AMBER Archive (2005)

Subject: Re: AMBER: generation of conformations for multi-conf. RESP fitting

From: FyD (fyd_at_u-picardie.fr)
Date: Tue Sep 06 2005 - 11:20:52 CDT

Quoting Kateryna Miroshnychenko <kateryna_mirosh_at_ire.kharkov.ua>:

> I want to perform a multi-conformational and multi-orientational
> RESP charge fitting for my ligand molecule consisting of 74 atoms.
> But it's difficult for me to decide:

RESP|ESP charges are affected by mol. orientation & conformation
-1- mol. conformation _greatly_ affect charge values
=> multi-conformation RESP|ESP fit has been used to make the charges more
general & of _high quality_ (which is suitable when MD is performed)
-2- mol. orientation also affect charge values (but _not_ as much as mol.
conformation)
=> Controlling mol. orientation (using the Rigid-Body Reorient. Algo.
implemented in R.E.D.) allows to get _reproducible_ charge values (+/-.0001)
independently of the QM software or initial structure used
=> Using multi-orientation RESP|ESP fit allows to average the charge diff.
observed for one mol. orientation over several orientations

> 1. how many conformations/orientations should I consider in order to
> obtain reproducible results?

I do not think it is a question of "number".

> 2. what is the best way to obtain such conformations? As far as I've
> understood each of these conformations must be not far from the
> minimum structure so that geometry optimization calculations would
> be not too long.

Pick up several (two or three) _stable_ (the most stable ones? with small dE ?)
conformations (not TS) _without_ H-Bond (leads to hyperpolarization). More
generally, I would follow "a" _rigorous_ strategy; whatever is "this"
strategy since many strategies (diff. in mol. orientation and/or mol.
conformation) are possible (being aware of the mol. orientation/mol.
conformation charge dependence and avoiding H-Bond). Then, if you can explain
what you wish using MD & the charges you got, "you won" since the
charges you developped take into account of many parameters (orient. & conf.).
Now, if you cannot explain what you wish by MD, I would try recomputing a new
set of charges (using different conformations and/or a new set of dihedral FF
parameters). In the later case, this is more easy since you know/are aware the
(orient)conformations you previously used...

You can find a lot of imformation about RESP charge derivation in the R.E.D.-I
FAQ @ http://www.u-picardie.fr/labo/lbpd/RED/FAQ-I.htm

Charge derivation is a difficult task, but also an important step in the
development a good force field since atomic charges are (with dihedral & vdW FF
parameters) primary determinants of mol. conformation & inter-molecular
interactions. This is why we developped R.E.D. which makes the RESP|ESP charge
derivation of high quality and high reproducible quite straighforward.

I hope this helps, regards, Francois 

-- 
 • F.-Y. Dupradeau •
DMAG EA 3901 & Faculte de Pharmacie, Amiens, France
       ••••
http://www.u-picardie.fr/labo/lbpd/FyD.htm
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu