[gmx-users] non-bonded interactions in GROMACS

[Erik Lindahl]

T. Zaraiskaya wrote:

>
>hello all,
>I wonder if you can help me to understand nonbonded interactions
>in GROMACS?
>
>1.Are coulomb interactions excluded from 1-2, 1-3 and 1-4 interactions?
> L-J are excluded from interactions between first and second bonded neighbours
>and scaled for 1-4 pairs, what about Coulomb, there is
>nothing in manual about it.
>
All nonbonded interactions (including Coulomb) are excluded from the 
first three bonded neighbours by default. You can select an arbitrary 
number of neighbours to exclude in the force field file.

The 1,4-interactions are either given explicitly (e.g. for Gromacs and 
Gromos96 forcefields), or you can scale the normal parameters with the 
"fudge" options in the
force fields. Note that they are multiplicative factors, so use 0.5 to 
decrease it a factor 2. This can be done separately for LJ and Coulomb.

>
>
>2. I am doing energy minimization, steep algorithm, lipid FF is taken
>from ffG43a2x, generate pairs = no.
>When some pairs ( e.g. H - CH1, both atoms with positive charges) are included
>in [pairs] section of the topology with c6 and c12 =0, i am getting a
>positive  pot.energy; when these
>pairs (with c6,c12=0) are not listed in [pairs] the pot. energy has a negative
>value, the term which is changing is Coulomb1-4. This is confusing
>because if Coulomb1-4 interactions were excluded or scaled
>by default i would expect to see smaller Coulomb1-4 term when these pairs are included.
>
The [pairs] section just means to include these pairs as 1,4 
interactions. Optionally, you can specify special LJ parameters there, 
but don't it's better to have those in the [pairtypes] section in the 
force field specification.

The coulomb 1,4 interaction is always determined by the charge of the 
atoms in the pair, although you can scale it with the fudge parameter.

If you remove a pair from that section it will not be included at all in 
the simulation, since there is no 1,4 interaction and the default 
interaction is excluded - both for LJ and Coulomb.

>
>  How to include nonbond.1-4 interactions correctly? Include pairs
>with c6,c12=0 in [pairs]? Place them into [exclusions]?
>something else...?
>
The parameters are already specified in the forcefield. Unless you are 
building your own forcefield it is not a good idea to change them - 
things will be inconsistent.

Just list the 1,4-pairs of atoms in the [pairs] section - the parameters 
should be read automatically by grompp.

The [exclusions] section is only useful if you want to exclude 
additional things (in contrast to the automatic exclusions of neighbours 
up to three bonds away). This is probably only useful when you use dummies.

>
>
>What exactly happens when I use option  -check14 of grompp?
>Are Coulomb interactions in 1-4 pairs removed as well,
>I guess this question recurse to previous.
>
Again - a normal 1,4 interaction involves both LJ and Coulomb, and the 
parameters are taken from the force field. If the LJ parameters happen 
to be zero we will calculate a normal Coulomb-only interaction. If you 
don't want any 1,4 interaction at all if the LJ parameters are zero you 
can use the -check14 option.

Cheers,

Erik