- cctbxbb - phenix-online.org

Re: [cctbxbb] Problems with conversion from conventional to primitive cell
by Jörg-Rüdiger Hill 25 Nov '05

25 Nov '05

Thanks for your script. I am doing this in C++. I can't find a xray::structure class in the C++ documentation. Does this only exist in Python ? Nevertheless, I run your script in Python and the result is somewhat different from what I get, but I still think it is wrong. If I create and visualize a supercell from the coordinates given by the script (e. g., with gdis) I no longer see the hexagonal arrangement of the atoms typical for the conventional cell. Since the conversion from conventional to primitive cell should only redefine the cell but not the atom positions (in Cartesian coordinates) I believe the output is incorrect. Best Regards, Jörg-Rüdiger Hill >--- J?rg-R?diger Hill <Joerg-Ruediger.Hill(a)scienomics.com> wrote: >> Dear All, >> I am trying to convert from a conventional cell to a primitive cell using >> cctbx, but I am having some problems. > >Are you doing this in C++ or in Python? > >> What I do is the following: >> 1) I construct a spaceGroup object using the space group symbol >> 2) I retrieve the change of basis operator using z2p_op() from the space >> group object >> 3) I apply this operator to the conventional cell >> 4) I take a list of symmetry unique atoms and apply the operator to their >> fractional coordinates >> At this point I have a primitive cell which agrees exactly with what I get >> from other codes. > >Note that there are ambiguities in the choice of the change-of-basis operator >to the primitive cell. You may get different results from different programs, >which are all correct. It is a coincidence if you get the same result. > >> The problem comes when I am trying to construct all atoms >> from the symmetry unique ones. I do this by the following procedure: >> 5) I take the space group object and call its change_basis() method with >> the z2p_op() operator >> 6) I loop over all order_z() symmetry operations in the space group >> applying the operation to the fractional coordinates making sure that I >> don't create two atoms at the same position > >You could use the site_symmetry and sym_equiv_sites classes to deal with the >special positions. > >> The whole procedure works well, e. g., for cubic crystals. However, if I >> take a hexagonal crystal (e. g., Cr2O3, R -3 c) the crystal build by steps >> 5 and 6 above is not correct. In this case the conventional cell has a=b != >> c, >> alpha=beta=90, and gamma=120, but the primitive cell has a=b=c and >> alpha=beta=gamma. > >That's what I'd expect. > >> I have tried to add ":R" to the space group symbol and >> constructing a new spaceGroup object in step 5 above, but that does not >> make any difference. What do I have to do to build the crystal in this case >> or what is wrong with my procedure ? > >Attached is a copy of the new cctbx/cctbx/examples/cr2o3_primitive_cell.py >script (in the CVS already; will also be in the next set of auto-build >bundles). Could you please compare your results to the output of this script? > >The script shows a high-level implementation. I could add a lower-level >equivalent that is more in line with what you'd do in C++. Let me know if > this is what you need. > >Cheers, > Ralf > > >from cctbx import xray >from cctbx import crystal >from cctbx.array_family import flex > >def demo(): > """ > Result of ICSD query: > N * -Cr2O3-[R3-CH] Baster, M.;Bouree, F.;Kowalska, A.;Latacz, Z(2000) > C 4.961950 4.961950 13.597400 90.000000 90.000000 120.000000 > S GRUP R -3 C > A Cr1 0.000000 0.000000 0.347570 0.000000 > A O1 0.305830 0.000000 0.250000 > """ > crystal_symmetry = crystal.symmetry( > unit_cell="4.961950 4.961950 13.597400 90.000000 90.000000 120.000000", > space_group_symbol="R -3 C") > scatterers = flex.xray_scatterer() > scatterers.append(xray.scatterer( > label="Cr1", site=(0.000000,0.000000,0.347570))) > scatterers.append(xray.scatterer( > label="O1", site=(0.305830,0.000000,0.250000))) > icsd_structure = xray.structure( > crystal_symmetry=crystal_symmetry, > scatterers=scatterers) > icsd_structure.show_summary().show_scatterers() > print > icsd_structure.show_distances(distance_cutoff=2.5) > print > primitive_structure = icsd_structure.primitive_setting() > primitive_structure.show_summary().show_scatterers() > print > p1_structure = primitive_structure.expand_to_p1() > p1_structure.show_summary().show_scatterers() > print > print "OK" > >if (__name__ == "__main__"): > demo() -- ------------------------------------------------------------------------------- Dr. Jörg-Rüdiger Hill Tel.: +33 1 53 43 51 05 Director R & D Tel. (direct): +49 89 613 05 700 Scienomics Fax: +33 1 53 43 92 92 17 Square Edouard VII E-Mail: Joerg-Ruediger.Hill(a)scienomics.com 75009 Paris, France Web: http://www.scienomics.com/ -------------------------------------------------------------------------------

2 1

Problems with conversion from conventional to primitive cell
by Jörg-Rüdiger Hill 24 Nov '05

24 Nov '05

Dear All, I am trying to convert from a conventional cell to a primitive cell using cctbx, but I am having some problems. What I do is the following: 1) I construct a spaceGroup object using the space group symbol 2) I retrieve the change of basis operator using z2p_op() from the space group object 3) I apply this operator to the conventional cell 4) I take a list of symmetry unique atoms and apply the operator to their fractional coordinates At this point I have a primitive cell which agrees exactly with what I get from other codes. The problem comes when I am trying to construct all atoms from the symmetry unique ones. I do this by the following procedure: 5) I take the space group object and call its change_basis() method with the z2p_op() operator 6) I loop over all order_z() symmetry operations in the space group applying the operation to the fractional coordinates making sure that I don't create two atoms at the same position The whole procedure works well, e. g., for cubic crystals. However, if I take a hexagonal crystal (e. g., Cr2O3, R -3 c) the crystal build by steps 5 and 6 above is not correct. In this case the conventional cell has a=b != c, alpha=beta=90, and gamma=120, but the primitive cell has a=b=c and alpha=beta=gamma. I have tried to add ":R" to the space group symbol and constructing a new spaceGroup object in step 5 above, but that does not make any difference. What do I have to do to build the crystal in this case or what is wrong with my procedure ? Best Regards, Jörg-Rüdiger Hill -- ------------------------------------------------------------------------------- Dr. Jörg-Rüdiger Hill Tel.: +33 1 53 43 51 05 Director R & D Tel. (direct): +49 89 613 05 700 Scienomics Fax: +33 1 53 43 92 92 17 Square Edouard VII E-Mail: Joerg-Ruediger.Hill(a)scienomics.com 75009 Paris, France Web: http://www.scienomics.com/ -------------------------------------------------------------------------------

2 1

mmtbx source tree moved to SourceForge CVS
by Ralf W. Grosse-Kunstleve 24 Oct '05

24 Oct '05

Today we've finally moved the mmtbx source tree from our private CVS server (cci.lbl.gov) to SourceForge. **** Old working copies checked out from cci.lbl.gov are no longer valid and should be removed. **** To check out a new working copy from SourceForge: Developer access: cvs [email protected]:/cvsroot/cctbx co -P mmtbx replace "user" with your SourceForge user name. Anonymous access: cvs -d:pserver:[email protected]:/cvsroot/cctbx co -P mmtbx It may take a few hours before the anonymous access is available and before the mmtbx tree is visible in the viewcvs web interface. For the records: we were hesitant to move more source trees to SourceForge because the CVS server was often heavily overloaded; a cvs update could take a significant amount of time. However, a recent hardware upgrade at SourceForge has improved performance dramatically. Performance has been very good for a couple of months. Therefore we feel it is a good opportunity to finally move the mmtbx sources to the proper place. This way, browsing the tree via viewcvs is available to everybody, and the automatic archiving of cvs activities (via email) should now work reliably for all developers. Cheers, Ralf __________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com

