AMBER Archive (2007)

Subject: Re: AMBER: RESP charge fitting using antechamber

From: FyD (fyd_at_q4md-forcefieldtools.org)
Date: Mon Dec 31 2007 - 09:40:11 CST


Quoting Francesco Pietra <chiendarret_at_yahoo.com>:

> What about approaching RED with NWChem? In my experience NWChem affords
> reliable RESP charges and, on other basis, in particular plane waves in the
> latest versions, I never considered to install Gamess or buy Gaussian.

Please, see the R.E.D. web site & look for News about the R.E.D. IV
version status: "R.E.D.-IV news".
Or look directly @ http://q4md-forcefieldtools.org/RED/news.php
As you can see, we are working on this new feature ;-)

This topic was already discussed in the AMBER mailing list.
A quick google search with "NWChem RED" gives many links:
http://structbio.vanderbilt.edu/archives/amber-archive/2007/0211.php
http://structbio.vanderbilt.edu/archives/amber-archive/2007/0213.php
etc...

That being said, interfacing NWChem/RESP with R.E.D. should not be
that difficult. However, we are also interested in finding keywords
allowing to get identical charge values with
GAMESS-US/PC-GAMESS/Gaussian-XX/NWChem (& if not possible describe the
differences obtained).

For instance it is possible to generate RESP charges with NWChem or
Jaguar (I mean without the RESP program), but from what I
heard/tested/read the results are quite different from regular RESP
charges generated with a GAMESS/RESP or GAUSSIAN/RESP system. Thus,
characterizing the problems is at least as important as just
interfacing "a" new program.

I would like to emphasize here that originally R.E.D. was developed to
get highly effective & reproducible RESP or ESP charge values
independently of the QM program & initial structures. We work to make
results obtained with GAMESS-US & Gaussian identical (+/-0.0001 e)
(See the R.E.D.-II manual page 16 where the tricky cases are fully
characterized). This is particularly true for metal compound with Fe,
Zn or Cu (among many others). We originally selected Gaussian
(proprietary & quite expensive) as used by the AMBER developers, and
GAMESS as academic & provided at no cost. Recently, PC-GAMESS was
interfaced because it is fast/powerful/fully compatible with Linux.

Now, you are right interfacing NWChem is quite interesting for many
reasons... This feature will be available in R.E.D.-IV & it might even
take a central role in the next R.E.D. Server.

Until the release of R.E.D. IV and/or R.E.D. Server, you might
consider using PC-GAMESS as it is simple to use & fast.

regards, Francois

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