SMILES output wrong for strange CO species.

[rwest]

I'm not sure what this represents (still not used to the lone pairs column) but I'm pretty sure there are only two atoms, a C and an O

1 *4 C 1 0 {2,S}
2    O 0 2 {1,S}

However, the SMILES (generated via RDKit) is [CH2]O

In [19]: print m.fromAdjacencyList('1  *4   C  1 0 {2,S} \n  2 O 0 2 {1,S}').toAdjacencyList()
1 *4 C 1 0 {2,S}
2    O 0 2 {1,S}


In [20]: print m.fromAdjacencyList('1  *4   C  1 0 {2,S} \n  2 O 0 2 {1,S}').getFormula()
CO

In [21]: print m.fromAdjacencyList('1  *4   C  1 0 {2,S} \n  2 O 0 2 {1,S}').toSMILES()
[CH2]O

Strangely, RDKit agrees that there are two atoms:

In [22]: r = m.fromAdjacencyList('1  *4   C  1 0 {2,S} \n  2 O 0 2 {1,S}').toRDKitMol()

In [23]: r
Out[23]: <rdkit.Chem.rdchem.Mol at 0x1048ee600>

In [24]: r.GetNumAtoms()
Out[24]: 2

so I'm not sure what's going on. Perhaps start with some more unit tests on SMILES output from strange things?

[bbuesser]

I think the problem is a missing check in fromAdjacencyList() that rejects wrong or incomplete adjacency lists like the one you mentioned:

1 *4 C 1 0 {2,S}
2    O 0 2 {1,S}

Where did you find it?

[rwest]

We made it up as a test.
Is the problem that we have a Carbon with 1 radical, 1 bond, and 0 lone pairs? (thus 2 valence electrons instead of 8)
Or an Oxygen with 0 radical, 1 bond, and 2 lone pair electrons (thus 3 valence electrons instead of 8).
Would this end up with an 11+ change? Or is it just entirely invalid?

[bbuesser]

I think that is what is happening. fromAdjacencyList() is reading that adjacency list assuming all hydrogen atoms are explicitly present as saturateH=False is standard for that function. Can you print the charge attribute of the two atoms? Also can you try it with fromAdjacencyList(saturateH=True)?

[connie]

This is a deprecated issue. The adjlist should have explicit hydrogens or if it is carbon monoxide it should be a triple bonded adjlist with 1 lone pair on both carbon and oxygen.