[gmx-users] g_rmsf

Xavier Periole x.periole at chem.rug.nl
Mon Feb 9 15:29:00 CET 2004


hugo verli wrote:

>Initially the simulations were 4.8ns long. But when I did something like your
>suggestion, i.e. rmsf analysis of each 400ps I observed the effect of the
>equilibration in some loops of the simulated protein, with a very high rmsf
>(<0.8nm) in the first 4 ns that decrease considerably in the last 400ps
>(<0.2nm). The radius of gyration also decrease in this first 3 or 4ns, and I
>understand to means the equilibration of some loops suffering of crystal packing. 
>
I would consider that 400 ps is small to generate fluctuation. 1ns is 
probably the
minimum to use considering the lenght of simulation you can generate 
nowadays.
When I mentioned the convergence of your fluctuations I meant to accumulate
your fluctuations. i.e. compare 0-1ns, 0-2ns, 0-3ns, etc ...
At one point increasing the lenght of the simulation should not change 
your fluctuation.
That is of course in the case that you are exploring one state of your 
protein. (exploration
around one minimum). If there is conformational change of your protein 
that becomes
more complicated because the average structure is a mix of differents 
"states" ...
At that point you need to explore much more the conf space to get 
reasonable results.

In your case the loops are reorganizing because of the release of the 
crystal packing
so I guess that till they reach their new conf your simulation the 
fluctuations are
big but are not really reflecting the reality of the flexibility of your 
system "equilibrated".
Looops are "always" flexible.

>
>
>The first thing I did was to increase the conformational sampling of the
>simulation, going to 8.0ns. The average of this 4.8 - 8.0ns simulation has a
>rmsf value more stable, always lower than 0.2nm.
>
>But looking the simulation in windows of 400ps I observe some kind of periodic
>movements, i.e. a loop that increase and decrase its rmsf value along the
>simulation. This profile is not shown in the average file. Because of this
>behaviour I thought it would be nice to make some .xpm matrix with the residue
>number, the time and the rmsf value in color, as the third dimension. So the
>frequency of the peaks could indicate the periodicity of the movement.
>
I don't know about the relevance of the periodicity on your fluctuations.

>
>I don't know if I was able to make myself clear, but if don't, please, tell me.
>
>Hugo.
>
XAvier

-- 
----------------------------------------------

   Xavier Periole - Ph.D.

   Dept. of Biophysical Chemistry / MD Group 
   Univ. of Groningen
   9747 AG Groningen
   The Netherlands

   Tel: +31-503634329
   Fax: +31-503634800
   email: x.periole at chem.rug.nl
   web-page: md.chem.rug.nl/~periole
   
----------------------------------------------






More information about the gromacs.org_gmx-users mailing list