AMBER Archive (2009)

Subject: Re: [AMBER] Nonstandard molecules simulated with parm99EP ?

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Tue Jun 30 2009 - 02:31:53 CDT


Dear Marek,

> I just took a look on the article which you suggested below. Unfortunately
> the authors do not discuss much in detail why they prefared parm99 for
> their experimental/computational "alcohol story"
> . They just mention that GAFF provided unsatisfactory results in some
> cases (MeOH, t-BuOH) which probably comes mainly from too low
> value of the O-H force constant.

Yes

> On the other hand after choosing
> parm99, this force constant was anyway subject of their reparametristion
> (as well as in case of relevant charges) but probably already before
> this reparametrisation parm99 behaves a little better than GAFF for
> their molecular systems and relevant experiment data (FTIR spectra).
> Intuitively this could be clear since GAFF parametrisation is based
> on really wide set of suitable chosen molecules (200 if I am not wrong)
> to cover wide variety of molecular structures,
> so some inacuracy regarding to specific molecules is "penalty" for it's
> transferability/universality.

200 % agree with you ;-)

> So OK, this is one particular example, which describes that parm99
> could be for some class of non-nucl.acids/non-proteins molecules better
> regarding to
> specific experimental data than GAFF. Are there some another examples
> of comparison of parm99/parm99EP versus GAFF in cases of
> non-nuc.acids/non-proteins available
> (especially some related to free energy of binding calculations) ?

I do not know.

> Is there some general recommendation available to decide which
> forcefield (ff99, ff99EP, GAFF) could be more suitable for the given
> "nonstandard" molecule
> regarding to given physical property ? Probably no and one should make
> some short simulation tests.

Yes, do your own study.

> But in this context there is still in the game my original question:
>
> "How it is possible to force the antechamber to use some different
> forcefield from GAFF ( like for example parm99EP
> for parametrisation of the relevant residui (generating of
> PREPIN/FRCMOD files) ?"

Here the Antechamber developers should answer.

My understanding is that the number of force field (FF) atom types in
GAFF is superior than these in parm99. Consequently, this might
generate problems when you want to generate FF atom type
automatically; I mean for parm99, you will get "blanks"...

> One possible way, although not much comfortable is of course to let the
> antechamber generate PREPIN's/FRCMOD's with
> GAFF forcefield atom types and then manually change this atomtypes to
> the corresponding atomtypes from
> desired forcefield (parm99/parm99EP) but as I mentioned before it is
> really not the most comfortable way
> which of course provide some chance to do mistakes on this level. So I
> hope there is some easier (automatic) way how to do this job or not ?

This is the reason why the FF libraries in R.E.DD.B. are FF atom types
independent; i.e. FF atom types are always added using a LEaP script.

See for instance:
http://q4md-forcefieldtools.org/REDDB/projects/W-46/script1.ff

Personnally, I always use parm99 FF atom types for organic
molecules/protein/nucleic acid and add FF atom types manually; pick up
only _classical_ force field parameters from GAFF (bond & angle force
constants, aromatic dihedrals) and adapt them to parm99 way. Get
equilibrium values from high ab initio geometry optimization or X-ray
structures (not from GAFF). Key dihedrals are always refitted (not
obtained from GAFF).

What you said above summarize exactly the problem "so some inacuracy
regarding to specific molecules is "penalty" for it's
transferability/universality". When you use GAFF automatically all is
fine if your molecule has been well parametrized but you do not know
if it was indeed well parametrized...

Just my personal opinion: Others might have other experiences.

regards, Francois

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber