[phenixbb] phenix.multi_crystal_average with twinned data

Thomas C. Terwilliger terwilliger at lanl.gov
Wed Mar 11 12:06:20 PDT 2009

Hi Peter and Mary,

For the cross-crystal averaging procedure, no refinement is done, so no
problem about using detwinned data in phenix.refine. in this case.  If
detwinned data is supplied then it will be used for density modification.
That is probably the best procedure at present.

(Note that when running autobuild, you can input a specific
"input_refinement_file" for refinement (raw data) and separately specify
an input_map_file with a map and a data file with (optionally) detwinned

All the best,
Tom T


Detwinning is possible with xtriage or phenix.refine (on way or the
other). Integrating this into density modification might be hard,
especially when the twin fraction is high. What is your twin fraction?

Tom, Is it possible to use a different set of amplitudes for density
modification then for refinement? You wouldn't want to refine against
detwinned data.



2009/3/11 Thomas C. Terwilliger <terwilliger at lanl.gov>:
> Hi Mary,
> I have not tried the phenix multi-crystal averaging procedure with
> twinned
> data, but here are some suggestions of how to proceed.
> 1. This routine does not do any refinement, only density modification.
> The density modification step does not know about twinning, so it will
> be
> necessary (if your data are significantly twinned) to supply structure
> factor files with detwinned data. I will check with Peter Zwart on
> producing detwinned data in phenix; it may be necessary to do that with
> other tools as normally phenix.refine uses the original data and does
> the
> calculations internally.
> 2.  The models you supply are used to (a) identify NCS symmetry and
> cross-crystal symmetry. Therefore it should not really matter if these
> models are incomplete, and therefore the approach you are using should
> work for improving the region where you have no model. The region where
> NCS is applied is defined as a region centered near the center-of-mass
> of
> the supplied model over which the NCS correlation is high.
> I hope that helps!
> All the best,
> Tom T
> I started running this function and then realized that I probably need
> to
> do something to take the perfect twinning into account.  I'm not sure of
> the best way to do this. I believe the map coefficients output by a run
> phenix.refine should be detwinned as I give a twin law when running
> phenix.refine.  Should I detwin the mtz containing the data and then use
> those files for the multi crystal averaging?  Is there a way to output
> detwinned data mtz files in phenix or do I need to use another program?
> Also, I have two ncs copies in the ASU.  There's a region of density in
> both ncs copies that I'm trying to improve the maps for with the multi
> crystal averaging.  I don't have anything built in the region yet.  Is
> there a way to use the NCS and multi crystal averaging at the same time
> without any pdb coordinates in that region?
> Even with my goof up about the twinning, the eff script I created
> appeared to run without problems.
> Thanks,
> Mary X. Fitzgerald
> Postdoctoral Associate
> Arnold Lab
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb

P.H. Zwart
Beamline Scientist
Berkeley Center for Structural Biology
Lawrence Berkeley National Laboratories
1 Cyclotron Road, Berkeley, CA-94703, USA
Cell: 510 289 9246
BCSB:     http://bcsb.als.lbl.gov
PHENIX: http://www.phenix-online.org
CCTBX:  http://cctbx.sf.net

More information about the phenixbb mailing list