AMBER Archive (2005)

Subject: Re: AMBER: RESP: Question

From: FyD (fyd_at_u-picardie.fr)
Date: Mon Dec 05 2005 - 15:31:32 CST

Quoting Carlos Jaime <carlos.jaime_at_uab.es>:

> Our group is using Amber since long ago. Nowadays we are sistematically
> using Amber v.7, and one of my Ph.D student face with one problem with
> RESP that I want to expose.
> We are using RESP to get a set of atomic charges that are 'independent
> from the conformation' of the substrate. We usually compute several
> conformations with an ab initio program (usually Gaussian), get the
> structure and energies, and we use the structure, the energy and the
> atomic charges into RESP, with the multiconformer option, to get an
> unique set of atomic charges.
> When working with fluorine compunds, we realized that averaging two
> conformers, the fluorine atoms got a charge almost zero.

We computed RESP charges for TFE:
-1) conf-1
-2) conf-2
-3) conf-1 + conf-2
The charges are highly similar in the three sets of charges i.e. ~ -.18

> We checked the
> behaviour of the RESP program with individual conformations and is
> working properly, but as soon as we are using more than one conformer,
> the charges are greatly reduced and several values are finally near zero.

This seems quite strange... We could imagine some tricks in your case (presence
of a HBond in one conformation and not in another one). But I would 1st check
the RESP inputs...

> We were really surprised and we started to search through the RESP code.
> We found thet RESP do not average charges but average the molecular
> electrostatic potentials.

Well "fitting" is not "averaging": It is normal. However, from our tests when
the conformations are _well_ selected, the resulting charges make sense.

> This means that the average of the meps of two
> conformers that are enantiomeric is almost null, and atomic charges are
> then also almost zero.

"conformers that are enantiomeric": Here I do not understand you.
Two 'real' enantiomers are only enantiomers and not conformers...
Now, you can compare two conformations of two enantiomers, but this is
different.

We never fit two (real) enantiomers together; very interesting ;-). I would not
be surprised to see some funny effects...

> Finally we end up with two possibilities:
> 1.- we are using RESP in a wrong way
> 2.- something in RESP is not properly working

I would strongly suggest you to use R.E.D. version II for this. You would be
sure if your way of using RESP is correct. In particular, see the manual
section VII "NEW TUTORIAL FOR R.E.D. II: Three conformation-two orientation
RESP fit (Alanine dipeptide)".

And then proceed as follows:
- fit each conformation independently & look at the RRMS
- fit the conformations to gether & look at the RRMS
How are the diff. in RRMS ?

> Has anyone faced with this problem before?
> Can someone give us some information about this point?

We will release soon (I hope) an important set of charges for many
molecules: This should help you. Contact me personnally if you want to have
access to this data in advance.

Regards, Francois

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu