[gmx-users] Forcefield question

Mark Abraham Mark.Abraham at anu.edu.au
Mon Feb 13 13:58:51 CET 2012

On 13/02/2012 9:46 PM, Matthias Ernst wrote:
> Hi,
> I am currently working with the AMBER forcefields provided by GROMACS. 
> I noticed there are atoms in the forcefields (e.g. amber99 or amber03) 
> with LJ parameters of 0.0 for sigma and epsilon. For example, the HO 
> atom type is sometimes used (e.g. in SER or in the terminal Hs of DNA 
> residues) but has such paramaters (see below).
> I do not understand (yet) how that can give physically meaningful 
> results. For sigma, I could understand that with combination rule 2, 
> the resulting sigma for a bond is not zero. But the contribution to 
> the total energy vanishes because of epsilon being zero and as such, I 
> think clashes between such an atom and some other atom cannot be 
> avoided because there is no energetic "penalty" for this.
> Am I getting something wrong? Are there some default values that are 
> applied if sigma and epsilon are zero or are those values stored 
> somewhere else but not in ffnonbonded.itp?

No, but the HO atom type is for modelling X-O-H where X is not H. The 
water model(s) which which AMBER are parameterized do not have VDW 
parameters for H in water, and likely the absence of them on alcoholic H 
is for consistency with that. The theory goes that H is too small and 
too labile by comparison with the O, so it may as well not occupy volume.

You may find the issue discussed in the AMBER papers, or confirmed in 
the AMBER force field files (which I hope are freely downloadable 


> Thank you for your help,
> Matthias
> ===============
> This is part of amber ffnonbonded.itp:
> [ atomtypes ]
> ; name      at.num  mass     charge ptype  sigma      epsilon
> Br          35      79.90    0.0000  A   0.00000e+00  0.00000e+00
> HO           1       1.008   0.0000  A   0.00000e+00  0.00000e+00
> HW           1       1.008   0.0000  A   0.00000e+00  0.00000e+00
> FE          26      55.00    0.0000  A   0.00000e+00  0.00000e+00

More information about the gromacs.org_gmx-users mailing list