[gmx-users] AFM-pulling

[Frauke Meyer]

Hi Alicia, hi Justin,

first of all: thank you, Justin, for the suggestions, I now have a cvs
code, with which the pulling works in parallel (on an SGI cluster) -
thanks a lot!

> As you suggested a previous mail a few months ago, we installed the updated
> CVS version of gromacs (20.09.2002), and we found some weird things:
> without constrains and a timestep = 0.0005 ps. In these conditions, the
> spring position dislocated 0.01nm in 10 ps of simulation (as expected),
> however the pulled_group (Pull-atm) has translated 14 nm by this time, which
> is absurd. Either we are doing something wrong, or something is not working
> on with the pull code.

I am trying to use this cvs code for pulling, too, and have the same
problem: a much too fast dislocation of the pulled groups (in my case, I
pull two groups away from each other).
Furthermore, while group_1 moves away from ref_group, group_2 moves in the
direction of ref_group!
These are the settings:
verbose = yes
runtype = afm
group_1 = nterm
group_2 = cterm
reference_group = Protein
reftype = com
reflag = 1
direction =  0.0 0.0 1.0
reverse = from_reference
update = 1
pullrate = 0.0001
forceconstant = 1500

According to the pdo file, the spring dislocation is correct (the two
springs move away with the correct speed, -0.05nm/ps and
+0.05nm/ps), but this is not the case for
the pulled groups (-0.9 and -2.9 nm/ps, so into the same direction!).

The 3.1.4 version gives reasonable results, though...
Have you already tried to solve the problem?
Hoping for your reply,
Frauke