[gmx-users] FEP + Ham. REMD

[Yuri Garmay]

Hi all!

I have a problem. So as it can be seen in the theme I am trying to improve
sampling of free energy calculating simulation by using replica exchange.
In the gmx4.6 it is simple while using FEP technique, but does not
implemented for umbrella sampling. But I want evaluate potential of mean
force between two flexible molecules, so I have decided to do this with
FEP+REMD.

It is desirable to select distance between the mass centres as a reaction
coordinate, but FEP has to do with bonds or constraints between single
atoms. Luckily there virtual sites are possible to define in gromacs. So I
have a thought of using it.

But I have never been using virtual sites and so have some questions.
I  introduced two virtual sites in topology with these lines:

--------------------
*.top:
[ atomtypes ]
;name mass charge ptype sigma epsilon
DD 0.0 0.0 V 0 0 ; nb interaction off?

[ implicit_genborn_params ]
; Atomtype sar st pi gbr hct
DD 0.0 0 0.0 0.0 0.0 ; interaction with solvent off?

[ atoms ]
.......some lines .....
   111         DD      9    DDD     DD    111          0
   112         DD     10    DDD     DD    112          0

[ virtual_sitesn ]
; Site from funct a d
111    2       1    2 ... ; virtual site at mass centre of some groupe
112    2       82   83 ...

[ bonds ]
....
  111   112     6           0.3           1000            1.3
1000 ; harmonic bond
....

*.gro:
    9DDD     DD  111   0.000   0.000   0.000 /// position have to be
determined automatically?
   10DDD     DD  112   0.000   0.000   0.000
------------------

The energy minimization showed correct constraint length at first sight.
But does this method solve problem I have? I will be very much appreciated
if someone experienced explains me whether this is correct or there are
pitfalls at this way.


-- 
Best regards,
Yuri