build(deps): bump chematic from 0.16.0 to 0.19.0

[dependabot]

Bumps chematic from 0.16.0 to 0.19.0.

Release notes

Sourced from chematic's releases.

v0.19.0

— 2026-08-23

Added — chematic-chem (Tautomer & Parent Identity round 2C, ROADMAP.md Phase 2)

• canonical_tautomer/tautomer_parent now canonicalize the aromatic lactam/lactim class (e.g. 2-pyridone O=c1cccc[nH]1 / 2-hydroxypyridine Oc1ccccn1) that was previously not invariant across input tautomer spelling. Implemented as a new directional step (apply_exocyclic_lactam_shift_tracked in tautomer.rs) — an exocyclic O donor/acceptor across an aromatic ring atom at odd ring-path distance — rather than fed into the existing score-ranked candidate pool, which measurably selects the wrong (lactim) tautomer for this class. See docs/rfcs/tautomer_parent_identity_phase2_rfc.md section 4.4a for the full mechanism and the measurement behind that design choice.
• Measured, not assumed: the shift fires correctly on all 5 of the RFC's design molecules, but end-to-end convergence holds for only 3 of 5 (2-pyridone, 4-pyridone, uracil) — cytosine and guanine do not converge end-to-end, because their carbonyl carbon is flanked by two ring nitrogens (RFC section 1.7, a second, distinct, out-of-scope tautomer defect found while fixing this one). Correction (round 2C-N, diagnosis-only, no production-code change): hypoxanthine was originally reported here as a clean holdout that generalized the fix — this was wrong. Re-measurement found hypoxanthine has the identical two-candidate ring-N-H ambiguity as cytosine/guanine (RFC section 1.7); the original holdout check happened to test only one of its two valid keto spellings. Closing section 1.7 for all three molecules needs ring-internal N-position normalization, deliberately not part of this change (same reasoning as excluding the nitroso/oxime defect, section 1.6). Neither section 1.6 nor section 1.7 is resolved as of this release; Phase 2 (Tautomer & Parent Identity) is not complete, and TautomerScoringConfig/ Python/WASM Parent-API bindings remain unimplemented.
• New negative controls confirmed unaffected: phenol, anisole, aniline, pyridine N-oxide, plain amides, and — the two checks that specifically probe the fix's scope boundary — 3-hydroxypyridine (excluded: even/meta ring distance, no valid Kekulé path) and 4-/2-aminopyridine (excluded: the analogous N-acceptor case is evidenced-out-of-scope, not swept in by generalizing the acceptor element).
• Isotope labels and remote stereocenters are confirmed preserved through the shift (atom-level checks, not string comparison).
• Hardened after review, before merge: the matcher is fail-closed — bridge must be a neutral aromatic carbon (not any aromatic element), donor a neutral, degree-1 oxygen with exactly one transferable H, and acceptor a neutral, degree-2 (pyridine-type, valence-compatible after +1H) aromatic nitrogen. Every candidate is verified post-generation against a full atom/bond invariant check (nothing changes except the one H-count pair and the one bond order) before being accepted, and the candidate tie-break cross-checks an independent fingerprint against canonical SMILES rather than trusting the SMILES string alone. Additional confirmed-excluded negative controls: a charged/3-connected pyridinium

... (truncated)

Changelog

Sourced from chematic's changelog.

Fixed — documentation (correction to a v0.19.0 changelog entry)

• v0.19.0's own benchmark-refresh entry below states 3D conformer quality was corrected to match docs/rdkit-migration.md's "Experimental, no RMSD/TFD comparison exists" characterization — that characterization was itself wrong, found during this round's OpenEye/materials-science competitive audit. validation/results/mmff94_bci_gap_227_phase2_ report.md already measures RMSD (mean 1.685 Å) and TFD (mean 0.2228) against RDKit's ETKDGv3+MMFF94 on the project's 265-molecule corpus. docs/benchmark.md, docs/rdkit-migration.md, and the same stale "15 SMARTS rules" pKa-rule-count claim (actual: 23 rules, 6 acid + 17 base, crates/chematic-chem/src/pka.rs) copied across README.md/ README_ja.md/docs/rdkit-comparison.md/docs/rdkit_cheatsheet.md/ docs/benchmark.md are corrected. Not rewriting v0.19.0's already- released changelog text itself, per this project's convention of never silently editing a shipped version's own historical record — see the entry below for what v0.19.0 originally said.

Known limitations in this release

• generate_coords (the legacy 3D placement engine) is not routed to the new connectivity-ordered engine — no existing caller's default behavior changed. See "Added — connectivity-ordered 3D coordinate generation" above.
• Issue #210's remaining 4 residual molecules (naproxen_S, ibuprofen_S, testosterone, cholesterol) are still unfixed via the UFF-rescue bridge specifically — stereo_safe (this release's own headline feature, above) already resolves them through embed_pipeline_v2's own path.
• Issue #390 (a single-molecule coupled/shared-bond E/Z correctness residual, independent of every fix in this release) remains open.
• The best-of-10 conformer benchmark establishes that embed_ensemble_v2 works robustly at scale; it does not establish RDKit conformer- selection parity — see "Added — benchmark" above for the exact distinction.
• This release's evidence is drawn from measurements already taken during development (29-molecule × 5-seed stereo sweep, 33-molecule connectivity differential, 265-molecule best-of-10 benchmark, 58-molecule UFF-rescue sweep, the 9.47M-compound downstream identity investigation) — no fresh full-corpus re-measurement was run solely for this release, per this project's own "minimize heavy measurements" policy.

[0.19.0] — 2026-08-23

Added — chematic-chem (Tautomer & Parent Identity round 2C, ROADMAP.md Phase 2)

• canonical_tautomer/tautomer_parent now canonicalize the aromatic lactam/lactim class (e.g. 2-pyridone O=c1cccc[nH]1 / 2-hydroxypyridine Oc1ccccn1) that was previously not invariant across input tautomer spelling. Implemented as a new directional step (apply_exocyclic_lactam_shift_tracked in tautomer.rs) — an exocyclic

... (truncated)

Commits

• e6ebff3 Merge pull request #368 from kent-tokyo/release-prep/v0.19.0
• 448e08f release: prepare v0.19.0 -- benchmark/validation refresh, version bump
• 24a9239 Merge pull request #367 from kent-tokyo/diag/tautomer-ring-nh-dual-flank-sele...
• 7ebb1a1 fix(chem): harden round 2C-N §1.7 diagnosis after direct pre-merge review
• 81c0d37 diag(chem): full diagnosis + design for RFC section 1.7 ring-N-H dual-flank r...
• c5b981a Merge pull request #366 from kent-tokyo/fix/redact-local-paths-from-validatio...
• ff9cba6 Merge pull request #365 from kent-tokyo/feat/tautomer-parent-identity-phase2-...
• 884f60c docs(validation): sync round 2C hardening into validation/README.md
• ebe7af8 fix(validation): redact local machine paths from committed diagnostic artifacts
• a0fdf0a fix(chematic-chem): harden round 2C-2's exocyclic lactam/lactim shift (post-r...
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

[dependabot]

Labels

The following labels could not be found: dependencies, security. Please create them before Dependabot can add them to a pull request.

Please fix the above issues or remove invalid values from dependabot.yml.

[dependabot]

OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting @dependabot ignore this major version or @dependabot ignore this minor version. You can also ignore all major, minor, or patch releases for a dependency by adding an ignore condition with the desired update_types to your config file.

If you change your mind, just re-open this PR and I'll resolve any conflicts on it.