[gmx-users] GROMACS OPLS-AA question

Barnett, James W jbarnet4 at tulane.edu
Sat Sep 12 16:17:49 CEST 2015

On Thu, 2015-09-10 at 11:32 -0600, Eric Smoll wrote:
> Hello GROMACS users,
> I am interested in using the OPLS-AA as implemented in GROMACS.
> The OPLS-AA forcefield omits all nonbonded (Coulomb and LJ)
> interactions 1
> and 2 bonds away. Nonbonded interactions 3 bonds are away are scaled
> by
> 1/2. Nonbonded interactions are applied without modification for
> atoms 4 or
> more bonds away or between atoms in different molecules.
> In GROMACS, this is accomplished by excluding Coulomb and LJ
> interactions
> 1, 2, and 3 bonds away with the following directive:
> [ moleculetype ]
> ; name nrexcl
> System 3
> The scaled intramolecular interaction at 3 bonds away is added in
> separately as a "pair" interaction. This is either done manually with
> "[
> pairtypes ]" and "[ pairs ]"  directives or automatically by
> including
> "yes" as the third entry under the "[ defaults ]" directive.
> *Is this correct?*

Essentially this is correct. 1-2, 1-3, and 1-4 nonbonded interactions
are excluded when the number of exclusions is 3. With [ pairs ] and [
pairtypes ] 1-4 nonbonded interactions can be explicitly added back
(the point is that you 1-4 nonbonded interactions typically have
different parameters, so the normal non-bonded parameters for 1-4 are
excluded and then different parameters are added back with [pairs]).
OPLS doesn't do that, but instead generates the 1-4 nonbonded
interactions as specified in the [ defaults ] section.

> The manual states that "the use of [Ryckaert-Bellemans potential
> functions]
> implies exclusion between the first and last atom of the dihedral."
> *This
> statement is referring to the use of  nrexcl=3 in the "[ moleculetype
> ]"
> directive as shown above, correct? The use of RB dihedral functions
> in
> GROMACS does not imply that the scaled "pair" interactions are not
> applied,
> correct?*

My understanding is it is saying 1-4 nonbonded interactions should be
excluded by setting nrexcl=3 when using RB dihedral functions. Clearly
OPLS uses the RB form in GROMACS and generates pairs, so pairs (either
generated or set with [ pairs ]), still apply.

That's my understanding - but you could test it out by setting generate
pairs to no in the [ defaults ] section and see if GROMACS complains
that [ pairs ] is missing even with the RB form of the potential. My
guess is that it will.

James “Wes” Barnett, Ph.D. Candidate
Louisiana Board of Regents Fellow

Chemical and Biomolecular Engineering
Tulane University
341-B Lindy Boggs Center for Energy and Biotechnology
6823 St. Charles Ave
New Orleans, Louisiana 70118-5674
jbarnet4 at tulane.edu

More information about the gromacs.org_gmx-users mailing list