Dear Dr. Terwilliger,
Its a pleasure writing to you. I have good memories of the EMBO workshop in Grenoble last year.
As you commented, I have used combinations of HA plus mtz, mtz alone, model alone, model and mtz. When the map coefficients come from a 2DFo-mFc synthesis generated by phenix.refine there is no problem as phenix.find_ncs readily detects the labels. The problem arises when I try to use an mtz from CCP4 coming from density modification in DM or directly from phasing as the column labels are not what the program wants. I suppose changing the labels should do it?.
With the heavy atom and mtz case another peculiarity is that it finds operators with cc of 0.64 from the ha sites and then moves to the map for searching operators giving a cc above 0.8 and finally reports operators with a cc of 0.54 ( presumably from the map). Unfortunately the ha sites require a mtz with map coefficients to work. I was wondering if there is a way to use the ha sites alone in phenix.find_ncs?( apart from getting the matrix from superposition in coot or the programs in CCP4 suite)?. I was unable to find a way from the manual. The present case is a bit tricky because of presence of a tNCS.