So this last CASP9 refinement target will not be easy.
The CASP9 organizers have told us that the best submitted model for this target had a GDT_TS of 56 which is quite low (a score of 100 is a perfect match to the native).
This means that you will need to heavily refine the starting model, especially in the regions that they mention (residues 10, 20 and 50) shown in note mode.
To make this even harder, the starting model they have selected is the Rosetta model submitted by our human group in the Baker Lab, which is trapped in a local minimum and highly minimized (essentially wiggled to death) so it will be very difficult to improve the score unless you get very close to the native state.
This is why we had to bring back the conditions for this one. It is in your best interest to focus on the regions with conditions immediately from the start (as opposed to wiggling/shaking the starting model) so that your starting very best score will be as low as possible.
This way, hopefully once you've made enough improvements on the 3 regions with conditions you will be able to beat the score of the starting puzzle.
Because we realize how much of a challenge this is, we will also post another version of this puzzle without conditions that will hopefully be useful.