[gmx-users] domain decomposition

[Mark Abraham]

On 5/04/2011 5:09 PM, mohsen ramezanpour wrote:
> Dear Mark
>
> Actually I don't know why.
> I just did the normal process as other my simulations.
>
> Let me discribe my work in details:
> I had a protein and a drug,I separated all residues around my drug (2 
> nm in radius) by PYMOL

You can't do that and hope for sensible results. The protein won't be 
happy if you expose its hydrophilic core to either a box of solvent or 
vacuum or implicit solvent.

> Then I saved the result as protein-new.pdb
> So.I used pdb2gmx to generate .top and .gro file for this .pdb file

pdb2gmx assumes you're giving it reasonable input - i.e. no missing 
residues in the protein. However, you've generated several chunks of 
missing residues above. pdb2gmx assumes there's a backbone peptide bond 
as normal, and this is too long. One can work around this issue, but 
still be crippled by the first problem.

> I entered the following commands for pdb2gmx:
> pdb2gmx  -f  protein-new.pdb  -o  protein-new.gro   -p   topology.top  
> -water  spc  -ignh
> and I used Gromos 43a1 force field.
>
> when I want to do EM there are an additional error that  results in 
> crashing the mdrun:
> Warning: 1-4 interaction ...  your system is exploding
>
> it says modifying interaction tables!
>
> Besides,I checked my pdb file,atoms 922 and 943 and 2 others who have 
> bad interactions,all of them are N atom of residues!

Yep, pdb2gmx has generated backbone peptide bonds that aren't sensible.

Mark