RE: AMBER: ff03 and ff99SB force fields

From: Yong Duan <>
Date: Fri, 4 Aug 2006 13:58:51 +0200

Hi Vlad:

I don't believe it is necessary to use the ff03 charge model for ligands
because the charges are indeed reasonably compatible with HF/6-31G* charges
(to which the am1-BCC charges try to mimick). In the case of amino acids,
since (torsion) parameter tuning has been targeted to specific charge set,
mixing two charge models can have unexpected results. This is particularly
important if you consider those short-range electrostatic (e.g., 1-4) terms
where the specific charge distribution (rather than the overall property of
the amino acid) plays quite important roles. In my opinion, this is probably
the most important reason for not mixing two charge models.

As for non-bonded interactions in ligand binding, we are actually very
liberal. We sometimes use ff03 charge models, sometimes HF/6-31G* models,
and, yes, sometimes am1-BCC models. Although a detailed study is needed to
compare these models, I am optimistic that they are reasonably compatible
and the difference is mainly at the level of accuracy. In the end, I
personally like the ff03 charge models for obvious (biased) reasons :). On
the other hand, HF/6-31G* model has been used widely and good results have
been obtained. However, it is probably unwise to treat HF/6-31G* model as
some sort of gold standard. Afterall, the reason that it works is really
because of the (lack) of accuracy in the underlying QM electrostatic field
obtained from HF/6-31G* method which can hardly be a solid reason for such
notion that we all should compare other models against it.

Of course, as you can see, I am very biased :). Hope this helps.


> -----Original Message-----
> From:
> [] On Behalf Of Vlad Cojocaru
> Sent: Friday, August 04, 2006 10:18 AM
> To:
> Subject: Re: AMBER: ff03 and ff99SB force fields
> Dear Yong,
> Thanks a lot for following up on this ... It is very useful to clarify
> all my questions regarding this issue. I have another small
> question. I
> have been doing some simulations of protein-ligand (small molecule)
> complex and use ff03 for the protein and HF/6-31g* charges for the
> ligang (derived with RED II - GAMESS). From your experience, would you
> think that even for small molecules one should be strict and derive
> charges according to the ff03 model? ...
> And just a short follow up on this, can one use the
> antechamber derived
> am1-bcc charges for small ligands in combination with ff03
> for proteins?
> Thanks again
> Best wishes
> vlad

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Aug 06 2006 - 06:07:14 PDT
Custom Search