Hello,

I am a relative newcomer to Phenix. I have a large multimeric protein structure docked into an electron density map, and I would like to further refine it by rigid-body refinement using Phenix. Is this a realistic strategy?

yes, this is a perfectly valid intent. Have a look at

https://www.dropbox.com/s/imv2yzwodik13in/afonine_etal_wcpcw2015.pdf?dl=0

that overviews and summarizes Phenix refinement tools for cryo-EM.

I have made some attempts, first by converting the .mrc map file to .mtz as outlined here:
https://www.phenix-online.org/documentation/reference/map_to_structure_factors.html

If you follow the above link you will see that it is best to refine the model directly against the map (in real space, that is), instead of taking a "detour" through reciprocal space.

So my suggestion to you is:

- get latest Phenix from nightly builds (it is essential that you use recent Phenix):

http://phenix-online.org/download/nightly_builds.cgi

- run phenix.real_space_refine (sorry, no GUI yet!):

phenix.real_space_refine model.pdb map.ccp4

If you expect the model to make large movements to fit the map, you can enable morphing or/and rigid-body, and also increase number of refinement cycles:

phenix.real_space_refine model.pdb map.ccp4 run=minimization_global+rigid_body+morphing macro_cycles=10

Then I tried to refine using rigid body refinement in Phenix. However, the program quits but doesn’t really tell me why. Here’s the last part of the log file:

Number of unique models: 24
  Time building chain proxies: 36.85, per 1000 atoms: 0.26
  Number of scatterers: 142296
  At special positions: 0
  Unit cell: (342, 342, 342, 90, 90, 90)
  Space group: P 1 (No. 1)
  Number of sites at special positions: 0
  Number of scattering types: 4
    Type Number    sf(0)
     S     264     16.00
     O   27720      8.00
     N   25872      7.00
     C   88440      6.00
    sf(0) = scattering factor at diffraction angle 0.

Writing MTZ file:
  /Users/xxx/Documents/Phenix/TEST/Refine_1/Test_refine_data.mtz

Sorry


Can anyone give me any ideas as to what went wrong?

Still, I would like to debug this.. Could you please send me the following files:

- original map;
- MTZ file that you get as result of conversion of this map into structure factors and the command you use to do this;
- input PDB file with the model;
- phenix.refine command (or GUI setup: .eff file and .log file) that crashes with the above error message.

All files will handled confidentially. Files may be large, so please use Dropbox or any similar file sharing utilities.

Also, is there a better way to do this?

Yes, see suggestions above!

Please let me know if you have any questions or need any help!

Pavel