Re: [AMBER] General question on applying restraints during production MD

From: Jason Swails <>
Date: Tue, 26 Jan 2016 07:57:58 -0500

On Tue, Jan 26, 2016 at 5:16 AM, George Tzotzos <> wrote:

> My system consists of a homodimer receptor and two identical ligands, one
> in each subunit of the receptor. The receptor also consists of one water
> molecule per subunit. The latter are assumed to be conserved. The system
> was minimised, heated and equilibrated applying constraints on the complex
> (restraint_wt = 2.0). The receptor waters form H-bonds with receptor
> residues and the ligand and are in proximity of the dimer interface.
> I’m currently conducting a 100 ns production MD in which the restraints
> have been removed. My goal is to determine whether the ligands have an
> effect in the formation of the homodimer interface, so I want to have a
> reasonable estimate of the relative binding of the two subunits in the
> presence and absence of the ligands.
> The water of one of the subunits dissociates from the receptor after ~25
> ns. The 2nd water molecule did not dissociate from the receptor.
> My question is whether it is legitimate to repeat the production MD
> maintaining the constraints on the water molecule but not on the rest of
> the system.

​Define "legitimate". You are biasing your simulation and introducing an
assumption into the model. Therefore, this reduces the potential
information that the model can provide. As an obvious example, you clearly
cannot draw any conclusions about the behavior of the restrained water
molecules (e.g., you cannot claim any evidence of their being conserved,
since you forced them to be so).

​his does not necessarily invalidate your entire study, but if the behavior
of the water molecules is critical to understanding the behavior of your
system, introducing this kind of restraint will limit the strength of the
conclusions you can draw from your model. To what extent, and whether it
is acceptable, are important questions.

One other thing I'll point out is that "conserved water" does not mean that
those waters don't exchange with bulk solvent. If you are seeing those
water molecules exchange, I would argue that is normal behavior and you
should just leave it alone. After all, the free energy change associated
with swapping any two water molecules exactly is, by definition, zero. A
water molecule that is not permitted to exchange has a higher free energy
than one that is (since it has a much smaller entropy).


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Jan 26 2016 - 05:00:05 PST
Custom Search