[phenixbb] Important change in phenix.refine output: MTZ file with data and map coefficients

Pavel Afonine pafonine at lbl.gov
Thu Mar 3 21:01:49 PST 2011

  Hi Everyone,

starting development version dev-686:


phenix.refine does not output *map_coeffs.mtz file. Instead, it creates 
prefix_serial.mtz file (example: lysozime_001.mtz) containing four sections:

- section 1: Original input data (Iobs or Fobs, and HL coefficients if 
provided) and flags (as supplied in input);
- section 2: Experimental data used in refinement (labels: 
F-obs-filtered, SIGF-obs-filtered). This can be different from "Original 
data" due to Iobs->Fobs conversion (if input data are intensities) and 
outliers filtering;
- section 3: Contains total model structure factors (labels: F-model, 
PHIF-model) used everywhere in refinement (for the formula, see: 
http://www.phenix-online.org/documentation/fmodel.htm). In the future, 
this section might be extended to explicitly list all contributions to 
Fmodel, such as Fcalc (corresponding to atomic model), Fbulk (bulk 
solvent contribution), and so on.
- section 4: Fourier map coefficients (exactly the same as used to be in 
*map_coeffs.mtz file).

This change is meant to reduce the confusion when people tended to use 
*map_coeffs.mtz file for refinement and PDB deposition, and facilitates 
the reproduction of refinement statistics (R-factors, for example).

Also, thanks to Jeff Headd, by default phenix.refine now converts input 
intensities (Iobs) to amplitudes (Fobs) using French & Wilson method 
(Acta Cryst A34, 517, 1978), so no need to run a specific program 
(Truncate, for example) to do this.

To see the content of a MTZ file use phenix.mtz.dump command:

phenix.mtz.dump data.mtz


More information about the phenixbb mailing list