The patch in https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=CCP4BB;325e1870.1112 solves the problem with 3u57.
A few days ago I read the interesting paper http://journals.iucr.org/d/issues/2011/01/00/dz5216/dz5216bdy.html referring to a crystal structure containing a diselenide bond; I downloaded the model and structure factors for 2xsk, and cif2mtz refused to convert them. This turns out to be because the Bijvoet pairs measured in the header were described as _refln.pdbx_F_meas_plus _refln.pdbx_F_meas_plus_sigma _refln.pdbx_F_meas_minus _refln.pdbx_F_meas_minus_sigma while both the ccp4-6.2.0/lib/data/cif_mm.dic and the http://mmcif.pdb.org/dictionaries/ascii/mmcif_pdbx_v40.dic dictionaries require these to be _refln.pdbx_F_plus _refln.pdbx_F_plus_sigma _refln.pdbx_F_minus _refln.pdbx_F_minus_sigma It was trivial to fix the problem with a text editor (and this is a case where the right answer is to get wwpdb to fix the sf.cif file at their end), but this led to some discussion at Global Phasing as to what could be done to make cif2mtz handle sf.cif files with unusual data items without requiring modifying and recompiling the source code each time. A summary of where we got to is at http://www.globalphasing.com/buster/wiki/index.cgi?CCP4cif2mtzImproveIdeas I would appreciate any comments on how to proceed in this direction. Tom
