[gmx-users] Normal Mode Analysis

[Bert de Groot]

Matthew Lardy wrote:
> 

Matthew Lardy wrote:
> 
> Hi all,
> 
> Another quick question.  I have started performing normal mode analysis
> (nma) of proteins from the PDB.  So they should already be in a minimum
> energy configuration right?  

No. it's a minimum energy conformation under the experimental conditions
(temp, salt concentration, pH, crystal contacts etc) but NOT (necessarily)
under the computational conditions (force field, vacuum, etc). So you'll
still have to *thoroughly* minimize your structure before NM. Usually you'll
need to use conjugate gradients in double precision for that.


> If so how does one perform normal mode
> analysis of a protein alone?  Assuming that I have already deleted
> ligands, metal ions, and other proteins that might be in the pdb file.
> 
> What should the mdp file look like?  I have been using a mdp file that
> looks like the following:
> 
> title                    = protein in water
> cpp                      = /lib/cpp
> include                  = -I../top
> define                   = -DFLEX_SPC
> integrator               = nm
> emtol                    = 0.0001
> emstep                   = 0.0001
> nstcgsteep               = 10
> nsteps                   = 10000
> nstxout                  = 5
> xtc_grps                 = Protein
> energygrps               = Protein
> nstlist                  = 10
> ns_type                  = grid
> rlist                    = 1.4
> vdwtype                  = Shift
> coulombtype              = PME
> rcoulomb                 = 1.4
> rvdw                     = 1.4
> 

See the post of Berk Hess a few weeks ago. Use cut-offs with value 0 (infinity)
both for the minimisation and NM.

> For one, the nm run won't finish, not on my linux box at work.  :(  I dump
> core about halfway through the run, a filesize limitation error of some
> sort, so I figure I am making a major mistake before I start the run.
> What is really interesting, is that I can complete the run on my Powermac
> at home, but then I cannot diagonalize the matrix (that is a calloc error
> there).  I know that I should have posted the output of these two dumps,
> but I want to make sure that my general strategy is alright before I worry
> any more about these other errors.
> 

for NM you need to store the full Hessian matrix (both in mdrun and in g_nmeig).
This is a 3Nx3N matrix, with N the nr of atoms. This can easily increase above
your system's memory limits for larger proteins.



Bert

____________________________________________________________________________
Dr. Bert de Groot

Max Planck Institute for Biophysical Chemistry
Theoretical molecular biophysics group
Am Fassberg 11 
37077 Goettingen, Germany

tel: +49-551-2011306, fax: +49-551-2011089

email: bgroot at gwdg.de
http://www.mpibpc.gwdg.de/abteilungen/071/bgroot
____________________________________________________________________________