[gmx-users] how to mimick explicit hydrogen bonding

[Mark Abraham]

ms wrote:
> Hi,
> 
> I need to implement a FF which includes directional hydrogen bonding. It
> seems however, to my understanding, that Gromacs has no simple way to
> include an explicitly directional contribution. All I can find is that,
> in standard force fields, h-bond terms are implicitly described by
> coulombic and vdw interactions, but I am a bit puzzled by how these
> terms are combined to give a directional interaction.

There's nothing directional about the physics of a hydrogen bond, unless 
your model makes it so. There'd be nothing intrinsically valid or 
invalid with that either, so long as you parameterized the force field 
under that assumption. If using rigid water models and/or other atomic 
constraints where H-bonded atoms can't have a geometric distortion, I 
suppose such a model might be necessary. It's still far from clear that 
H-bonding is sufficiently different from other electrostatic 
interactions to warrant special treatment (and thus reparameterization 
of charges).

> Can anyone give me hints on how to describe a directional
> hydrogen-bond-like interaction?

You would not achieve this in GROMACS without code modification. A 
special non-bonded list for H-bonded water (and maybe H-bonded 
non-water) that didn't call the standard water inner loops would be 
required.

MArk