Re: Constraints protocol

From: Peter Freddolino (
Date: Mon Oct 13 2008 - 13:34:58 CDT

Hi Kamilla,
I would highly recommend doing some NPT equilibration in the presence of
restraints (that is, harmonic restraints, or what namd calls
constraints) on protein atoms, because most MD systems start out
undersolvated and thus have boxes that shrink somewhat, and you don't
want to have that stage alter the starting structure of your protein
significantly. If you equilibrate with very strong protein (>10 or so
kcal/mol A^2) restraints it is probably good to remove them gradually,
but yours are probably ok to remove in one step. It can't hurt, of
course, to remove your restraints gradually. Different people tend to
have different equilibration protocols; the key is to make sure that at
each step you truly do equilibrate the system (subject to the current
constraints placed on it), and as much as possible you avoid
perturbations of your solute during early stages of solvent
equilibration. Speaking of which, you should probably also add an
unconstrained NPT equilibration phase to your simulation prior to taking
data that you consider part of a production run, to allow initial
relaxation of your protein and any final adjustments to the periodic
cell size without including this data in your analysis.


Kamilla Kopec wrote:
> Hi,
> I am currently using the following "four step" protocol to generate MD
> trajectories in NAMD with the Charmm22 forcefield with CMAP corrrections:
> 1) Constrained minimization (constraint of 500 kcal/A2 on all protein
> atoms) 2000 steps
> 2) Unconstrained minimization 2000 steps
> 3) Constrained heat (0-300 K over 30ps) in NVT ENSEMBLE (constraint of
> 10 kcal/A2 on all protein atoms)
> 4) Production run in NPT ENSEMBLE no constraints
> If constraints are used during heating is it necessary to lift them
> gradually? Or is it OK to lift them all at once as I have done? If it
> is necessary to lift gradually - should this be done before or after
> the switch from NVT to NPT and over how long?
> Any suggestions for a suitable protocol would be greatly appreciated.
> Best regards
> Kamilla

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:58 CST