, that it does'nt! It An B3LYP asymmetric rotational potential from Gaussian, i.e. an DFT one, cannot be fitted to such a symmetric function like the cosine ore a sum of cosine. Sine and cosine functions are symmetric per se. There seems to be one chance only: switch on the 1-4 non-bonding interactions, which are switched off normally. I discussed this question with a college of me and we find out that this seems to be the only one opportunity describing a chiral potential correctly.
Cheers, Peter
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology