AMBER Archive (2009)

Subject: Re: [AMBER] Regarding Hbond interactions

From: Thomas Cheatham (
Date: Thu Jul 23 2009 - 17:57:57 CDT

> How Amber handling Hydrogen bond interactions in
> calculations?. If it is included in electrostatic interactions, how one
> can differentiate non bonded electrostatic interactions from H-bond
> interactions?

There is a paper by DM Furgeson and Kollman ~2002 in JCC on how the 6-12 +
electrostatics can mimic h-bonds (as compared to a 10-12 potential).
Essentially, the h-bond represents both electrostatic attraction and
dispersion attraction/repulsion. One can artificially decompose this
energy to represent only the interactions between the hydrogen bonding
atoms, however this can also be misleading as the full interaction
represents a balance of many interactions. This can give qualitative
insight into the relative strength. For example, you could think
of zeroing all charges except those involving the h-bond and approximate
the electrostatic components. However to more precisely get at this this
energy would require free energy perturbations to non-h-bonding base
analogues (see work by E Kool and others) or a pulling experiment where
you pulled apart the base pair to estimate a PMF or effectively calculate
the melting temperature / profile. These latter two approaches are
research questions and likely very tricky since you need to consider the
unfolding states and interactions with water, whereas the former may
provide insight.

As a simple example, consider guanidinium and acetic acid as a plus
charge and minus charge interacting in water...

  + -

The ion pair (similar to a h-bond) has favorable electrostatic energies of
> 50 kcal/mol. However, if you simulate the assocation of the two ions
(or measure it experimentally) the equilibrium constant is near 1 and
overall there is only a very small free energy for ion pairing. To get at
the whole picture you need to worry about entropy, desolvation and the
direct ion-ion interaction. With energetic decomposition, this is tricky;
with umbrella simulations or free energy perturbations, it is more
straightforward (in principle).


AMBER mailing list