[phenixbb] refinement problems

[Christine Gee]

Hi
This is a more general crystallography question maybe.  I am using  
phenix.refine and I have a model structure refined against data at  
3.4A.  The model was built from experimental phasing.  I reprocessed  
the same dataset, to include another angstrom resolution (3.3).  It  
has 50% of the data from 3.4 to 3.3 and the I/sigI is about 2.  When  
I refine my structure at 3.4 the stats are 32 and 36 for R and Rfree,  
but when I use the phase extended data (and the same reflections are  
the test set up to 3.4A as long as CCP4i did the job right with  
setting the Rfree) the stats suddenly jump to 49 and 56 when I refine  
the model against the higher resolution data.  Is there something I  
am missing?  I am doing a rigid and an anneal before individual sites  
and adp refinement.  The space group is P3121.
Regards
Christine