AMBER Archive (2006)

Subject: Re: AMBER: System in polarizable waters

From: David A. Case (
Date: Mon Sep 11 2006 - 13:12:04 CDT

On Mon, Sep 11, 2006, Martin LEPSIK wrote:
> First, the order of atoms is different for TIP3PBOX and POL3BOX in
> solvents.lib. It is O H1 H2 for the former, while H1 O H2 for the
> latter. This results in two water types when adding POL3BOX to several
> explicit TIP3P waters. Should I change the order of atoms and/or residue
> name WAT-->PL3 of the explicit waters to solve it?

I'm assuming that you want all waters to be pol3(?) In that case, try the
commands like those you listed in your email before reading in the first pdb

> Second, I find it misleading having the default water model in
> leaprc.ff02 and .ff02EP TP3 instead of POL3 (although I know about
> simulations of polarizable solute in TIP3P water). In order to have
> POL3P waters with PL3 names, do I add the 3 lines (shown below) after
> loading leaprc, before loading pdb and adding POL3PBOX or differently or
> twice??
> WAT = PL3
> set WAT.1 name "PL3"
> loadamberparams frcmod.pol3

This sounds correct. I agree that the question of what the default water
model should be is a tricky one. We might add a new model in the next release
(.e.g "ff02r2" or something like that) that has the POL3 model as default.

As an aside: I would really recommend that you use leaprc.ff02.r1 in place of
the original ones. This provides updated backbone torsion parameters for
proteins, analogous to those needed for non-polarizable Amber potentials. If
you don't (yet) have Amber 9, you can download the parameters from (The "r1" revision is included in Amber
9, so no extra download is needed).

Also, we have been seeing problems (such as shake failures) with the "EP"
versions of this force field. Yong or Piotr can add comments here, but for me
the non-EP versions seem to be the better choice. [As I understand it, the
"r1" revisions were only tested for the non-EP version.]


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to