[gmx-users] constant_force pulling

[Thomas Schlesier]

Hi all,
i tried out the constant_force pulling, to simulate a force clamp 
pulling experiment. But there are now some questions. But first describe 
the system and so on...
The system consists of two molecules which can bind through hydrogen 
bonds with no water. Each molecule has one atom which will be consibered 
for the pulling. ZUG is that from the molecule that should be pulled and 
REF is the one from the other molecule (in umbrella pulling it would be 
the reference group). So the system looks the following
REF----hbonds-----ZUG         ->   pulling in this direction.

In the *.mdp file i had the following parameters:
pull                =  constant_force
pull_geometry       =  direction
pull_dim            =  Y Y Y
pull_ngroups        =  1
pull_group1         =  ZUG
pull_k1             =  -500
pull_vec1           =  1.631 0.196 0.279

where pull_vec1 is the vector from REF to ZUG.
The manual states that for constant_force there is no reference group, 
so i took for pull_geometry direction, because it looks like that all 
the other options need a reference group.

grompp complained that the pulling is in absolute coordinates and that 
this can lead to artefacts (i think when the molecule starts to rotate). 
In the end both molecule were pulled together through space, because REF 
wasn't fixed (so the molecule could move freely).

Then i fixed the position of REF via position restraints and a freeze 
group (not both together, were two simulations). In both cases grompp 
didn't complain about the pulling in absolute coordinates. (I fixed the 
position of REF, because that's the same what the pullcode does in 
umbrella/position pulling with the reference group, the spring moves 
relative to the reference group so the reference group is fixed for the 
pulling).

So now to my questions:

1) Would this (to fix the position of REF) be the right way to simulate 
a force clamp experiment?

2) If so, what would be better for fixing freeze group or position 
restraint? I tend to the former because then REF is fixed totally, in 
the later cases the two molecule would move a little bit, till the force 
from the pulling and restraint will be equilibrated and then i have 
nearly the same case as in the freeze group, but REF could fluctuate in 
space (how much depends on the strength of the restraints, but i think 
it would not be very much).

3) What to do with COM? Remove nothing, or only the translation or 
translation and rotation? One remark, i disabled the pbc, but i would be 
interesed what one should use best without and with pbc (in the later 
cases there could be also water present).

Hope somebody has any ideas.
Greetings
Thomas