AMBER Archive (2003)

Subject: Re: AMBER: Basic potential question:

From: FyD (fyd_at_u-picardie.fr)
Date: Wed Dec 24 2003 - 12:35:22 CST


Quoting Wayne Dawson <dawson_at_nih.go.jp>:

> > can anybody tell me what are the potentials cff91,cvff,esff good
> > for?(if they have some specific purpose) are they good RNA+proteins?
>
> In all humility and due respect, they were certainly not good for
> RNA. I think the Amber option is a far better choice if you are using
> Discovery. If I remotely recall correctly in the Discover manual, I
> think they say these potentials were mainly optimized for calculating
> proteins, so _perhaps_ they work ok for most of the usual proteins
> you'll find in PDB. In my own case, I did not find that they worked
> all that well, but that may because of the type of proteins I had to
> deal with.
> The main problem in general with Discover is that it is largely a
> "black box" to the user and therefore rather difficult to rip open and
> see what is really inside --- particularly when you wonder why
> something doesn't make much sense in a calculation.

I totally agree with Wayne Dawson... We worked with Discover/Discover_3 for more
than 5 years and we even converted AMBER and GLYCAM FF for Discover_3 (see
http://www.u-picardie.fr/labo/lbpd/AmberFFC/). Our conlusion: we switched to
AMBER... The fact that Discover is a black box limits you a looot... Moreover,
if a bug is discovered in such product you just have to pray that the company
provide you the patch... (See Bugs in Computational Chemistry Software and their
Consequences: The Importance of the Source Code,
J. Mol. Model., 2003, 9, 271-272)

Best regards, Francois

F.-Y. Dupradeau
 --
The Scripps Research Institute, San Diego, CA
Faculte de Pharmacie, UPJV, Amiens, France
 --
http://www.u-picardie.fr/labo/lbpd/fyd.htm

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu