[cctbxbb] rotate_translate_map function causes "splinching"

[Pavel Afonine]

Hi Nick,

could you please send complete script that reads the map, shifts it and 
then writes it out? It isn't clear what files you used..

Also, if you could reduce this to a small toy example (small box, just 
one blob) then it will help too.

Pavel


On 11/15/17 08:11, Pearce, N.M. (Nick) wrote:
> Hi all,
>
> I’m having a problem with the rotate_translate_map function from 
> cctbx.maptbx.
>
> I have a 3D array (a mask of 0s and 1s, in this case) that I’m writing 
> as a CCP4 map file. If I write it out “on the origin” (without running 
> rotate_translate_map), then it’s fine (see attached “before.png”).
> However, if I apply the rotate_translate_map function and then write 
> it out, odd features appear at the edge of the unit cell (see 
> “after.png”).
>
> It is NOT the case that the map is going through the “periodic 
> boundary conditions” of the unit cell (the shift is less than one unit 
> cell, as shown by the structure which is transformed by the same amount).
>
> The minimal code is:
> ———————————
> array = cctbx.maptbx.rotate_translate_map(
> unit_cell=self.unit_cell(),
>  map_data=array,
> rotation_matrix=scitbx.matrix.rec([1,0,0,0,1,0,0,0,1], (3,3)).elems,
> translation_vector=(-1.0*scitbx.matrix.rec((X,Y,Z), (3,1))).elems
> )
>
> # Write the map
> iotbx.ccp4_map.write_ccp4_map(
>                               file_name=f_name,
> unit_cell=self.unit_cell(),
> space_group=self.space_group(),
>                               map_data=array,
> labels=flex.std_string(['’])
> )
> ———————————
> where (X,Y,Z) is a tuple for the origin shift.
>
> Is there something stupid that I’m missing? Has anyone else had the 
> same problem?
> The only code for a rotate_translate_map function I could find was 
> actually in maptbx/average_densities.h and I couldn’t see anything 
> obviously “wrong”… though I’m not that familiar with C++.
>
> The maps and files can be found at 
> https://www.dropbox.com/sh/xqrsk5lcn603416/AABiVO74g6A_jZMGn_DjxOmka?dl=0
> all “grid” files are without translation, and all “ref” files are 
> after translation. both are written out exactly the same (the 
> iotbx.ccp4_map call above).
>
> I’m using the cctbx bundled with a ccp4 nightly from 2016-10-22.
>
> Thanks,
> Nick