[gmx-users] b-factor

[Martyn Winn]

On Wed, 2010-06-30 at 13:07 +0200, Erik Marklund wrote:
> leila karami skrev:
> > Hi all
> >  
> > pdb file for my protein was obtained by solution NMR. this file is as 
> > follows :
> >  
> > ATOM      1  N   GLY A   1     -25.349  -8.577   4.055  1.00  0.00
> > ATOM      2  CA  GLY A   1     -24.037  -8.099   4.448  1.00  0.00
> > ATOM      3  C   GLY A   1     -23.580  -6.913   3.622  1.00  0.00
> > ATOM      4  O   GLY A   1     -23.652  -6.939   2.393  1.00  0.00
> > ATOM      5  H1  GLY A   1     -26.109  -7.957   4.035  1.00  0.00
> > ATOM      6  HA2 GLY A   1     -24.067  -7.811   5.488  1.00  0.00
> > ATOM      7  HA3 GLY A   1     -23.324  -8.902   4.328  1.00  0.00
> > ATOM      8  N   SER A   2     -23.111  -5.869   4.298  1.00  0.00
> >  
> > pdb file obtaind from g_rmsf  -f  pr.xtc  -s  pr.tpr  -n  pr.ndx -oq 
> > command is as follws :
> >  
> > ATOM      5  CA  NGL     1      20.794  51.547  38.010  1.00 272.65
> > ATOM     12  CA  SER     2      23.444  51.157  35.390  1.00 257.41
> > ATOM     23  CA  SER     3      25.254  48.487  33.290  1.00 189.09
> > ATOM     34  CA  GLY     4      28.474  47.777  31.510  1.00 114.27
> > ATOM     41  CA  SER     5      27.814  48.657  27.860  1.00 195.51
> > ATOM     52  CA  SER     6      31.164  48.167  26.020  1.00 217.99
> > ATOM     63  CA  GLY     7      31.934  49.987  22.770  1.00 300.70
> > ATOM     70  CA  LYP     8      32.204  48.057  19.510  1.00 248.64
> >  
> > so can I compare experimental B-factor with that calculated from MD?
> >  
> > any help will highly appreciated. 
> I'm currently calculating b-factors too, and there may be two things 
> that you must think about. One is, as I understand it, that the b-factor 
> in x-ray scattering experiments is unaffected by motions in the 
> scattering directoion, effectively reducing the observerd mds of the 
> atoms so that B = pi^2 * 8 * msd_p, where msd_p is only the motions 
> perpendicular to the scattering direciton. Another thing to consider is 
> wether isotropic b-factors is suitable in your case.
> 
> Erik

I don't think the scattering direction is a significant factor. Each
atom occurs in multiple asymmetric units, and so in multiple
orientations in the crystal, depending on the spacegroup. In any case,
the diffraction intensities are averaged over multiple observations,
taken at different crystal orientations with respect to the beam.

More serious is that the B factor soaks up many sources of uncertainty:
molecular motion (e.g. from crystal phonons), internal motions, static
disorder, errors in the data, effects of partial occupancy, etc. In my
experience, the rmsd calculated from a B factor is much larger than that
calculated from MD, and quantitative comparison is impossible. But you
may well be able to see qualitative similarities.

Note also that you are comparing the molecule in solution with the
molecule in the crystal, and there will certainly be differences near
crystal contacts.

There is no B factor for NMR structures. The closest equivalent is to
look at the variation between MODELs in a PDB entry.

Cheers
Martyn

-- 
***********************************************************************
*                                                                     *
*               Dr. Martyn Winn                                       *
*                                                                     *
*   STFC Daresbury Laboratory, Daresbury, Warrington, WA4 4AD, U.K.   *
*   Tel: +44 1925 603455    E-mail: martyn.winn at stfc.ac.uk            *
*   Fax: +44 1925 603634    Skype name: martyn.winn                   * 
*             URL: http://www.ccp4.ac.uk/martyn/                      *
***********************************************************************