Re: [AMBER] torsion parameters in parm99.dat

From: Jason Swails <>
Date: Wed, 15 Dec 2010 09:55:56 -0500


On Wed, Dec 15, 2010 at 1:42 AM, xqkong <> wrote:

> Dear amber users,
> As a novice of amber, i have some questions about the torsion parameters
> in parm99.dat. I noticed that sometimes there are more than one line of
> torsion parameters corresponding to the given atom types.Such as
> CT-CT-CT-CT 1 0.18 0.0 -3.
> CT-CT-CT-CT 1 0.25 180.0 -2.
> CT-CT-CT-CT 1 0.20 180.0 1
> Therefore, I have the following questions,
> First of all ,why do we need three kinds of parameters to describe a
> given torsion angle ? what does the

Think of this not as 3 different parameters, but rather 3 terms that compose
a single parameter. The negative periodicity at the end indicates that it's
a "multiterm" dihedral parameter.

> chemical meaning of this configuration ?I noticed that the difference in
> the periodicity range from 3 to 1.But we
> know that when we rotate a bond consists of four sp3 carbons from 0 to
> 360 degree, the number of the

In the Amber force field, it would be a mistake to try and glean chemical
insight out of the torsion parameters; in general there is none. Instead,
think of the torsion as a simple correction. This is because 1-4 non-bonded
interactions (which are interactions between atoms separated by 3 bonds; or
those that compose a dihedral angle) are included in the force field
(although the Amber force field does scale them). Therefore, part of the
torsion potential is already taken care of by other terms! Because the force
field is not *perfect* in this regard, however, and because there are
occasionally quantum mechanical effects (such as breaking conjugation in
certain conjugated systems) that can't be modeled by a classical, non-bonded
potential, the difference between the torsion potential accounted for by 1-4
terms and the *real* torsion potential (which you can calculate via
high-level quantum mechanical calculations, for instance) is what has to be
accounted for by the torsion terms. Long story short: no chemical meaning
behind torsion terms.

> maximum energy should be 3. so the
> periodicity should be 3 rather than 2 or 1.
> Secondly, How does the program such as sander to use these
> parameters in the simulation? In amber force field , the torsion energy was
> defined as follow,
> ET=sum{Vn/(2*idiv)*(1+cos(|pn|*phi-phase)}
> Taking the CT-CT-CT-CT as an example, should the total torsional
> energy for this dihedral be
> E(torsion for CT-CT-CT-CT )
> = ( pk1*(1 + Cos[pn1*phi - phase1]) + pk2*(1 + Cos[pn2*phi - phase2]) +
> pk3*(1 + Cos[pn3*phi - phase3]) ).where pk1 = 0.20; pk2 = 0.25; pk3 = 0.18;
> phase1 = Pi; phase2 = Pi; phase3 = 0; pn1 = 1; pn2 = 2; pn3 = 3 ?

Yes. The torsion terms are simply a Fourier sum. Torsions are well-suited to
being fit with Fourier sums since they are periodic (clearly). As is always
true with any kind of expansion, if a function doesn't happen to be a single
term, then the more terms you include, the better fit you can get. That is
why many torsion parameters have multiple Fourier terms.

Hope this helps,

   Any suggestions will be greatly appreciated!
> Best regards,
> Xiangqian Kong
> Drug Design and Discovery Center,
> Shanghai Institute of Materia Medica,
> The Chinese Academy of Sciences,
> Shanghai 201203,P. R. China
> Email:
> _______________________________________________
> AMBER mailing list

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
AMBER mailing list
Received on Wed Dec 15 2010 - 07:00:04 PST
Custom Search