The problem is that the data is from a twinned crystal. Both orientation angles have been figured out manually for two lattices. Due to the overlapping, most of frames have relative high average c2 value (up to 6.0 sometimes). The data set reduced turn out to be good enough to yield right molecular replacement answer without any further modification of the data ( reduced by Scalepack). But for high resolution structure, the data is bad as regards to R value (28% after simulated annealing). So I do want to get a better data set. If one could modify the source code a little bit, it will do the work. The idea is:
- Do a prediction of spots from one lattice
- Pull out a box around each predicted spot, set the background to 0 (or averaged background value of the frame)
- Do everything else the same as original Denzo do for the reflections from the other lattice.
I tried to do it myself. But it is really difficult for me to decode the Denzo. Would you like to tell me if that is possible for you to do it for me or is there any help that I can get from you?
The algorithm needs to be somewhat more complicated. One has to include the possibilities of overlap between lattices. Maybe in the future.
All I need is a way of being able to process data from a twinned crystal and I'd be so happy.
Put detector as far back as possible to separate the two lattices. Process one at the time. Hope for the best. Do not expect outstanding results.