Support writing CX extensions in reactions

[rachelnwalker]

We are hoping to get support for writing CX extensions when writing to reaction SMILES/SMARTS. This is just a draft, I wanted to see if we think this would work. At the very least it would be nice to be able to read/write atom labels and stereo information.

Currently, there are a few things that aren’t supported

• Link nodes
• Coordinate bonds
• Writing to CX smiles (worried about atom/fragment reordering?)

I’m also a little confused about the substance group writing — it seems like the atoms are correct, however there are some extra commas and colons written compared to the input. RDKit seems to parse this correctly but I'm not sure if this is expected/a problem? For example:

Input SMARTS: [CH3:6][O:5][CH:3](-*)[O:2]-*>>[CH3:6][NH:5][CH:3](-*)[O:2]-* |$;;;star_e;;star_e;;;;star_e;;star_e$,SgD:1,0:foo:bar::::,SgD:7,6:foo:baz::::,Sg:n:4,2,1,0::ht,Sg:n:10,8,7,6::ht,SgH:2:0,3:1|

Roundtrip output SMARTS: [C&H3:6][O:5][C&H1:3](-*)[O:2]-*>>[C&H3:6][N&H1:5][C&H1:3](-*)[O:2]-* |$;;;star_e;;star_e;;;;star_e;;star_e$,SgD:1,0:foo:bar::::,,SgD:7,6:foo:baz::::,,,Sg:n:4,2,1,0::ht:::,,Sg:n:10,8,7,6::ht:::,SgH:3:1.1|

If this seems like a reasonable approach, I can try to extend this to work for SMILES as well.

[ricrogz]

I think this approach makes sense.

[ricrogz]

This is duplicated, it's declared again inside the loop.

[ricrogz]

Are you doing this to remove the RGroup labels from the output?

[greglandrum]

My two cents on the approach: it seems like it would be more robust and efficient (because you won't have to create the new object and copy all the atoms and bonds) to have a new getCXExtensions() which takes a vector of mol pointers (and to modify the current getCXExtensions() to call that one) and to modify each of the functions that calls to do the same.