We had a strange thing happening in one of our protein structure
refinements using X-PLOR 3.1 (2.5 A data):
The model was first refined as a rigid body, and both R and R-free
converged to about 31 %
When we applied the SA protocol (3000K) the Rfree ended up at
35% while the R-factor dropped to 27%.
Does this make sense ???