Re: [Jmol-users] crystal structures and symmetry

[Rich <rg...@el...>]

On 29-4-2006 14:23, Bob Hanson wrote:
> Jmol users,
>
> I am experimenting with adding the capability for Jmol to read the 
> symmetry operations in selected file readers. These look like this:
>
> x,y,z
> -x,1/2+y,1/2-z
> -x,-y,-z
> x,0.5-y,0.5+z
>   
These are the existing _symmetry_equiv_pos_as_xyz datatype. There is 
also a new CIF dataname

_space_group_symop.operation_xyz

which is the preferred format to replace _symmetry_equiv_pos_as_xyz. The 
mmCIF equivalent is _symmetry_equiv.pos_as_xyz

> I have a working prototype for CIF files that can be played with at
>
> http://www.stolaf.edu/people/hansonr/jmol/test/proto/sym.htm
>
> Questions I have:
>
> 1) Who is interested in such capability?
>   
Me and all the crystallographers I know that (might) use Jmol plus the 
medicinal chemists and modelers that use crystallographic results. A 
non-trivial population.
> 2) How do those models look so far? Do you think they look "normal" or 
> "natural" with the whole molecules extending a bit from the unit cell 
> like that?
>   
Some of these are aethestic questions. The base molecule should be one 
with its COG within the unit cell and close to the origin. But this is 
not always done anymore (and especially so if SHELX alone is used 
without any checking done). If you start with such a base molecule and 
apply the symmetry operators you may well end up with almost a complete 
molecule hanging out of the cell outline. People accept that if they 
have experience with it. The option is to always apply cell translations 
to the result of the symmetry operator to keep the COG within the cell 
boundries. This is also a commonly used approach.
> 3) Notice that the maleic.cif file has 17 completely different 
> structures, one per frame. That look OK?
>
> 4) If we were to implement something that allowed one to "grow" a 
> crystal -- designate how many unit cells to combine, and in what 
> direction, what's the best way to do that? I mean, what sort of commands 
> might accomplish this, in your mind?
>   
The traditional approach is to have three values (one for each of a,b,c) 
that you can increment between some reasonable limits (-9, +9). You can 
force integers so you move out it whole unit cells (easiest 
computationally) or you can permit tenths of a unit cell if you want to 
let users "grow" outward with just those atoms that fit within the 
expanded boundry becoming visible.
> 5) Is it important to be able to select specific molecules (i.e. 
> covalently bonded sets of atoms)?
>   
Yes.
> 6) Is it important to be able to select just the file's "primitive" data 
> set -- that is, without symmetry operations applied?
>   
Yes.
> 7) Is it important to be able to control how the file loads -- applying 
> symmetry or not? If so, should the default be "no symmetry appied", or 
> should it be what you see here, "symmetry applied and normalized," or 
> should it be "symmetry applied and all atoms within the boundaries of 
> the unit cell"?
>   
All of these.
1) Just load the atoms as given (may be a disconnected set - especially 
if the molecule comes from SHELX)
2) Load the atoms and apply symmetry to make the base molecule fully 
connected (can be tricky if atoms are close to others and no explicit 
bonding info is given in the input file - you can end up with infinite 
chains if you are not careful)
3) load the molecule as supplied and apply all symmetry operators to 
produce a unit cell
> 8) Am I right in my analysis that although SHELX files DO have the SYMM 
> line, the complications involving LATT make it VERY complicated to do 
> this correctly within that file type?
>   
No. The convention for defining the lattice type is easily dealt with 
and once yo have created the full set of symmetry operators the 
application of them is no different to that for the CIF
> 9) Are there any complication with mmCIF files? For example, these files 
> don't have _symmetry_equiv_pos_as_xyz listed, so perhaps none of this 
> applies to them?
>   
Are the atom coordinates in mmCIF files crystallographic or Angstrom? If 
the former then things are pretty easy. If the latter then then the 
correct conversion has to be done to transform them back to the 
crystallographic system. Then apply the symmetry operators once you have 
the "real" coordinates.

Rich