[gmx-users] Gromacs force field discussion

[Ran Friedman]

David van der Spoel wrote:
> patrick fuchs wrote:
>> Hi David,
>> since updated GROMOS force fields are now available within GROMACS, I
>> agree one does not need to use ffgmx, at least for standard simulations.
>> The only problem is for lipid/protein simulations. The only publicly
>> available I know that runs under GROMACS is the combination of Berger
>> lipids with ffgmx (on the website of Peter Tieleman). For someone
>> starting a project of lipid/protein simulations, there is a priori no
>> other public alternative (using GROMACS), although I know some new
>> parameters are about to be published.
>> Cheers,
>
> But the Berger et al force field is based upon OPLS, and hence it can
> be combined with OPLS all atom FF.
>
These parameters were developed with united-atom OPLS. Tieleman and
co-workers introduced modifications to allow the usage of these
parameters with all atom OPLS (Tieleman et al., J. Phys. Condens. Matter
18 S1221-34, 2006). I would use it with cautious though, the results
show significant differences between ffgmx and OPLSAA (Table 1 of the
paper above). In addition, there can be an error in the range of 10
kCal/mol in the calculated cyclohexane to water transfer energies with
OPLS (in any combination, Table 3 of Tieleman et al.). I found this
alarming, but haven't seen results from other FFs - the error may be
just as big.

Ran.