[gmx-users] How to performe a monolayer dynamics?

[Erik Lindahl]

Hi Hector,



On Sep 22, 2004, at 4:59 PM, Hector Mrz-Seara Monne wrote:

> First of all, let me thank in advance,
>
> After reading several references about monolayers , and bilayers 
> specially those performed by Erik Lindal and Peter Tieleman there are 
> several things that worry me. My system isn't a tipical lipid 
> monolayer, it is contituted by hydrocarbonated-azo-benzene with a 
> carboxylated head and to start we want to reproduce an experimental 
> isoterm.
>
> I have charge the carboxylix acid at the end of the chain and then 
> when I made the monolayer and run the dynamic without water, the 
> monolayer is destroyed by the coulomb forces, as I expected.  But as I 
> read many authors said that they start the simulation without water to 
> allow the monomer to get a better tilt of chains. How they perform 
> that? How were they able to maintain the heads at z=0 for afterwards 
> insert the waters bellow?

First, you probably don't need to worry about the chain equilibration - 
those comments were written in the days when 10 ns was a long 
simulation. Now it's probably easier to simulate with water until 
things settle.

Second, if you choose to equilibrate without water the stability will 
depend on the type of lipids. DPPC for instance is nice and stable even 
without water, as long as you have it in a reasonable starting bilayer 
configuration.

Still, there are quite a few other solutions: you could add position 
restraints to an atom in the headgroup so they can't move too much (my 
personal favorite), or you can freeze e.g. the z coordinate of an atom 
completely during the equilibration. The details are available in the 
Gromacs manual.


>
> Another question is that, as I know the best way of calculating 
> couloumb forces is with PME in gromacs. That is true? Why then lots of 
> people using gromacs use a shifted cut-off? In case of using PME I 
> have to neutralize my system with ions ( in the hydrated monolayer)?, 
> and in the case of using a cut-off?
>

"Best" depends on your needs. It's the most accurate and only slightly 
slower (30%), but cut-offs scale much better over high number of 
processors in parallel runs and long cut-offs are usually a perfectly 
fine alternative when it comes to accuracy.

The PME algorithm works even for systems with a net charge, but due to 
the way it is evaluated the average charge of the system is taken as 
the zero level (grossly simplified), so it kind of corresponds to a 
background uniform charge that neutralizes the system. Conceptually 
it's probably easier to add counterions if you don't want to think 
about those effects...

Cheers,

Erik