We would like to use Coot to fix up the geometry between the backbone and
the covalently bound chromophore within GFP variants, but our efforts to
convince Coot to connect the polypeptide and the chromophore have thus far
failed. This includes using MAKECIF to generate a new geometry library
entry with specific LINK connections from the backbone C of Phe64 to the N
(or N1) of the chromophore and likewise for the continuing connection from
the chromophore C (or C3) to N of Val68.
Given these definitions, Coot does refine the zone but ignores the
linkage and lets the atoms involved float. A couple of uninformative error
messages are issued. Coot also seems to ignore LINK cards in a PDB file.
We had this working once, but something has changed in the most recent
version(s) of Coot. Does anyone have a recipe for incorporating covalently
bound ligands to a protein, with proper geometry?
A related question is: how are the ideal bond and torsion angles defined
for protein-heteroatom linkages? MAKECIF defines only the bond length. We
have been unable to discover documentation describing such definitions.
Thanks for any advice,
Jim Remington
--