From: Nicholas Musolino (
Date: Tue Aug 03 2010 - 08:59:00 CDT

On Aug 3, 2010, at 7:45 AM, koushik.k wrote:
> My system is protein + water sphere (not water box).
> I have done SMD using constant velocity pulling.
> I want to save the coordinates for all frames.
> I have tried catdcd, and also the SAVE COORDINATES option in the VMD main
> window.
> But, i have been getting a PDB WRITE ERROR.
> when i went through the VMD mailinglist, i have found that the error may
> be due to unexpected values of X,Y,Z or Beta or Occupancy.
> So, i tried fitframes.tcl script to fix the system's transformation and
> rotation.Even that did not help.
> Then i tried to fix the center of mass of system to the center coordinates
> {0 0 0}. But, still i was getting a PDB WRITE ERROR.
> The PDB WRITE ERROR was always for an atom belonging to a water molecule.
> Upon close inspection, i found out that when you load the psf and dcd file
> in VMD and view the system through the trajectory, you find that some
> water molecules shoot out from the water sphere.
> This might be the reason for getting a PDB WRITE ERROR because when the
> water molecules shoot out their X,Y,Z values become too high for VMD to
> write.
> Can anyone suggest a solution to this problem.

Hello Koushik,

Is the solution that you're interested in:
(a) performing a production simulation in which water molecules don't fly off into space, or
(b) creating a PDB of your system as it is now?

If you would like to try approach (b), you can select a subset of atoms in VMD using the scripting commands
set spheresel [ atomselect top "within 30 of protein" ]
$spheresel writepdb mysystem.pdb

I haven't tested this and I'm not sure the exact selection text you'll need to use, based on your system and sphere size.

One general note: I think the cautious advice would be to revisit problem (a), and take the flying water molecules as a symptom of an SMD simulation that is being steered too quickly, or some other problem, but I'm not experienced with SMD and I'm not sure what you're trying to measure/accomplish with your simulations.

PS. Thanks for searching the list archives ahead of time.

Nicholas Musolino
Ph.D. candidate, Molecular Engineering Laboratory
Department of Chemical Engineering, MIT | 617-253-6675 | Room E19-528