[gmx-users] on non-equilibrium MD

Justin Lemkul jalemkul at vt.edu
Sat Apr 2 19:08:30 CEST 2016

On 4/1/16 10:51 PM, Brett wrote:
> Thanks Justin.
> Based on your answer, here I would like to ask,
> Suppose there is a PDB for a protein-ligand, and it is already known ligand
> would lead extremely significant conformation change of that specific
> protein. Based on the protein-ligand PDB, I would like to investigate the
> conformation of the protein without ligand by MD.
> First from the PDB for  the protein-ligand, I delete the ligand part and get
> the PDB for the protein part. Then with the protein part PDB and based on
> your lyzoyme tutorial protocol, through energy minimization, NVT and NPT
> equilibration, I start the production MD.
> Here my question is, as for the energy of the protein in the ligand binding
> and ligand free state is significantly different, and based on my practice
> based on your lysozyme tutorial, the energy minimization, NVT and NPT
> equilibration steps could almost lead to no conformation change of the
> protein from the initial PDB conformation in the protein-ligand state, then

As is their purpose.

> in the production MD process, there should be a significant conformation
> change, which I regard that the conformation change can be checked by the
> energy change by command "gmx energy -f md_01.edr -o potential.xvg". If the

The potential energy will be dominated by water, and the potential energy of the 
protein itself is a force field-dependent, nonphysical quantity.  So no, this 
doesn't give you anything useful.

> protein conformation in the ligand binding status has much higher energy than
> in the ligand free state, then will you please tell me why the conformation
> change can only be observed in the production md process, but not in the
> initial energy minimization step before the production MD?

There is no thermal energy in EM, so energy barriers are not crossed.  The 
structural changes are all downhill in local minima.  So to observe 
conformational changes in an apo protein will require extensive (and multiple) 
simulations to observe, perhaps on the microsecond scale or longer.

You can't simply skip equilibration because you know the structure will change. 
  You still need to equilibrate the solvent around the initial state of the 
protein; if you don't, you'll get spurious forces that (if the simulation 
doesn't immediately collapse) will lead to potentially artificial structure changes.



Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441


More information about the gromacs.org_gmx-users mailing list