AMBER Archive (2007)

Subject: Re: AMBER: Water question

From: David Cerutti (dcerutti_at_mccammon.ucsd.edu)
Date: Sun Sep 23 2007 - 14:01:24 CDT


Hi Eddie,

    My personal favorite is SPC/E, but there are arguments why the TIP#P
models should be used with the AMBER force fields instead. The reason I
like SPC/E is that it is one of the few models (indeed, the only
non-polarizable point charge model) that reproduces the diffusivity and
dielectric constant while simultaneously obtaining good results for
density, heat of vaporization, heat capacity, and isothermal
compressibility. Another important thing to consider is the context in
which the model is to be used. The popular TIP3P model is NOT intended to
be used with Ewald summation, nor is the original TIP5P. Follow-ons to
these models are available, such as the extended TIP3P model "F" by
Price and Brooks (JCP 2004) are designed for use with Ewald sums,
but remarkably they still don't get the diffusion coefficient right
and the other properties are at best as good as SPC/E (the TIP3P-F frcmod
file will be available in the next release of AMBER but someone probably
has it already).
    So, that would be my suggestion--SPC/E or TIP3P-F if you're just using
ff03. TIP5P has extra points and so should only be used with a similar
forcefield (and even then you should use Stephen Rick's TIP5P-Ew, which
was modified for use with Ewald sums). Whatever model you choose, make
sure to read page 29 of the AMBER manual first--you must specify a frcmod
file to be sure that the water model you want really gets used in your
simulations. And, if this is all too complicated, don't worry most people
just use the plain old TIP3P by the AMBER default.

Dave

On Sun, 23 Sep 2007, Eddie Men wrote:

> Dear Amber users
>
> What is the 'best' water model (explicit) for a protein simulation?. I am
> using by the way the leaprc.ff03
> force field model (someone may want to suggest a different one?. Please do it
> before I start a million hour
> simulation and then have the reviewers reject my paper, lol).
> Is there any consensus as to what model should NOT be used?. Does some model
> have
> more impact in my calculations than other?. Which one is more accurate?.
>
> Thanks in advance
>
>
> Eddie
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu