unrefineable structure ?

schiweck at argon.biophys.mpg.de schiweck at argon.biophys.mpg.de
Wed Apr 30 04:36:37 EST 1997

Dear all,

I am stuck with the problem of an "unrefineable" antibody (Fab fragment)
structure. The situation is as follows.

I have collected data to 3 Angstroems from several crystals. All data can 
easily processed with DENZO and give consistent results (primitive tetrago-
nal spacegroup with dimensions of about 78, 78 and 143 Angstroems). In SCALE-
PACK processing with Laue symmetry P41212 gives a Rmerge (about 6.9 %) which
is only slightly higher than the one I get for Laue symmetry P41 (6.5 %).
Pseudo-precession images from the data processed in P41 also indicate that the
higher symmetric laue group is present. Assuming that either P41212 or the
enantiomorphic spacegroup P43212 is correct, one gets a Matthews coefficent of 
2.24 for one molecule in the asymmetric unit. 

For molecular replacement I used several Fab fragments all with different elbow
angles (as this is a mutant of a Fab fragment whose structure I have determined
previously at 2 A resolution, my search models should be really suitable).
In AMORE I get only one significant solution in the cross-rotation search 
for one of the Fab fragments. In accordance to that in XPLOR this search model
is the only one which gives a significant solution after PC-refinement.

In the translation search with AMORE a very dominant peak is obtained for 
this cross-rotation solution in P43212 with a correlation coefficent of 52.1 
and a R-factor of 45.5 % using data from 15 to 4 A (next peak CC=37.0, 
R 50.5%). In P42121 the solution is significantly lower with CC=36, R=50.5%).

Of course I then used the coordinates of this mol-rep solution for refinement
in XPLOR. Rigid-body refinement in P43212 for the whole molecule or the 
domains results in a drop from 48 to 42 % for the R-factor and from 47 to 44 %
in Rfree. In the following refinement the R-factor drops to about 30 % whereas
the Rfree first drop to about 41 % and then after the R-factor has reached a
value of about 35 % rises again to a final value of 45 % ! This behaviour is
independent of the refinement protocol. Even with torsion angle dynamics
which should solve my problem of underdetermination, (99.3 % completeness to
3 A but still about 12000 parameters to refine with 9000 reflections for 10 to
3 A) I cannot refine my structure. 
As everybody can imagine I checked the molecular replacement solution again 
and still feel that it is correct. 
What makes me sure about that is the fact, that residues which were not 
included in the search model (among those were the mutations I wanted to see) 
are visible in 2Fo-Fc and Fo-Fc electron density calculated from the "refined"

Does anybody know about a similiar case where low-resolution data 
(which of course are not brilliant but reasonable) were not sufficient for 
refinement or where this kind of overfitting occured. Any suggestions are
welcomed. (Unfortunately I can not collect better data, as e.g.no synchrotron
beam time is available at the moment).

Thanks, Wolfram


Wolfram Schiweck
Max-Planck_Institut fuer Biophysik
Heinrich-Hoffmann-Str. 7
60528 Frankfurt/Germany

e-mail: schiweck at mpibp-frankfurt.mpg.de     

More information about the Xtal-log mailing list