[gmx-users] MD - analysis

[Justin A. Lemkul]

Kavyashree M wrote:
> Dear Sir,
>  
> 
>     What does gmxcheck tell you about the .edr file that is giving weird
>     results? Do the plots look normal?  Perhaps a frame got corrupted
>     somewhere along the way.  The screen output should print how many
>     frames were considered in the analysis; if it does not match your
>     expectations based on nstenergy and the length of the simulation
>     then something went wrong.
> 
> 
> gmxcheck gave proper number of frames corresponding to 100ns, other plots
> temp, pressure, volume, density look fine
>  
> 
>     Most analyses do not need re-imaging.  Keeping the protein within
>     the confines of one unit cell is typically just a convenience for
>     visualization.
> 
> 
> Ok.
> 
> I had done simulations of four similar protein using the same mdp file. 
> in one of them
> the minimum distance between the periodic images went near 0.9nm, I had 
> used 1.0nm
> as distance between protein atoms and box wall. and cut offs were 1.0nm 
> for vdw and 1.4nm
> for columb. Till some 17ns the minimum distance was above 2nm the 
> gradually there was a
> dip after around 20-25ns. Now I ran 100ns simulation and I have to 
> discard this trajectory
> because of this error. I thought distance of 2nm between protein atoms 
> was enough as 1.4nm
> was the max cutoff.
> 
> How can we know prior to starting the simulation that we may get some 
> such errors for using
> such a parameter. I could have used larger box size but it will increase 
> the time. Is it trial and
> error basis to find out the optimum box size?
> 

Generally, no, you don't have to waste lots of time fiddling with the box size 
before you find the right one.  The choice is based on the cutoffs and the 
nature of the system.  For a well-folded, stable protein, I see no reason why 
what you've done isn't appropriate (unless you've set up the box incorrectly and 
only think you've set certain dimensions).  For disordered proteins or those 
capable of large conformational changes, then the box needs to be large to 
accommodate these possible motions.

The only other possibility I can think of is that the starting configuration 
compressed a lot over time, shrinking the box.  I don't know why this would 
happen for a protein in water, but I suppose anything is possible.  Most 
condensed phase systems should not be very compressible, but any such change 
would be obvious from plotting the volume over time.  If it is stable, then the 
protein must be doing something unexpected.

-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

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