[gmx-users] Re: DPPC lipid force field difference between upload one, benchmark one, and lipid.itp

[Erik Lindahl]

Hi,

> In the Gromacs website, forcefield section, you upload a ffG43a2x.tgz
> force field for the lipid. Also, in the Gromacs benchmark section, 
> there
> are one case for 1024 DPPC bilayer, in its topology file, it include 
> the
> force field. Also, in peter tieleman's webist, it has the lipid.itp 
> that
> is the force field of lipid from OPLS/Berger.
>
> After I compare the three forcefield, it seems the force field in the
> upload section of Gromacs website has some common place with Tieleman's
> lipid.itp. I am wondering the lipid parameter of forcefield from the
> upload section of Gromacs website are also from Berger et;; al, 
> Biophys.
> J. 72, pp. 2002-2013.as lipid.itp from Tielman's?

Both of the uses the method described in the Berger paper to 
reparameterize the LJ coefficients in the lipid tails.

The first generation (described in the Berger paper) used OPLS LJ 
parameters for the headgroups and reparameterized the tails to fit the 
properties on liquid pentadecane.

ffG43a2x builds on Gromos96 instead, to make it easier to use with 
proteins. The tails were reparameterized in the same way, but we also 
took care to get a correct trend between hydrocarbons of length 6 to 15 
(or 20, I don't remember that exactly).


> Also what is the difference of the force field used for the benchmark
> case of DPPC with the other two force field?

The forcefield is Berger, but IIRC the DPPC molecule description in the 
.top file is missing one dihedral. Since it is only used for 
benchmarking I don't want to change the topology and add interactions 
(it's no big deal, though - I can't imagine it affects things 
significantly).


> which one that do you recommend me to use for my DPPC analysis?
>
Can't say since it depends - "know your force fields" and make a wise 
selection :-)

The original Berger probably provides the best reproduction of 
experimental DPPC properties, at the cost of not having any possibility 
of doing protein interactions.

The modified G96 above can do proteins, but we haven't spent as much 
time optimizing it for DPPC.

And, you could always roll your own OPLS-AA/L version if you write a 
DPPC topology!

Cheers,

Erik