[gmx-users] Best Lipid parameters...

[Erik Lindahl]

jozef hritz wrote:

>	Hi Erik,
>
>>The ffG43a2x is an extension where one of my diploma students 
>>reparametrized carbons to suit the CH3 and CH2 for long chains. These 
>>parameters reproduce the area/lipid of at least DPPC within 1% of the 
>>experimental value (it is slightly dependent on the size of your system 
>>too). We've used it for tons of simulations with very good results.
>>
>>There are a newer set of Gromos96 carbon parameters which should improve 
>>the situation significantly, but I've not been able to test it myself 
>>and it's not included in the distribution yet.
>>
>
>I understood this point, but the question is if it is also possible use
>for protein/membrane/water systems without adding parameters describing
>interactions between atoms of proteins and membrane. Do You think
>combination rules for LJ are enough?
>

Yes, that should work just fine - you shouldn't have to define anything 
extra. In most places Gromos96 already uses combination rules, and where 
it didn't do it we simply scaled the new CH2/CH3-(other atom) 
interaction parameters by the same amount as they were scaled in the 
vanilla version of Gromos96 (relative to the combination rule value). As 
far as I know there isn't any forcefield where people have derived 
special parameters for (membrane)-(protein) interactions.

In any case I wouldn't worry too much about the protein; the changes to 
the parameters are quite small - the only reason it is important for the 
membrane is that we pack thousands of similar atoms in the chain region, 
so there the effect can be larger.

Cheers,

Erik