Hi Wolfram, each phenix.refine run creates a MTZ file that contains four sections of data: 1) original data as supplied at input; 2) data actually used in refinement (which may miss a few outliers, truncated by resolution and/or sigma if requested by the user, Fobs obtained from Iobs using French&Wilson method, etc); 3) Fmodel: total model structure factors including all scales, etc. They can be used to reproduce reported R-factors exactly; 4) Map coefficients: 2mFo-DFmodel (with and without filling missing Fobs), mFobs-DFmodel and anomalous difference map (f input data is anomalous: that is contains Fobs(+) and Fobs(-)). R-free flags are part of 1) and 2) above. That's the file that is supposed to go to PDB one way or another (as MTZ or CIF). Pavel On 1/22/13 9:28 AM, wtempel wrote:
Hello all, based on ${prefix}_001_f_model.mtz from phenix.refine, I would like to prepare a cif file with structure factors. FOBS, SIGFOBS, R_FREE_FLAGS, FMODEL, PHIFMODEL, FOM are available in (+) and (-) versions, presumably representing Bijvoet pairs. My questions are: 1) are FMODEL, PHIFMODEL the most appropriate columns for calculated amplitudes and phases for the purpose of PDB deposition? 2) I understand that the PDB sf_tool does not accommodate separate (+)/(-) columns for some of the items. A job for phenix.reciprocal_space_arrays and a reflection file without anomalous differences? Alternatively, I imagine that phenix has a tailored procedure just for the preparation of PDB deposition, but I may not have used the correct web search terms to find the instructions. Thank you in anticipation, Wolfram Tempel _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb