[gmx-users] Lipid forcefield to use with TIP5P/OPLSaa

[Marc F. Lensink]

On Wed, Apr 23, 2008 at 03:22:09PM +0400, Vasilii Artyukhov wrote:
> 
> I'm afraid I have to use TIP5P; I _am_ looking at temperature dependent
> information, and also, I need a good description of hydration of  small
> molecules I mentioned. These and other considerations make more or less
> every other water model I know of (including TIPxP-Ew) at least twice less
> attractive than TIP5P. So if TIP5P doesn't screw up the description of the
> membrane as much as any other model would the water part, I think it will
> do.

if you could share your (future) experience with tip5p and membranes
with us, that would be great.  some time ago i decided to switch to
all-atom simulations, and then had to make the choice between tip4p
and tip5p.  now for the life of me i don't remember *why*, but i do
know that there were some strong arguments against tip5p.

> 2008/4/22 Marc F. Lensink <lensink at scmbb.ulb.ac.be>:
> >
> > i'm very happily using berger/oplsaa/tip4p.  mind that this means ua
> > lipid tails and the rest aa, so unless you have a distinct lipid
> > phase, it's probably not what you want.  personally i'm not (yet)
> > convinced it's a better combination than berger/gromos/spc, though.
> 
> I'm sorry, I didn't quite understand the argument about 'having a distinct
> phase'. Could you please elaborate?

suppose there's a protein in your bilayer.  if the lipids tails are
berger, ie united-atom (ua), but the rest of the system, most
importantly protein, is all-atom (aa), then you have an ua-aa contact
region, which the force field is not parameterised for.  i'd like to
add that this is not necessarily a problem.  current-day united-atom
force fields already have ua-aa contacts when they use explicit
hydrogens for polarised hydrogens (SOL, PHE, etc).

> So, this leaves us with two votes for Berger/TIP4P and nothing at all about
> CHARMM27. But it would be really nice to hear from someone who'd been using
> the latter FF. (I realize that it might be smarter to ask such questions on
> the CHARMM forum, but on the other hand, I'm still willing to do this in
> GROMACS...)

i hear very good stories about the charmm27 lipid tail parameters, but
haven't used them myself.  keep in mind that you will typically have
to use surface tension terms to keep a correct area per lipid.  not
too long ago new dppc head group parameters were published, for use
with charmm27, that do not require this surface tension term: Biophys
J (2007) 92:4157.  worth to check out, i'd say.

cheers,
marc