[gmx-users] VDW switched off in CHARMM tip3p model

Christopher Neale chris.neale at mail.utoronto.ca
Thu Aug 15 00:04:47 CEST 2013

>From my inspection of the force field files included in gromacs-4.6.1, when you
use the charmm ff and the charmm tip3p (tips3p.itp) you get the hydrogen type HT,
which does have LJ on the hydrogen atoms, whereas you only get the hydrogen 
type HWT3 when you use the non-charmm tip3p (tip3p.itp).

If I understand Mark's comment to be that there is no good reason to use the charmm
tip3p water model with the charmm ff, then I disagree (See below).

I have posted an image ( http://tinypic.com/r/110ak5s/5 ) of the area per lipid (APL) 
for a POPC bilayer that I built with charmm-gui (83 lipids/leaflet; 43 waters/lipid)
and then simulated under the charmm36 lipid parameters for ~ 300 ns many times 
successively with different water models (2 repeats each, starting from different 
conformations out of charmm-gui). Note the large difference in the APL for the 
regular tip3p and the charmm tip3p.  In fact, some papers note that you get 
gel-phase bilayers with the charmm36 lipid ff and the non-charmm tip3p model when using 
certain cutoff treatments 
(Piggot and Khalid -- http://pubs.acs.org/doi/abs/10.1021/ct3003157 -- is one such paper).


-- original message --

You could do that, but it's extra cost for no known benefit. See

On Tue, Aug 13, 2013 at 7:15 PM, Weilong Zhao <wz24 at zips.uakron.edu> wrote:
> Hi,
> I noticed that for GROMACS versions of 4.5 of later, the force field for
> TIP3P water model in CHARMM27.ff package has a following line included:
>  The following atom types are NOT part of the CHARMM distribution
> ..
> .
> HWT3    1       1.008000        0.417   A       0.0     0.0     ; TIP3p H
> It seems that VDW interaction of the default hydrogen of water is switched
> off, which makes it not the correct TIP3P ff of CHARMM. However this
> hydrogen type HWT3, is used as default by gromacs as CHARMM water model.
> Even though the hydrogen without VDW is the original TIP3P water model, I
> wonder is there any problem to include it in CHARMM ff set? Really
> appreciate if anyone could share some opinion.

More information about the gromacs.org_gmx-users mailing list