Re: [AMBER] Partial charge discrepancy between AM1-BCC small molecule charge and FF library charges

From: Shiji Zhao via AMBER <>
Date: Wed, 27 Jul 2022 14:24:43 -0700

Hi Liao,

The charges for amino acids in the ff14SB force field were derived using a
method called RESP (REstrained Electrostatic Potential), which fits charges
to reproduce QM calculated electrostatic potentials (ESPs) around
molecules. In contrast, AM1-BCC is an empirical method that applies one set
of parameters into any molecule. In theory, RESP should give better charges
than AM1-BCC since it takes the molecular details of target molecules into
account. However, many studies have shown that AM1-BCC derived charges give
ideal performances in molecular modeling tasks such as MD simulation. For
example, this one

My advice is that, if you don't have easy access to QM software to
calculate QM ESPs, you can simply use the AM1-BCC method for charging. If
you have QM softwares such as Gaussian or GAMESS, you can use the RESP
method (There are two programs in the AmberTools package that can perform
RESP charging: *resp* and **), or you can compare the
performances of both methods.

Shiji Zhao

On Wed, Jul 27, 2022 at 1:45 PM Liao via AMBER <> wrote:

> Dear experts,
> Today when looking at partial charges on my molecules, I found quite a big
> difference, between the side chain N on Lysine, and an alkyl NH3+ on a
> small molecule. The small molecule has its partial charges calculated by
> sqm in UCSF chimera for AM1-BCC charges, and the Lysine charges are from
> ff14SB library.
> I drew the standalone lysine and added charges to it, and the big
> difference still stands, that the side chain N on Lysine is +0.374 (in NH3+
> state), and the newly calculated lysine treated as a small molecule is
> +0.851. Changing the small molecule lysine from NH3+ to NH2 the
> non-protonated state didn't make a significant difference, that the
> recalculated charge after minimization is +0.909.
> I remember from reading, and from my own experience, that good charges are
> one of foremost important factors in modeling. So I'm quite surprised by
> this difference, that AM1-BCC charges are known to be good, and so are well
> established FF charges, but they differ so much in the case of NH3+.
> Which one should I prefer?
> Thanks!
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Thu Aug 04 2022 - 13:37:31 PDT
Custom Search