Dear Phenix Community, I am having difficulties solving a protein-DNA complex structure and was hoping for some ideas or strategies. My data diffracts to 3A and depending on how I index my data fits well into I4 or P4. I have also been exploring everything in lower space groups. My MR model has 64% identity to my unsolved protein. The crystal protein is 24kD and runs as a monomer or sometimes a dimer. Xtriage strongly suggests that there is tNCS due to a strong patterson peak at 0.5, 0.5. 0.5. Using the phaser GUI, when I search around P4, I get very promising statistics (TFZ = 20's and LLG ~2000), and oddly, in these cases phaser places two subunits on top of each other (despite default packing limitations) with the protein beta sheet cores overlapping, and rotated 180 degrees of each other. The density looks believable, but obviously something is very wrong. The crystal packing would be feasible if another subunit were placed, but I have been able to coax phaser to find a complete solution. It would rather place it overlapping than in available spaces. My questions: 1) does this suggest that there is some form of twinning going on? 2) sometimes phaser (phenix GUI) tells me it is adjusting for tNCS and sometimes it says its NOT (even when there is an even number of subunits being searched for), is this a bug or deliberate? 3) I have tried using phaser from the command line in MR_AUTO mode with the keywords TNCS_USE ON TNCS PATT PERCENT 88 TNCS_NMOL 2 or 4 but dont see any indication that the TNCS adjustments are being made and I have not been able to repeat my phaser GUI results this way. 4) What is a good strategy for tackling this particular problem? Right now I am just chugging through my list of possible space groups, compositions and resolution ranges, but need to look through each possible solution individually because of the misleading statistics and so is very inefficient and time consuming. Thank you for any help or comments. Ian