[gmx-users] Simulation Box break into 16 domains => Gromacs 3.3.3

[Justin A. Lemkul]

Mark Abraham wrote:
> Chih-Ying Lin wrote:
>>  
>>  
>>  
>>  
>> Hi
>> Sorry that i have posted the same message for several times.
>>
>> I used Gromacs version 3.3.3.
>> My simulation system = one protein + 20 ligand + water molecules ( 7x 
>> 7x 7
>> )
>> MPI setting => #PBS -l nodes=4:ppn=4,arch=x86_64   => 16 nodes in total
>> After doing the energy minimization, => the potential energy is extremely
>> high ( say, ten to the 9th order )
>> I visualized the " Simulation-System-EM-solvated.gro " after the energy
>> minimization.
>> Then, I found that the Simulation Sysmtem is devided into 16 domains very
>> clearly and the molecules (protein, ligand, and water) break into 
>> atoms in
>> the boundaries.
>> I have checked that the 20 ligands are not overlapped each other and 
>> are not
>> overlapped with protein from the beginning.
> 
> Chih-Ying sent me before and after .gro files off-list, and the
> structure really has done something like this (pic attached). After EM,
> the cubic simulation cell has partitioned four times along each side,
> and within each small cube, the waters have contracted as if to minimize
> surface area. The result is as if each small cube did an EM in isolation
> of any other such cube.
> 
> Are you sure you are using a 3.3.3 mdrun? Consult the top part of the
> log file. I'd guess the most likely problem is that you're seeing some
> bizarre artefact of your MPI or installation. How was GROMACS
> configured? Did you use shared libraries? Try a version without. Try
> single-processor grompp+mdrun.
> 
> If there's no light shed by the above, please send me your input protein
> and ligand structure files, and your pre-EM .tpr and I'll see if I can
> reproduce your workflow or result.
> 
> Mark
> 
> 

I have a strong suspicion that this all derives from the incorrect editconf 
command posted earlier.  I saw something like this once, long ago, before I knew 
what I was doing :)  The problem I had stemmed from incorrectly assigning the 
center of the system.

I would suggest re-building the system, starting from the editconf step, and 
centering the protein with -c in a box defined *either* by -d or -box, to see if 
that fixes things.

-Justin

> 
> ------------------------------------------------------------------------
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================