[gmx-users] charge assign

[Justin Lemkul]

On 7/2/16 2:37 AM, elham tazikeh wrote:
> Dear Justin
> i really appreciate for your reply
> i think my question wasn't clear. i prefer to express my problem again:
>
> i'd like to accomplish protein-drug simulation by gromos force field and
> before, docked my ligand on the protein (because,there wasn't the
> crystallography file of this complex).
> Now, to reach my destination, i eliminated ligand.pdb file from the end of
> the complex.pdb and determined the charges of atoms by gaussian software by
> mulicken method  (without optimization ), but the numbers of atoms in this
> file and the files come from PRODRG server (the gromaces 87/gromacs
> coordinate file .polar /aromatic hydrogens (*.gro) and the gromaces
> topology (*.itp)) weren't the same.
>  for this reason, i pasted the section of complex.pdb as lig.pdb in PRODRG
> and achieved all formats of my ligand.
> next, i saved the pdb file (polar/aromatic hydrogens) as ligand.pdb and
> then give it to gaussian software to determination the charge of atoms.
>
> Is it correct?
> or
> i should adding hydrogens to my ligand by another way?
>

Providing Gaussian with a UA coordinate file isn't going to lead to a useful result.

The connection between GROMOS and QM is very limited, and as I understand from 
talking with one of their developers, the initial charge distribution is 
obtained from a simple DFT calculation of the electrostatic surface potential. 
Charges are then assigned empirically (not from Gaussian or any other QM 
software) until they reasonably reproduce the ESP, then again empirically 
refined against condensed-phase properties (because an in vacuo charge 
distribution is not the same as the one in solution due to solvent polarization 
effects).

In general, piece together your ligand from existing building blocks; this is 
commonly done in most additive force fields.  Anything that is unknown (weird 
chemical moiety, etc) should be parametrized itself as a suitable model compound 
(not the whole ligand, because probably most of it is known from existing 
groups).  Those unknown parts are treated via the QM approach above, but also 
require empirical data to target.  This is why it is sometimes quite difficult 
to parametrize molecules for GROMOS; it requires significant empirical effort 
despite the apparent simplicity of the UA representation.

Or try ATB (it's far superior to PRODRG, but never trust a black box 
implicitly), or use a different force field that has a much stronger and more 
straightforward connection to QM (CHARMM, AMBER, OPLS - all of which have 
published parametrization protocols and/or web servers that do most of the work 
for you).

-Justin

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

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
http://mackerell.umaryland.edu/~jalemkul

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