[phenixbb] gridding of maps & coot

Roger Martin roger at quantumbioinc.com
Tue Mar 17 11:22:12 PDT 2015


Hi,

in making ccp4 formatted binary files readable by coot there are two 
*determine_gridding methods on the c++ side of cctbx ; 
*http://cctbx.sourceforge.net/current/c_plus_plus/namespacecctbx_1_1maptbx.html
1)
template<typename IndexValueType>af::tiny< IndexValueType, 3 >     
determine_gridding (uctbx::unit_cell const &unit_cell, double d_min, 
double resolution_factor, af::tiny< IndexValueType, 3 > const 
&mandatory_factors=detail::one_one_one, IndexValueType max_prime=5, bool 
assert_shannon_sampling=true);
2)
template<typename IndexValueType>af::tiny< IndexValueType, 3 >     
determine_gridding (uctbx::unit_cell const &unit_cell, double d_min, 
double resolution_factor, sgtbx::search_symmetry_flags const 
&symmetry_flags, sgtbx::space_group_type const &space_group_type, 
af::tiny< IndexValueType, 3 > const 
&mandatory_factors=detail::one_one_one, IndexValueType max_prime=5, bool 
assert_shannon_sampling=true);

 From the python side these get called within
http://cctbx.sourceforge.net/current/python/cctbx.maptbx.html#crystal_gridding
__init__(self, unit_cell, d_min=None, resolution_factor=None, step=None, 
symmetry_flags=None, space_group_info=None, mandatory_factors=None, 
max_prime=5, assert_shannon_sampling=True, pre_determined_n_real=None);

In phenix code it appears that only resolution_factor is ever set and 
internally then only the first determine_gridding method ever gets 
called. For numerous space groups these grid dimensions aren't 
compatible with symmetry operations and if symmetry operations are added 
to the ccp4 binary file coot has issues with asu (asymmetric unit) 
fitting to the grid. http://www.ccp4.ac.uk/html/maplib.html
If the symmetry operator equations are skipped then coot will read as if 
the map is P 1 and not know the space group and asu's. symmetry_flags 
would have to be set to something non-None to call the second method.

And if the second determine_gridding method(which is spacegroup aware) 
would be used, the grid dimensions are compatible with symmetry 
operators and coot reads and works with the asu's (and my own custom 
code is also happy).

For all the maps produced from phenix.refine runs I've examined they 
have a space group number between 1 and 230, no symmetry operators 
included and grid dimensions from the first determine_gridding method.

Now there are a number of space groups which have different axes 
chosen.  For example 155 can be R=Rhombohedral 
http://img.chem.ucl.ac.uk/sgp/large/155az1.htm or H=Hexagonal 
http://img.chem.ucl.ac.uk/sgp/large/155bz1.htm axes.  The symmetry 
operators identify which one or some systems prepend a number on the 
space group number such as 155 -> 1155.  But coot (0.7 or 0.8.1) doesn't 
respond to that; only to the symmetry operators I suppose because the 
symmetry operators are the most trustworthy.

The functionality is already in cctbx. Is there a reason the second 
determine_gridding method(spacegroup aware) is not used by phenix?

And a reason symmetry operator equations are not added to the output 
ccp4 binary files?

How should a ccp4 binary know the R or H axes of the space group? 
(according to its format http://www.ccp4.ac.uk/html/maplib.html)

-Roger






-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20150317/45ce72e9/attachment.htm>


More information about the phenixbb mailing list