how do we store assignment

[kjappelbaum]

• do we want to duplicate the structure all the time or do we store only a reference
• how do we store the atom(s) reference: indices of a molfile?
• smiles with marked atom(s)

[djeanner]

For NMReDATA we used the index in the mol file. I heard that if a "external" library reader might changes the order of atoms - but we could argue that the mol reader should just be able to keep track of the index. But using a marked smile would allow to give any individual assignment a serious ontological basis. One could say that we developp the latter for report - and see if it flies - but (try) to rely only on the atom index.

[kjappelbaum]

The assignment types now assumes a SMILES and a list of indices w.r.t to this SMILES a bit like https://docs.chemaxon.com/display/docs/chemaxon-extended-smiles-and-smarts-cxsmiles-and-cxsmarts.md or how Luc stores NMR assignments at the moment

[kjappelbaum]

I'm moving assignment now to MolecularAssignment since peaks might have quite different assignments if we think beyond structure elucidation applications:

• in XRD, we rather assign to some hkl planes
• in UV VIS you might assign to some substructure with heuristic or to some specific transition between energy levels, especially when you do high-resolution spectroscopy
• In TGA it is rather an ensemble property, where a peak might be due to loss of some solvent

latter is something one would probably rather describe with the annotation

[kjappelbaum]

the only issue with SMILES for molecular assignment is that they work underwhelmingly for organometallic compounds

[kjappelbaum]

but somehow i prefer molfile over SMILES, because it involves much less interpretation (i do not like the implicit H). And if we decide to repeat the information to avoid corruption/interpretation errors, then we probably do not care to also have the coordinates there.