[gmx-users] why protein comes out of the box

[Mark Abraham]

sangeeta kundu wrote:
> *Dear Sir,*
> ** 
> *    I am facing one  problem, I gave the MD run of a protein of 
> approximately 70 residues for 10ns, after 1 ns run part of protein 
> was getting out of the waterbox and at the end of 10 ns I found that 
> almost half of the protein was coming out of the waterbox, I gave the 
> simulation under NPT condition at 46C and 1BAR pressure, I used  43a1 
> force field and spc water.*

The protein can't come out of the box... the box is periodic. The 
protein can *look* like it's come out of the box if your visualization 
software doesn't know that the system is periodic.

> *I used the 43a1 force field, and prepared the water box by the commands *
> *editconf -bt triclinic -f conf.pdb -o box.pdb -c -d 1.0 *
> *genbox -cp box.pdb -cs spc216 -o water.pdb -p topol.top*
> ** 
> *But I wonder that  while executing the command*
> *grompp -f em.mdp -c water.pdb -p topol.top -o em.tpr*
>  
> *it did not display any charge ,

Okay, take a step back here. If you didn't know whether the protein 
should be charged before reaching this point of the system setup, then 
you haven't stopped to consider whether your simulation is actually 
modeling anything close to a real physical system. What ionization state 
do you expect for the ionizable residues? If you put garbage into an MD 
simulation you get computationally expensive garbage out...

> so I did not add any counterion to my 
> protein, in the subsequent steps em.pdb and pr.mdp files were produced 
> and in both the cases the protein was almost in  the centre of the box, 
> while I took the snapshot after 1ns the protein was still at the centre 
> , but after 1 ns it was coming to one corner of the box, and finally 
> almost half of it was outside the box when I completed the run, I can 
> not detect why it is coming outside the box?*

It's not, but the trajectory is written without attention to the 
periodicity, since mdrun doesn't know in advance that you plan to 
visualize and it would be convenient if a particular molecule was always 
at the center of the simulation box, and you can always reconstruct the 
"correct" images later (and it would be inefficient for you to tell 
mdrun in advance). If you do care, you can use the options to trjconv to 
choose something sensible. See the man pages.

> *Where am I doing mistake? Another question is how can I understand that 
> the simulation has been completed successfully?*

First, you need to know what you are hoping to learn from the simulation...

Mark