[gmx-users] Docking validation

[Justin Lemkul]

On 3/9/14, 5:04 AM, Mostafa Javaheri wrote:
> Dear Justin
> I docked a ligand to a hetrodimer protein, I know the approximate
> position of the ligand but I don't know whether or not the docked
> ligand is relaxed in the binding site or is it taken the proper
> orientation? I run the ligand-protein complex for 5ns MD by using
> charmm force field and applied position restraint and constraint
> forall bonds (ligand parameters obtained from swissparam server).
> Should I apply restraint and constraint? Or I have to remove them for
> better flexibility if its yes how should I remove them? And will the

Constraints are fine and the CHARMM force field should be used with "constraints 
= h-bonds."  Restraints, on the other hand, are probably working against you. 
Restraining a protein during initial equilibration is done to prevent distortion 
due to unequilibrated solvent, but beyond that restraints are not normally 
applied for simple MD simulations of proteins in water.  In this case, I suspect 
they are a barrier against refining the position of your ligand.

> simulation be correct if I remove them? Could I mark some residue of
> binding site to force them to interact with the ligand? How?

Possibly the pull code, or the intermolecular distance restraints that are being 
tested within the development code.

> To understand whether simulation could or not predict the correct
> position of a ligand, I docked another ligand to its binding site
> which their crystallographic structure had been provided before, I
> used the worst docking pose for performing the same MDS, at last I
> compared the positions of two ligands (crystallographic and
> simulated), simulation hardly changed the position of the ligand after
> 5ns although its position energetically was wrong (based on docking
> and crystallographic pdb file). If MDS continue for several µs will
> the ligand find the correct position and orientation?  Given that I

I would not jump to the assumption that microseconds of MD would be useful here. 
  5 ns is a short simulation, you can always extend it a little bit more.  But 
do realize that a poor docking pose might not be refined by MD; if it is in a 
local minimum from which the energetic fluctuations in MD cannot displace it, it 
won't do you any good to try to brute force it.

> couldn't meet such a MDS how can I speed up this simulation? Will it
> be useful to mark some residues for interaction or put the ligand near
> the binding site by dragging or do a flexible docking?

If you have the ability to do flexible docking, do it.  I understand the need 
for faster rigid docking when doing high-throughput work, but if there are 
problems with the output of rigid docking, clearly the assumption of rigidity is 
wrong.

-Justin

-- 
==================================================

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

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================