Rhys, since you can obtain good native crystals, I would try Xenon phasing. I have recently heard several success stories at the Australian Synchrotron, and it just seems to be a straightforward and rational way.
Molecular Replacement should also be an option. I did this with beta-barrel proteins quite a bit, and even wrote a program for elliptic distortion of the model, to arrive at models that may be closer to the target. good luck, Kay
