[gmx-users] forcefield.itp file for gromos45a3 and oplsaa

[Thomas Piggot]

Hi,

In the GROMOS force fields the non-bonded interactions are explicitly 
given as some atom types have more than one C12 value. The different C12 
values are used for interactions with different atom types. This means 
you need to be very careful when modifying the ffnonbonded.itp for the 
GROMOS force fields. If you wish to do this I suggest that you read of 
the GROMOS 53A6 force field paper (DOI: 10.1002/jcc.20090).

Cheers

Tom

On 04/01/12 06:18, Mark Abraham wrote:
> On 4/01/2012 4:43 PM, XUEMING TANG wrote:
>> Hello
>>
>> I am trying to adjust compound LJ parameter in oplsaa (sigma&epsilon)
>> and also in Gromos (C6 and C12) forcefield for comparison. The
>> original ffnonbonded.itp file in oplsaa.ff has no pair potential
>> listed. While the original ffnonbonded.itp file in gromos**.ff listed
>> all the pair potentials. If I change one atom potential, do I need to
>> do anything to generate the new pair potential before running a
>> simulation in opls-aa force field? Or how to generate the pair
>> potentials? While for the case of gromos forcefield, the
>> forcefield.itp file shows no pair potential will be generated. Is
>> there any automatic ways to generate the modified LJ potentials?
>> Should I list every possible LJ pair potentials in nonbonded.itp file?
>> It looks like some very tedious work if several atoms are going to be
>> modified.
>> Also, what does fudge LJ and fudgeQQ mean in forcefield.itp file?
>> Values of those two are different for OPLS and Gromos.
> In general, two force fields work differently and have their own
> idiosyncrasies that will show up in how GROMACS implements them. When
> modifying them, you should be sure to check the original literature and
> the GROMACS manual and satisfy yourself how your changes will work. You
> should search the manual for discussion of the use of gen-pairs and the
> fudge values. I don't know why the list of [ nonbond_param ] values is
> exhaustive - you'd think the combination rules could be used. In
> practice, you only need to add parameters for interactions that can
> occur in your system, and grompp will complain about those that it can't
> find.
>
> Mark

-- 
Dr Thomas Piggot
University of Southampton, UK.