[gmx-users] LJ parameters

[Erik Lindahl]

Pedro Alexandre Lapido Loureiro wrote:

>
>
>Hello,
>
>I wish to simulate a DPPC bilayer, using the GROMOS96 45a3 LJ parameters for 
>the
>CH2/CH3 of the lipid tails (CL2/CL3).
>
Be warned that Alex de Vries in Groningen has tested this, and the 
Gromos96 force field
still results in much too low areas/lipid, even with the new tail 
parameters.

>
>1)I want to scale the 1-4 interactions *-CL2/CL3 with some fudge factor.
>2)I want to maintain all of the other 1-4 interactions of the GROMOS96 43a1
>forcefield (ffG43a1nb.itp file).
>3)If I say "generate pairs=yes" will GROMACS ignore ALL of the explicit  1-4
>C6/C12 parameters under [pair types] or only those that are not written in
>ffG43a1nb.itp file?
>

No, it works the other way around:

1. If an interaction is given explicitly in the [ pairs ] (1,4) or [ 
nonbond_params ] (normal interactions),
    that value will always be used.
2. If there isn't any entry for a 1,4 interaction in the pairs section, 
it will be set to zero, unless you
    have selected to generate pairs in the [defaults] section in the 
header of the forcefield. If you
    select to generate pairs, the fugde factors will be used.
3. If a normal nonbonded interaction is missing, the combination rule in 
the header will be used.
    (1 = geometric, 2 = lorenz-bertheholz). For instance, Gromos96 uses 
combination rules for almost
    all interactions, so instead of listing all of them we could rely on 
the combination rule and just  
    list the paramters for the exceptions.


Cheers,

Erik