Re: [AMBER] gaff-parametrisation - how many conformation for a flexible molecule

From: FyD <>
Date: Wed, 17 Nov 2010 09:18:27 +0100

Dear Thomas,

> i want to parametrise a not so small molecule (300 around half of
> them are hydrogens). the really flexible part of that molecule
> consits of a ring structure, which is build from 4 benzene molecules
> (B) and 4 alkane-chains (C) (14 carbon-units):
> B-C-B-C
> | |
> C-B-C-B
> so my question now is, how many different conformations one should
> use to derive the resp-charges?

For nucleic acid & proteins, one or two canonical conformations were selected.
See Cieplak et al. J. Comput. Chem. 16: 1357-1377, 1995.
     Duan et al. J. Comput. Chem. 24: 1999?2012, 2003.

For sugars, the Glycam developers used another approach where the
conformations involved in charge derivation are obtained from a MD
trajectory. This represents an efficient way to affect the charge fit
based on the conformations the most observed in MD simulations.
See Kirschner et al. J. Comput. Chem. 29: 622?655, 2008.

One can also use conformation(s) deduced from experimental data... or
the lowest minimum observed after geometry optimization or a set of
minima having small dE values versus the lowest minimum.

Another approach is to split your "big" molecule/polymer into
elementary building blocks where the conformation(s) of each
elementary building block is fully characterized. In fact the idea is
exactly the same than for a biopolymer such as a protein or a nucleic
acid. In your case, you build two connected fragments "-B-C-B-C-...-"
involving two small molecules defined as "a-B-b" and "c-C-d" and using
charge constraints during the fit so that the total charge of "B-C" is
generally an integer. The advantages of this approach is that you can
parametrize potentially any type of "big" molecule/polymer having more
or less heterogeneous chemical groups. The main disadvantage is that
this approach might not be that easy (when one is a new user) to
figure out what should be the "a", "b", "c" and "d" capping groups. To
follow this approach you can use R.E.D. or even R.E.D. Server; You
will find examples in R.E.DD.B. and tutorials as well. See

I hope this helps.

regards, Francois

AMBER mailing list
Received on Wed Nov 17 2010 - 00:30:03 PST
Custom Search