Re: [AMBER] High value for force constant is acceptable?

From: Karl Kirschner <>
Date: Fri, 19 Feb 2021 19:39:29 +0100

Hello Rinsha,

  I can briefly comment about the value of the force constant that we use
in performing torsion scans for optimizing force-field parameters. In these
projects, we only perform molecular mechanics (MM) minimization on a single
molecule in the gas phase, and thus cannot directly address the use of
large force constants in multi-molecule (e.g. protein-ligand)
condensed-phase MD simulations.

 To do an apples-to-apples comparison between the MM and quantum mechanics
calculations, we typically use a very large force constant (i.e.
significantly larger than 800.0 kcal/mol rad2). I believe this is necessary
since we are often optimizing conformations that are quite far from
stationary points on the molecule's potential energy surface (PES; e.g. a
high energy conformation found between a transition state and a minimum).
As a conformation approaches a stationary point, then I think the force
constants don't have to be so large, especially when the local PES is
shallow. If the MM minima reproduces the QM minima well, then these
conformations (i.e. those that reside at an energy minimum) shouldn't need
to be constrained at all when isolated (e.g. a single gas-phase molecule).

 Without knowing exactly what you are wanting to simulate, I can't say too
much more. If you are using the force constant to reproduce a transition
state in your simulations, then you might need a large force to keep it in
the desired conformation. However, if you think that the target
conformation is a minimum -- and you believe that the applied force field
models the molecule's PES accurately -- then I wouldn't think you need a
large force constant. Another condition that I can think of where you might
need a large force constant is when an induced-fit mechanism is occurring
and the given force field does not reproduce a very high energy
conformation well. Perhaps some insight can be gained through investigating
the target molecule using QM and seeing if the used force field reproduces
the QM data.

Hope that adds some food for thought.

On Thu, Feb 18, 2021 at 2:26 PM David A Case <> wrote:

> On Thu, Feb 11, 2021, Rinsha Chk wrote:
> >I am doing an MD simulation on azobenzene with dihedral angle restraint.
> >But it requires a force constant of 800.0 kcal/molrad2 to maintain the
> >desired dihedral angle value... My concern is that, is it acceptable to
> use
> >such a high value?
> That is very high. What is your evidence that such a large constant is
> needed? Could you live with some deviation of the desired torsion angle?
> I hope Amber users that regularly do torsion scans will chime in here.
> ....thx...dac
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Fri Feb 19 2021 - 11:00:02 PST
Custom Search