RE: Lone Pairs.

From: Bayly, Christopher <christopher_bayly_at_merck.com>
Date: Thu 08 Feb 2001 11:39:39 -0500

Where and if lone pairs are needed has returned as an area of active
discussion in the FF business. I would grossly summarize the arguments as:
1. General agreement that in principle they should be there (i.e. it is
better theory).
2. Disagreement on whether _in practice_ they make enough of a difference to
make it worth the
   extra lone pair centers, calculation, and parameterization.
3. Disagreement on where to place them if you want them (i.e. how far away
from the nucleus).
4. Concern about the extra parameterization pain: how you treat torsions
involving the lone pair, should lp's have a tiny vdW radius (if
so=non-physical; if not you risk strong positive charges getting pulled into
them = singularity).

The next-generation polarizable forcefields tend to have lone pairs. For
the two-body additive FFs like Cornell et al., the issue is addressed
directly in a paper by Dixon and Kollman (J. Comp Chem 1997, 18, p1632),
with the conclusion being you can't get in-plane and out-of-plane H-bonding
energies, and free energy of solvation, correct for pyridine unless you have
them. Our experience with AM1-BCC charges is that lone pairs would improve
things in general for heteroaromatic rings.

Fred: The Dixon and Kollman work is directly relevent to your histidine-Cu
bond situation, especially vis-a-vis relative energies of in-plane and
out-of-plane coordination to the metal. Overall, my $0.02 is that you may
want to put them on a la Dixon and Kollman, but caveat emptor re point 4
above.

Christopher Bayly


> ----------
> From: Fred P. Arnold[SMTP:fparnold_at_chem.nwu.edu]
> Sent: Wednesday, February 07, 2001 5:19 PM
> To: amber_at_cgl.ucsf.edu
> Subject: Lone Pairs.
>
> Hello,
>
> Pardon if this is a frequent and silly question, but I am new to Amber.l
>
> In the parm94.dat and later files, I see that the Lone Pair type from the
> earlier parm91.dat has disappeared. I would like to model a
> donor-acceptor type interaction (LP on histidine to Cu) without explicitly
> putting in the bond. Will the current FF definition with the RESP charges
> be sufficient, or should I refit presuming a dummy atom at some position
> approximating where an LP would be? I read the '95 Cornell paper, but
> didn't find a clear answer. My inorganic roots say that I should have a
> locus of charge displaced from the N nucleus explicitly present, in order
> to achieve the correct directionality.
>
> Thanks for your time.
>
> =Fred Arnold
>
> Frederick P. Arnold, Jr.
> NUIT, Northwestern U.
> f-arnold_at_northwestern.edu
>
>
Received on Thu Feb 08 2001 - 08:39:39 PST
Custom Search