1 0

strange silicon
by Marc Gailhanou 17 Oct '05

17 Oct '05

Hello, I have some trouble with cctbx, I try to understand why and I do not succeed. I need your help ! I replaced quartz with silicon in the "beach in the box" example/tutorial. This is the resulting code : from cctbx import xray from cctbx import crystal from cctbx.array_family import flex silicon_structure = xray.structure( special_position_settings=crystal.special_position_settings( crystal_symmetry=crystal.symmetry( unit_cell=(5.4307,5.4307,5.4307,90,90,90), space_group_symbol="Fd3m")), scatterers=flex.xray_scatterer([ xray.scatterer( label="Si", site=(0.,0.,0.), u=0) ])) silicon_structure.show_summary().show_scatterers() for scatterer in silicon_structure.scatterers(): print "%s:" % scatterer.label, "%8.4f %8.4f %8.4f" % scatterer.site site_symmetry = silicon_structure.site_symmetry(scatterer.site) print " point group type:", site_symmetry.point_group_type() print " special position operator:", site_symmetry.special_op() f_calc = silicon_structure.structure_factors(d_min=1).f_calc() f_calc.show_summary().show_array() and running the script gives the following result result : Number of scatterers: 1 At special positions: 1 Unit cell: (5.4307, 5.4307, 5.4307, 90, 90, 90) Space group: F d -3 m :2 (No. 227) Label, Scattering, Multiplicity, Coordinates, Occupancy, Uiso Si Si 16 ( 0.0000 0.0000 0.0000) 1.00 0.0000 Si: 0.0000 0.0000 0.0000 point group type: -3m special position operator: 0,0,0 Miller array info: None Observation type: None Type of data: complex_double, size=9 Type of sigmas: None Number of Miller indices: 9 Anomalous flag: False Unit cell: (5.4307, 5.4307, 5.4307, 90, 90, 90) Space group: F d -3 m :2 (No. 227) (0, 2, 2) 0j (0, 4, 0) (120.217658126+0j) (1, 1, 1) (-84.1884955081+0j) (1, 3, 1) (65.4837172721+0j) (1, 3, 3) (57.4543830832+0j) (1, 5, 1) (-51.4875968695+0j) (2, 2, 2) (128.545406989+0j) (2, 4, 2) 0j (3, 3, 3) (-51.4875968695+0j) The 220 structure factor, at least, does not look correct . I do not understand the 16 multiplicity either. Is there something I miss ? Best regards Marc

2 3

Re: [cctbxbb] differing build and install directories?
by Ralf W. Grosse-Kunstleve 28 Sep '05

28 Sep '05

Hi Bill, Sorry that I dropped the ball on this question before. Erik outlines the best way of creating a portable cctbx bundle, but in this case there may be a simpler alternative: - simply cp -r the installation into the final destination - re-run python .../libtbx/configure.py mmtbx in the new build directory location. This will update all full paths. Note though that this simple approach will preserve all intermediate files created by the build procedure (i.e. all the .o files and some auto-generated C++ files). But please try this out first. If it does what you want we can think about ways of getting rid of the unwanted files. > The other issue is that I need to hard-code in the DYLIB path I hope that we don't have to do this, although I'd be willing to modify the central SConscript to optionally support hard-coded paths if there really is no alternative. > instead > of relying on an environment variable, which I am not allowed to set. All cctbx commands on PATH go through "dispatchers". To see how it works, look at any of the auto-generated scripts in the "bin" subdirectory of the build directory. The dispatchers are basically all the same; only the last few lines differ. Each dispatcher updates PYTHONPATH and DYLD_LIBRARY_PATH "just in time" to finally start the target command. I.e. there are no permanent settings in the environment. You just have to add the "bin" directory to PATH if you don't want to type the full path name to each command all the time (which will also work). Is this going to work for you? Cheers, Ralf

4 4

differing build and install directories?
by William Scott 27 Sep '05

27 Sep '05

Dear Ralf: Sorry to keep pestering you. I'm now trying to build debian packages, which means the build directory differs from the install directory. Because of this, a lot of the shell scripts, python scripts, and compiled binaries in cctbx have a path like /sw/src/fink.build/ ccp4-5.99.3-1/ccp4-5.99.3 hard-coded into them, and I'd like the final one to look like /sw/share/xtal/ccp4-5.99.3 (for example). Similarly for Debian linux. Is there a simple way to build in one directory but use the final install-path for configuring? I've tried changing $install_root in the file cctbx_install_script.csh, but that didn't work. Similarly, I noticed the install_paths hard-coded into the dynamic libraries for cctbx are relative, rather than absolute like the others, for examle: % otool -L cctbx_dmtbx_ext.so cctbx_dmtbx_ext.so: /System/Library/Frameworks/Python.framework/Versions/2.3/ Python (compatibility version 2.3.0, current version 2.3.5) lib/libboost_python.dylib (compatibility version 0.0.0, current version 0.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 88.0.0) /usr/lib/libstdc++.6.dylib (compatibility version 7.0.0, current version 7.3.0) /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0) Is there a simple way to change that? All the best, Bill William G. Scott Associate Professor Department of Chemistry and Biochemistry and The Center for the Molecular Biology of RNA Sinsheimer Laboratories University of California at Santa Cruz Santa Cruz, California 95064 USA

2 3

Testing.....
by Erik Mckee 26 Sep '05

26 Sep '05

Hello :) Does this work? I'm surprised there aren't any test messages in the archives before now.... However, perhaps I'm just talking to myself here..... *Tests for echo.....

2 1