Hi all,
I have run into an unusual problem and am looking for some help. I am
working with a small protein that is 76 amino acids in length. I have
collected anomalous data and have located the 4/5 selenomethionine sites
using SHELX C/D and SHARP. The map that is produced from SHARP after dm is
pretty nice. I can see secondary structure and side chains. I have
submitted this map to PHENIX autobuild, and it was able to place about 85%
of the model. I have extended the model to be 90% complete. Upon
inspection of the model, the selenomethioine locations match both the
sequence and the peaks in the anomalous difference fourier map. I was
surprised to find that the model actually falls into 2 ASUs. When I start
my refinement using phenix.refine and/or refmac, the R-factors start off in
the mid .40s, but after a round of simulated annealing the R-free rises to
the mid .50s. I can go back and remove a section of the model, run a round
of refinement, and have the density return for that section of the map,
which I interpret as the phases being some what correct. Does anyone have
any ideas on what would cause the R-free to be so high? Again, I emphasize
that the experimental map was of rather good quality.
Thanks for your help,
Vlad
|