Non-bonded pair parameters for an inserted molecule (modifying the topol.top))

User discussions
1

GROMACS version: 2018.7
GROMACS modification: No
Dears,
I know it might be a very basic question, but I am a little puzzled at this point.
I have a query regarding an issue with non-bonded interaction between two molecules in my system.
I am looking at the interactions between cellulose-hemicellulose. To build the topology, I have used GLYCAM06 parameter sets from Amber and converted the topologies to GROMACS format for each configuration separately. So I have two topology files for cellulose and hemicellulose. In each [moleculetype] I have full sets of [ atoms ] , [ bonds ], [ pairs_nb ], [ angles ] and [ dihedrals ].
My question is: does it make sense to only include both the [ moleculetypes ] into one topology file? In that case, how does GROMACS deal with the non-bonded interactions between the two molecule types? I mean, there would be no explicit c6, c12 defined for inter-molecular interactions in this case. Am I right?
If yes, how can I solve this issue?
I very much appropriate your help and guidance.
Kind regards,
Ali

2

It’s not necessary and is not normally how one would construct a topology.

Of course there would. All LJ interactions are computed according to the force field combination rules using the atom type LJ parameters in ffnonbonded.itp. One does not need to combine [moleculetype] entries to get correct nonbonded interactions.

3

Dear Justin,
Thank you for your prompt reply.
So how should I construct it? Shan’t I have two [ moleculetypes ] sections each for the cellulose and hemicellulose, for instance?
And what about the LJ interaction between two pairs of atoms, each from a different molecule type?
In this case, shall I ask GROMACS to generate the pairs through gen-pairs? If yes, what should be the value for fudgeLJ? 1?
Thank you again for your help.

4

A [moleculetype] doesn’t have to have any physical meaning, but are usually assigned to a complete molecule or polymer. You have two different species, so you’ll have a total of two [moleculetype] directives, one for cellulose and one for hemicellulose.

What about them? Every atom interacts in the simulation; it doesn’t matter what [moleculetype] they are in.

If the force field calls for it, yes. AMBER force fields have gen-pairs set to yes in forcefield.itp.

No idea. What scaling factor does GLYCAM06 require for 1-4 interactions?

5

Thank you very much Justin. You made it all clear now.
Kind regards