AMBER Archive (2008)

Subject: Re: AMBER: Antechamber/mopac am1-bcc charge

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Sat Jun 28 2008 - 02:47:54 CDT

Quoting Francesco Pietra <chiendarret_at_gmail.com>:

> OK, simplicity might br illusory in this area, as I also said.

Yes.

I think you could read some original papers from Kollman & coworkers.
For instance, the paper corresponding to parm96.dat & parm98.dat. This
is clearly explain that the strength of AMBER force fields is THE
charge model, i.e. RESP and not Mulliken (at that time am1-bcc was not
developped; I personally avoid am1-bcc because this is a black box).

BEHIND the word "RESP" is not just picking a conformation and applying
a 2 stages fitting approach. On the contrary, a well defined strategy
is defined:
* rigorously define the conformation(s) used with well defined rules
   - small difference in energy between conformations, avoiding
conformations with hydrogen bond(s), etc ... -
* rigorously define the basis set used
* rigorously define the 2 fitting approach
This means a user has to pay the price (computer simulation time, time
spent) to study the system (before MD simulation).

This is the strength of the parm94, 96, 98 and 99 AMBER and GAFF force
fields. Then, more complex strategies were applied such as in ff03 FF
and Glycam FF, but still based on a RESP fitting approach.

> One is
> not prevented by current theories to try to be rigorous, however. You
> know the work by your co-nationals on tetracycline (JACS 2008, 130,
> 114). How many others are willing (or allowed) to invest so much time
> for every new family of "unusual" compounds they want to investigate?

Well my personal opinion this is a normal work if one wishes to
produce interesting papers.

> Personally, I see the danger to be simplistic, but I am tankful for
> GAFF.

GAFF is based on the RESP charge model.

> More than that, what I would like to see is a comparison of the
> said work on tetracycline carried out with GAFF. What about if that
> turns out in a substantial agreement of the results (and, of course,
> agreement with experiments, if they exist or can be done)?
>
> If not, unless it is an oddness for such a class of compounds. we
> should be suspicious about any simulation carried out for "unusual"
> compounds with simplified FF. Not a good perspective, also because
> packages for electronic calculation are not well devised to help
> setting up a fitting FF. They can do that, but along tricky work.
>
> Well, I feel that existing literature comparing in a clear,
> unequivocal, practical way "rigorous" with "simplified" approaches
> should be put at the forefront. This is a wish by the common user, as
> I am.

Well, every user can learn, little by little. The AMBER mailing list
helps a lot.

More a user will spend time to prepare her/his FF (FF parameters;
charge values - before MD simulation), more she/he will control/be
able to understand her/his results after MD simulation.

regards, Francois

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu