Hi JD,
Unfortunately, Chimera isn't really being as smart here as you are giving it credit for.  Many PDB files specify coordination bonds in CONECT records.  Chimera identifies those bonds and converts them to pseudobonds (otherwise its atom-typing code, which depends on correct covalent bonding, would get the bonded atoms wrong).  So PDB files that do not identify the coordination bonds with CONECT records (e.g. 1lwx) will not have any coordination bonds in Chimera.
I'm wondering if what you are hoping to do is along the lines of this previous mail to chimera-users:

http://www.cgl.ucsf.edu/pipermail/chimera-users/2007-May/001529.html

--Eric

                        Eric Pettersen
                        UCSF Computer Graphics Lab
                        http://www.cgl.ucsf.edu

On Nov 23, 2007, at 9:30 AM, Jean-Didier Maréchal wrote:

Hi Eric,

Thanks for that. I have been able to get a bit further. 

To follow with my problem, I have one question more though. Basically, I
would like to know how chimera "Recognizes" metal chelation and
coordination bounds. 

What I see is that when you open a pdb with a metal ion, the program
places correctly "semi bounds" corresponding to coordination
interactions. Unfortunately, I work with a series of protein structures
where metals do not adopt a typical geometries (ex: 1lwx). First, I
would like to be able to improve the number of acceptable metal
geometries, but my questions are: how does chimera recognize the right
geometry of the metal ions? Is it implemented in a python module? It is
a C++ code? 

All the best,
JD