Haem-Fe and cobalamines-Co are not recognised by the CDKAtomTypeMatcher. I'm not sure if they are salts (there is a patch for Co3+ in the tracker). For example, here is PDB ligand code 1FH (a haem):
C[CH]1CHC2=Cc3n4[Fe++]5|6|N7=C(C=c8n5c(=C(c9cccc
c9)C1=N2|6)c(C=C)c8C)C(=C(CCC(O)=O)C7=Cc4c(CCC(O)=O)c3C)C
and here is PDB ligand code COB (a cobalamine):
C[Co++]|12|3|N4=C5C=C6N|1=C(CHC6(C)C)C(=C7[N-
]2=C(CH[C]7(C)CCC(=O)NCCHOP(=O)
O[CH]8CHCHn9cnc%10cc(C)c(C)cc9%10)
[C]%11(C)N|3=C(CH[C]%11(C)CC(N)=O)C(=C4C(CC(N)=O)[CH]5CCC(N)=O)C)C
note that the liganding nitrogens are all recognised as N.planar3 in 1FH, but one of the 4 nitrogens around the Co in COB is also null. Confusingly the other 3 liganding nitrogens in COB are of type N.sp2.3.
Gilleain, is this report still valid? Or now outdated and fixed by Nimish's work?