implement data model for metabolite identifiers

[hredestig]

• design data-model for metabolite identifiers. Suggestion is to create internal metabolite identifier, and attach all new identifiers to that internal identifier, to make all mappings a 1-hop query (query speed decreases rapidly with the number of hops that need to be made)
  • add indices on chebi
  • unique constraints on 'strong' identifiers e.g. inchi / chebi / smiles / cas..?
• write a proper data-import procedure which adds all identifiers from
  • metanetx (draft implementation existing)
  • use chebi to add to all identifiers
    - annotation: trivial names, mass
    - strong identifiers: inchikey, inchi, with and without hydrogen and charge layer, cas, smiles
    - sync with iloop to make it clear which chebi identifiers are known to iloop
  • own manually curated list of pairs of identifiers
  • merge sets of identifiers to single set if
    - any strong identifier is shared between them.

Would solve #7 and #6