[gmx-developers] Exclusions

[Jason de Joannis]

Quoting Berk Hess <hessb at mpip-mainz.mpg.de>:

> Ana Sofia wrote:
>
>> Dear Gromacs-Developers,
>>
>> I'm trying to change my system between two states (A and B) by applying
>> the TI formalism based on the coupling parameter approach. Since the
>> conversion from state A to B involve breaking of a covalent bond between
>> two atoms, the excluded neighbours as well as the third neighbours (1-4
>> interactions) of these atoms will be different in state A and B.
>> As an example: a hexane molecule (state A) is transformed in two propane
>> molecules (state B). This implies the breaking of the bond between C3 and
>> C4 and the redefinition for state B of all the excluded neighours.
>>
>> Apparently, GROMACS does not allow the change in the exclusions between to
>> states.
>>
>> How can redife my exclusions?
>>
>> thanks in advance,
>>
>> Sofia Oliveira
>>
> This issue could be more complicated.
> What is your full thermodynamic cycle?
>
> There are two solutions.
> One is to remove the exclusions that are going to disappear
> and replace them by 1-4 interactions, where you probably
> manually have to enter the A and B state parameter to make
> them resemble the proper LJ interactions. Also the fudgeQQ
> of your force field has to be 1. If your force-field does not
> have a fudgeQQ of 1 and your have excluded Coulomb interactions
> it will not work.
>
> Another option would be to add all atoms of the B state as virtual
> sites, exactly on top of the A-state atoms. You can then make
> a separate B-state topology.
>
> Note that you might have trouble sampling at intermediate states,
> as the non-bonded and bonded lambda dependencies might
> not be well balanced.
>
> Berk.
>
>
For hexane<->2propane you could define special pairs, eg,

;ai aj funcA c0A c1B funcB c0B c1B
1 4 1 1.0 1.0 1 0.0 0.0
2 4 1 1.0 1.0 1 0.0 0.0
2 5 1 1.0 1.0 1 0.0 0.0
3 4 1 1.0 1.0 1 0.0 0.0
3 5 1 1.0 1.0 1 0.0 0.0
3 6 1 1.0 1.0 1 0.0 0.0

and therefore leave the exclusions on.

There are no charges to worry about in this example, but I guess it 
would be a problem if you had charges that need to be excluded in A and 
not in B.

Are you planning on using bond constraints or real bonds? It could be 
hard to use constraints for this. And that would be compounded by the 
change in degrees of freedom.

-- 
\ Jason de Joannis  \
\ Emory University  \
  \ jdejoan at emory.edu \
   \ 404-402-1332      \