Re: AMBER: RESP charge fitting for HFIP

From: FyD <fyd.u-picardie.fr>
Date: Wed, 20 Dec 2006 08:21:42 +0100

Quoting Mingfeng Yang <mfyang.gmail.com>:

> The problem now is the equivalencing. Usually, the methyl and methylene
> atoms are optimized in the second stage. But I remember that in the Amber
> mailing list, somebody mentioned Fluorine atoms should be equilvalenced in
> the first stage because of its strong electronegativity.
>
> In the input file generated automatically by RED II program, nothing is
> constrained in the first stage. In the 2nd stage, the charges of CHOH atoms
> are hold constant, and the 6 F atoms and the 2 C atoms in the two CF3 groups
> are equivalenced, respectively. I feel this procedure is a little weird, and
> guess the reverse order should be the correct choice.
>
> Another problem is that HFIP molecule can predominately adopt two
> conformations, in one of which, the H atom at the OH group can form weak
> intra-molecular hydrogen bond with F atom. Shall I include this conformation
> to fit the charge? According to people's experience, "Ligand coformers with
> intramolecular hydrogen bonds should not considered (this was mentioned in
> the original Cornell et al paper). RESP fitting of these conformers places
> far too much electronegativity on the hydrogen bonding heavy atoms." I
> tried to include and not include, and it seemed that the latter gave
> slightly better result, but both can not generate correct charge
> distribution to reproduce the liquid properties.
>
> The results from both trials and my intuition told me that if I use a
> Boltzman weight (~30% and ~70%), which approximately is the experimental
> ratio of the two conformations, to include the two conformations, the
> resulted charge will be perfect.
>
> Now if my charge fitting procedure has no significant problem. Then the
> choice for me is to manually fine tune the charges to reproduce the
> experimental thermodynamic parameters. But this is kind of against the
> philosophy behind amber force fields which always leave RESP charges alone,
> and tune up other parameters, such as torsions.

I think you summarized very well what is known so far, i. e.
- Using or not conformation(s) with H-Bond in RESP charge derivation,
- Using or not several conformations to get more general charge values,
- Using Bolzmann weight,
- Using or not several orientation(s) to get charge values less
dependent on molecular orientation,
- Using or not the default RESP charge derivation, i. e. where only
the charge values of the methyl & methylene groups are recomputed in
the 2nd RESP input.

Why do not you try to run MD simulations & see what you get ?

In R.E.D., RESP inputs are automatically generated. However, you CAN
FULLY control/modify the RESP charge input generation procedure. For
instance, if you do not want to follow the default RESP charge
derivation procedure, and equivalence the CF3 group in the 2nd RESP
input, below is what you could follow to modify the R.E.D. code; In
fact, not-following the default RESP charge derivation was one reason
of implementing this strategy in R.E.D. ;-)

Look for:
- the table '%Tatoms' line 302 in R.E.D.-II, & add the 'F' atom in the
corresponding list with its atomic number, i. e. something like: F =>
"9"
- the table '%Elements' line 1524 in R.E.D.-II, & add the 'FT' atom in
the corresponding list with its atomic number, i. e. something like:
FT => "9"

You should get what you wish with the following type of atom name
conventions in the initial PDB file:
C1 F2 F2 F2 C3 H3 O4 H4 C1 F2 F2 F2 : default RESP charge derivation
CT1 FT2 FT2 FT2 C3 H3 O4 H4 CT1 FT2 FT2 FT2: non-standard RESP charge
derivation.
You can obviously try many other possibilities...

I hope this helps, regards, Francois


-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Dec 24 2006 - 06:07:03 PST
Custom Search