[gmx-users] Several general questions about correctness of molecular modeling

[X.Periole]

Hello Misha,

> 1) What RMSD value, obtained in molecular dynamics as a 
>function of time, is believed to be normal? I received, 
>that after 1 ns of simulation RMSD value augmented up to 
>2 A for backbone atoms. It seems for me very large.
> 

This a reasonable value. As David pointed out it depends 
on
the size of your protein.
  
> 2) Whether molecular dynamics should reproduce a 
>B-factor of protein residues, obtained experimentally in 
>X-ray analysis. I received in MD that B-factor for side 
>chains of protein residues is sufficiently bigger then 
>that from X-ray structure.
>

This normal when you consider that:
1) an X-ray structure is extracted from a crystal where 
the
protein is certainly constrained by the crystal paking.
2) the experiments are generally made at very low
temperature, which again will reduce the protein 
"dynamics".
In general the B-factor should be qualitatively reproduce.
Significant differences can be observe in loops, or 
solvent
exposed regions that rearrange from the crystal packing.

> 3) What minimal time is needed to perform a molecular 
>dynamics of a protein in water? How the time, required 
>for equilibration, should be determined? Is it sufficient 
>to see, that Potential energy reached a plateau, and all 
>the trajectory after this time can be used for analysis?
> 

The minimal time necessary for an MD simulation is 
completly
arbitrary ! At the stage MD simulations are now there is
almost no definition of minimal time. The equilibration of
the potential energy is very fast compared to the
simulation time we can reach now (100s of ns). A 
simulation
of a pure protein in a solvent ... reaching equilibration
...  is now a days of no much use. We know we can do it
for any protein for which the structure has been resolved. 
 This actually gives very little information about the 
protein itself. You need to ask a question conserning the
function of the protein and generate (several) simulations 
that would give the best answer you can give. The analysis
performed on the simulation is generally what makes the 
interest. What you get out of it !!
To answer your question, yes you need at the Epot 
equilibrated to consider that you have reached a region
of "equilibrium" and begin any analysis.

Good luck,
XAvier