Misleading pseudobond at RNA chain break in mmCIF 5an9

[Tristan Croll]

Opening 5an9, residues N2700 and N2925 (the 3' and 5' sides of a chain break) are messed up. The pseudobond across the break goes from the O3' atom of N2700 to O3' of N2925.

[Eric Pettersen]

The problem description most closely matches the behavior of the mmCIF version of 5an9, though the current daily build does correctly bond N2925-26 (the pseudobond is nonetheless still wonky). The PDB version has different incorrect behavior, so I will open a separate ticket for that since Greg handles mmCIF and I do PDB.

--Eric

[Greg Couch]

The mmCIF reader tries to generate a pseudobond connecting the linking atoms of N2700 and N2925. Unfortunately, N2925 is missing the P, so the pseudobond is connecting to the closest atom. A possible solution would be to use the residue template to find the closest existing atom to the P to connect the psuedobond to.

[Greg Couch]

Fixed. For residues with templates and known linking atoms, if the linking atom is missing, use the closest (by template connectivity) existing heavy atom.