Skip to content

Consider treating a hydro prefix on a bridge atom as a no-op (blocks morphine and ~35k names) #307

Open

Description

@guillaume-osmo

Not a bug — OPSIN's current behaviour is defensible — but a rule change that would recover a large, chemically important class of names for what looks like little risk.

A hydro prefix that cites an atom which is saturated by construction is rejected:

hydrogen addition at locant: 13 was requested, but this atom is not unsaturated

Locant 13 of 4,12-methanobenzofuro[3,2-e]isoquinoline is the methano bridge CH2. It is sp3 because the bridge prefix made it so, so citing it in the hydro prefix is redundant rather than contradictory — IUPAC P-25.4 already accounts for bridge saturation in the bridge prefix itself.

This blocks morphine and the whole morphinan class:

(4R,4aR,7S,7aR,12bS)-3-methyl-2,4,4a,7,7a,13-hexahydro-1H-4,12-methanobenzofuro[3,2-e]isoquinoline-7,9-diol
  -> REJECTED   (PubChem CID 5288826, morphine)

same name with '13' removed from the hydro list and hexa -> penta
  -> BQJCRHHNABKAKU-KBQPJGBKSA-N   which is morphine exactly

Scale

35,880 names in PubChem's IUPAC set fail this way, all at locant 13 of that ring system. Deleting the bridge locant recovers 34,740 exact StdInChIKeys (96.8%) and 35,085 correct skeletons (97.8%). It is the second-largest single cause of OPSIN failing to interpret a PubChem systematic name.

That OPSIN's structure-building is already correct

CAS Common Chemistry names morphine as a morphinan, not with the fusion nomenclature, and OPSIN parses the CAS name perfectly in both formats:

(5alpha,6alpha)-7,8-Didehydro-4,5-epoxy-17-methylmorphinan-3,6-diol       BQJCRHHNABKAKU ✓
Morphinan-3,6-diol, 7,8-didehydro-4,5-epoxy-17-methyl-, (5alpha,6alpha)-  BQJCRHHNABKAKU ✓

So this is purely about tolerating a redundant locant in someone else's naming style, not about getting the structure right.

Suggested rule, and the reason for caution

"A hydro prefix naming an atom that is already saturated is a no-op, not an error" — but only where the atom is saturated by construction (a bridge, a spiro atom), not where it was saturated by another hydro prefix in the same name. The narrow version matters: a hydro prefix landing on an atom saturated by an earlier hydro prefix usually signals a numbering mismatch, and silently accepting that would turn a clear rejection into a possibly-wrong structure. Given #300, I assume you would want the narrow reading.

I have not implemented this — the distinction between the two cases is exactly the sort of thing where I would rather defer to you than guess.

How it was found

Auditing OPSIN against PubChem's 123.5M Lexichem-generated IUPAC names. Of 403,090 names still uninterpreted with -s, only 3.9% are grammar failures; 96% parse cleanly and fail during structure building. This is 8.9% of that residue.

Tested with OPSIN 2.9.0, JDK 26 on macOS.

Activity

  1. dan2097 commented on Oct 25, 2026

    @dan2097
    Owner

    This is almost certainly a bug in Lexichem. The name hasn't just got a redundant hydro indication, technically it's missing one as well:
    (4R,4aR,7S,7aR,12bS)-3-methyl-2,4,4a,7,7a,13-hexahydro-1H-4,12-methanobenzofuro[3,2-e]isoquinoline-7,9-diol
    should be
    (4R,4aR,7S,7aR,12bS)-3-methyl-2,3,4,4a,7,7a-hexahydro-1H-4,12-methanobenzofuro[3,2-e]isoquinoline-7,9-diol

    The 3 position is the nitrogen. OPSIN would implicitly add hydrogen here as the atom is surrounded by saturated atoms, but it's still not correct.
    This sort of thing is really tricky as there are some pragmatic benefits to supporting malformed names as they do end up in the literature.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions