Thanks to Dr.Herman (on NO NO fit between model and map)
barani at mace.cc.purdue.edu
Mon Apr 15 23:52:19 EST 1996
-From hermanschreuder at mmd.com Tue Feb 27 09:27:21 EST 1996
-On Feb 26, 10:35pm, Barani wrote:
-> The protein is 18 KDa. Diffracts to 2.8 A. After homology
-> modelling, simulated annealing and positional refinements
-> the R-factor = 22.0 and R-free = 29.0. The geometry is
-> good (0.01 and 2.0). The refinements are well behaved. No
-> water molecule added yet.
-> But the electron density map is a NO NO fit with the model!!
-I had a similar experience with an enzyme-inhibitor complex where I
-had 2.0 A data and isomorphous crystals. The statistics were excellent,
-but the maps were garbage. It turned out that the data set used was
-rather strong and all the strongest reflections were rejected by the
-defaults of FWIND (I did not put any FWIND statement, because I wanted
-to use all reflections). Apparently, these defaults are quite low and
-the solution was to add a statement like: fwind=0.1=10000000. After
-including the strongest reflections as well, the maps looked excellent.
I would like to thank Dr.Herman for the suggestion which greatly
improved the quality of the map. Also a 2 sigma cutoff in reflections
in making the map was useful, as the weak reflections were highly
erroneous. The final map was incredibly superior and all pieces of
the puzzle fell in place. I hope that a larger value for fwind
(default value) gets introduced in the next version of xplor.
chumma at purdue.edu
The all new SUBROUTINE DIE calling card, yours for free.
More information about the X-plor