Re: Solvent model other than TIP3 for use with CHARMM FF in NAMD??

From: Marcos Sotomayor (
Date: Sun Jun 26 2005 - 14:04:28 CDT


Its been very interesting to read this discussion about water models...
some additional comments:

- As Blake and Aaron mentioned before, the CHARMM force field
has been optimized to work with TIP3 and therefore working with other
water models (F3C, TIP4, TIP5, SPC, etc) seems to be inadequate unless
enough testing is done. TIP3P is NOT "just the most popular one to go
with CHARMM". Note that NAMD uses flexible water molecules, and
that not only the degrees of freedom employed for the protein and the
water molecules should be the same, but also the potential form has to be
the same, as mentioned in Levitt's paper. It seems that the
F3C model uses a potential form slightly different to that used by CHARMM
(switching functions different?). Also, Brian's question "Which protein
force field was used?" to parameterize tip3, tip4, tip5 seems to be
"ill-defined", since, as far as I know, those water models were developed
to reproduce properties of liquid water first. Then CHARMM developers used
tip3 for their parameterization. Although these water models are far from
perfect, one should try to be consistent and use the appropriate model for
the corresponding protein force field.

- I wonder if the reason of the stability of the protein in Leonardo's
and Adrian's NVE simulations is due to the water density used. Note that
at such temperatures water is probably vapor! It also seems *risky*
to use a protein force field and water force field that are not
temperature dependent. In particular, TIP3 was designed to reproduce
properties of *liquid* water, not ice or vapor! It would be interesting to
test the properties of those models at such high temperatures.

- Finally, within NAMD you can assign different temperature coupling
constants to different atoms, and thus to solvent and solute, by using the
LangevinFile command.

PS: Grande la U!

On Sun, 19 Jun 2005, [ISO-8859-1] Leonardo Sepulveda Durán wrote:

> Hello everyone!!!
> I just have finished some dynamics at 498K using explicit solvent NPT
> or NVE in NAMD, to unfold a protein. Nevertheless, it maintain its
> stability in my 2ns simulations. I have done implicit solvent
> simulations in CHARMM at 498K and the protein easily unfolds. So I
> think there would be some issue In the TIP3 solvent model (or better,
> he relation beetween protein and water paremeters) wich favours native
> state ensemble even under unfolding conditions used in literature (but
> with other solvent model and forcefield, F3C, flexible 3 center water
> model, and ENCAD), That is why I was wondering if there is another
> water model compatible with CHARMM FF able to be used in NAMD, to see
> if with it unfolding is easier to achieve.
> I have not revised the diferences beetween F3C and TIP3 yet, but I
> Think if the VdW parameters are more atractive unfolding can be
> favoured, I guess it would be more important than the use of bond and
> angle degrees of freedom, but I am not sure of that.
> So any comment about the issue would be very helpful
> Thanks
> Leonardo Sepúlveda Durán
> Universidad de Chile

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:39:37 CST