On Wed, Sep 14, 2011 at 9:51 AM, Fredrick Devadoss <
Fredrick.Devadoss.uni-konstanz.de> wrote:
> Dear Amber Users/Developers,
>
> I am facing a strange problem while doing targeted MD (TMD) simulation. To
> elaborate my work, I have to write this (big) mail.
> Here we go:
>
> I have done the targeted MD for a Protein-DNA complex, by taking its open
> conformation as the starting structure and closed conformation as the target
> structure. I tried to run this simulation for 10ns with 10 input files (1ns
> each – from TGTRMSD 8.286 to 1.500). After 2ns, the simulation was stopped
> after starting the 3 input file. When I looked into the coordinate file,
> there was no space between the two adjacent columns. This is due to the fact
> that the complete system was moving very fast in one direction and, thus,
> after 2 ns the coordinates of all the atoms are more than 10000 Angstrom
> away from the origin. My explanation of this is that the additional force
> introduced by the targeted MD leads to a fast center-of-mass motion. I tried
> to use the NSCM flag to remove the center-of-mass motion but this did not
> help.
>
> My second idea was to use the IWRAP flag in my input file to translate the
> molecules back into the original box of the periodic boundary conditions.
> Please find the attached input file (TMD1.in). After addingthe IWRAP flag,
> the simulation run all the 10 inputs (ie. 10ns). When I looked into the RMSD
> and restraint energy values, there were some sharp peaks in the starting of
> the 4ns, 5ns and 10ns in the simulation. Please find the attached files for
> the plots. When I looked into the structures at these time steps, the DNA
> molecule moved out of the protein-ligand complex, but inside (the other
> corner of ) the water box. But in the next snap-shot, its forming the
> complex again. This is caused by the fact that the DNA molecule but not the
> protein is wrapped in the time step. But here problems with the alignment to
> the reference structure and the following rmsd calculation of the targeted
> MD appear. During one simulation the wrapping does not seem to influence the
> rmsd calculation. But when the restart file of a previous calculation is
> read in and the DNA as wrapped one more time than the protein, not the
> complex is compared to the target structure but the DNA on one side of the
> box and the protein on the other side as separated distance molecules. Since
> the targeted MD uses residues of both the DNA and the protein in tgtfitmask
> and tgtrmsmask, the alignment as well as the rmsd calculation is spoiled.
>
>
> At this juncture, I would like to know the following:
>
> (1) Is there a possibility to avoid the acceleration of the system due to
> the artificial force of the targeted MD?
> (2) Or is it possible to remove the center-of-mass motion?
> (3) Or is there a possibility to tell amber that the complex is an entity
> which should be wrapped completely or not at all?
>
You could combine them into the same molecule (see SOLVENT_POINTERS and
ATOMS_PER_MOLECULE in your topology file). You can get a description of the
topology file here: http://ambermd.org/formats.html for description about
what each means. Essentially, sander/pmemd wrap by "molecule", so if you
tell them that your complex is a single "molecule", they will be wrapped
together.
However, I don't know how this would affect your simulation results since
amber uses the minimum image convention, so it shouldn't matter how they get
"wrapped" into the 'center' box...
Perhaps someone else knows why you're seeing what you're seeing.
HTH,
Jason
(4) Do you have any other idea to solve the problem (unfortunately using one
> 10ns simulation is not an option due to computer time limits of the queuing
> system of the used computer cluster)
>
> Expecting your reply and thanks in advance.
>
> Warm regards
> Fredrick
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 14 2011 - 07:30:04 PDT
