Dear Albert,
First, R.E.D. does not only generate RESP charges but MEP-based
charges; thus the number of charge models handled by R.E.D. is far
broader than only RESP charges. See the list of charge models (you)
reported on the q4md-forcefieldtools web site.
Then, the RESP charge model bears the Amber stamp; I am not sure that
using (i) this charge model and (ii) directly in CHARMM is a good
idea: for sure politically, if I dare to say that, this is might be a
little dangerous if one wants a paper to be accepted by the CHARMM
community ;-)
But, MEP-based charges can be directly used after validation in MD
simulations or involved in some refinement approaches.
More generally, did you do a literature search about all types of ESP
charges and CHARMM force fields? or even ESP/RESP charges and OLPLS? I
am sure you will find ideas: but for sure one should be all but naive
here...
If you use a MEP based charge model in CHARMM following a well defined
reference and if you demonstrate that this charge model correctly
reproduces some experimental parameters, I think this should be fine.
You might be interested in:
C. Jenson & W. L. Jorgensen, J. Am. Chem. Soc., 1997, 119, 10846–10854.
R. H. Henchman & J. W. Essex, J. Comput. Chem., 1999, 20, 483–498.
E. Mayaan, A. Moser, A. D. MacKerell Jr. & D. M. York J. Comput.
Chem., 2007, 28, 495–507.
J. L. Knight & C. L. Brooks III, J. Chem. Theory Comput. 2009, 5, 1680–1691.
I am sure you can find many other interesting papers...
I am sure Amber developers have their own opinions here.
regards, Francois
> Recently I found the following statement from CHARMM website
> (http://dogmans.umaryland.edu/~kenno/cgenff/faq.html#esp):
>
>
> /I obtained Mulliken, NPA, AIM, Hirschfeld (aka stockholder) or
> Merz-Kollman (or CHELP or other (R)ESP fitted) charges. Can I use
> them as they are?/
>
> /No! Charges that are directly derived from the wavefunction (such as
> Mulliken, NPA, AIM, Hirschfeld,...) are generally considered incapable
> of reproducing accurate electrostatic interaction energies in the
> framework of an atom-centered point charge model (as used by Class I
> additive force fields such as CHARMM) without emiprical corrections such
> as the "BCC" scheme. Force fields that do use unmodified charges based
> on vacuum QM calculations typically use an indirect method, fitting the
> atomic charges to reproduce the ElectroStatic Potential (ESP) at many
> points around the molecule. However, //*CHARMM falls into neither of
> these categories*//. For charges to be compatible with the L-J (and
> dihedral) parameters in CHARMM, they must be optimized based on
> interactions with water, as described in the CGenFF paper and tutorials.
> The Merz-Kollman charges mentioned in said paper and in the old tutorial
> are //*only*//meant as an easy way to obtain an //*initial guess*//, and
> have been superseded in this role by the CGenFF charges produced by
> //the CGenFF program
> <http://dogmans.umaryland.edu/%7Ekenno/cgenff/program.html>//. Same for
> other (R)ESP methods and AM1-BCC and the likes; they may be appropriate
> for other force fields, but not for CHARMM. Different force fields are
> based on different parameterization philosophies, and all parameters are
> interconnected and have to be consistent. If you don't have the time or
> resources to perform charge optimization based on interactions with
> water, it is strongly recommended to assign charges by analogy to
> existing CGenFF compounds, which were properly optimized. In fact, //the
> CGenFF program
> <http://dogmans.umaryland.edu/%7Ekenno/cgenff/program.html>//does this
> in an automated fashion, and is presently the recommended source of
> initial guess charges (and parameters); see //this related FAQ entry
> <http://dogmans.umaryland.edu/%7Ekenno/cgenff/faq.html#procedure>//. /
>
>
>
> It seems that we cannot use RESP charge for CHARMM FF? However, when we
> read the paramter comments for RED-VIII.5
> (http://q4md-forcefieldtools.org/REDS/popup/popkeyword.php), we can find
> the following:
>
> / -6- *ESP-A1*
> HF/6-31G*//HF/6-31G* - Connolly surface algo. used in MEP computation -
> 1 stage ESP fit qwt=0.0
> Used in some AMBER, OPLS & CHARMM force field based simulations //
> //
> //// -7- //*ESP-C1*//
> //HF/6-31G*//HF/6-31G* - CHELPG algo. used in MEP computation - 1 stage
> ESP fit qwt=0.0/ /
> //Used in some OPLS & CHARMM force field based simulations/
>
> So, Iam quite confused for such different comments. Does anybody have
> any idea for this?
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 29 2012 - 13:30:02 PST
