AMBER Archive (2006)

Subject: Re: AMBER: H-Bond Terms in amber force field

From: Thomas Cheatham (
Date: Wed Nov 15 2006 - 17:51:36 CST

> I have a system which is formed by unnatural base pairs; it is a
> (5'iGiGiCiC3')_2 helix. This helix is more stable than (5'GGCC3')_2
> structure. This is experimentally verified.
> When I use implicit solvent model to simulate both these structures, GGCC
> stays helical while iGiGiCiC falls apart.
> Amber force field does not have an explicit term for H-bonds in the
> potential energy function. In the old amber force fields, there is an
> explicit H-bond term present in the function. I wonder why the H-bond term
> is taken away from the force field, and how the 'H-bond' term is
> incorporated into the new amber force field. This might help me to figure
> out why iGiGiCiC is falling apart and how I can solve this problem. Thanks
> in advance for any suggestions.

"Falling apart" is not very descriptive, but likely something has become
misbalanced in the structure. As the changes are very modest (flipping of
the amino and carbonyls of GC base pairs), I am surprised by your
observations. Things I would consider:

(a) Are the iG/iC models very different (in terms of charges, vdw,
impropers, ...) from GC?

(b) Are you sure that the implicit model is "correct" in that the observed
stability of GGCC compared to the modifed nucleotides was real (i.e. in
multiple runs with different initial conditions is the same behavior

[In my experience with implicit solvent and DNA, results are sometimes
funny, i.e. in a long simulation of a 14-mer duplex is was stable for 500
nanoseconds except for end group fraying then spontaneously denatured.]

As these are small duplexes, you could run each in explicit solvent where
barring any funny problems with the parameterization I would expect both
to be stable. You could even do the TI to convert i-modified nucleotides
to standard...

Finally, there is a paper by Veenstra/Kollman or Ferguson/Kollman, J Comp
Chem ~1996 that discusses how point charge + 6-12 vdw is fine for
reproducing h-bonds; this is when Amber got rid of the 10-12 in
parameterizations. As Sponer's group has shown via benchmarking to high
level QM, the Amber charge model does an excellent job of reproducing
h-bonding in nucleobases...


\-/ Thomas E. Cheatham, III (Assistant Professor) College of Pharmacy
-/- Departments of Med. Chem. and of Pharmaceutics and Pharm. Chem.
/-\ Adjunct Asst Prof of Bioeng.; Center for High Performance Computing
\-/ University of Utah, 30 S 2000 E, SH 201, Salt Lake City, UT 84112
/-\ (801) 587-9652; FAX: (801) 585-9119
\-/ BPRP295A

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to