[CP2K:1426] Re: Specifying QM/MM system

[Laino Teodoro]

Hi

>  This is not always possible. One of the reasons that you chose QM/ 
> MM is to eliminate the need for parameters for a certain region of  
> you system (e.g metalloproteins)

even for metalloproteins you need inevitably to equilibrate the  
system at MM level. You may want in these cases keep part of the
region of interest frozen but you've anyway to create potential and  
topology files for the all system, in order to allow your
favorite program to run your system.
Have nice examples regarding metalloproteins where not even 1 ns of  
MM equilibration are enough to have a proper thermalized
system.. so thinking to do that at QM level is just too much optimistic.

Let me also stress that as long as you don't want to study chemical  
reaction classical forcefield can nowadays do an extremely
good job..

> Just to sweep away all the doubts.. charges for QM atoms, though
> requested are
> never used. The only classical contribution coming from QM atoms are
> the QM-MM non-bonded.
>
> I guess that you refer to van der Waals interactions, since the  
> electrostatic interaction are computed at a QM level.

as I said in the sentence above, MM charges corresponding to QM atoms  
are ignored. Only classical non-bonded (which could
also be different  than van der waals) for QM - MM atoms are kept.
But this are general rules that you can find in whatever basic paper  
on QMMM.

Teo