Hi, Bob,
This is Ping. I like to talk to you about the possible implementation of
QM/MM method with PMEMD.
The current QM/MM method adopted from Weitao Yang's group has the advantage
of simplicity.
The total energy of the system: E(total)=E(QM)+E(MM)+E(QM/MM)
E(QM/MM) can be broken down to
E(QM/MM)=E(electrostatic,QM/MM)+E(vdw,QM/MM)+E(bonded terms,QM/MM)
In calculation, there are several tasks to be performed at different stages:
1. the energy calculation requires the return of E(MM)+E(vdw,QM/MM)+E(bonded
terms,QM/MM) from PMEMD;
2. the optimization of QM geometry requires the return of force on QM
(gradient) generated from E(vdw,QM/MM)+E(bonded terms,QM/MM);
3. the frequency calcualtion (and some optimization methods) requires the
return of hessian on QM atoms from E(vdw,QM/MM)+E(bonded terms,QM/MM);
(the above calculations requires the vdw and bonding MM parameters of QM
atoms and its geometry)
4. the optimization of MM atoms with respect to the cooresponding QM
geometry requires both geometry and partial charges, as well as other MM
parameters of QM atoms.
(this step is somewhat similar to the belly type calculation; QM atoms fixed
in space and run the MM optimization)
5. free-energy-perturbation calculation requires evaluation of E(QM/MM) for
different set of QM geometries and partial charges, and calculate the
difference.
A simple way to achieve this is to assign additional set of group for QM
atoms that its geometry and partial charges will be updated from external
files when appropriate.
There are some minor issues that need to be addressed, and we can talk about
it later.
I am looking forward to talk to you about it in details.
BTW, since the upgrade of the baobab system, I am having some trouble
compling the parallel version of PMEMD there. I finally get a working
version using ifc7.1 and mpich. I am wondering if you have had the success
on it.
Thanks,
Ping
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Tue Jan 11 2005 - 19:53:01 PST
