web app (#39)

web app

Author: AdrianM0

src/openclatura/assembly_parts.py

@@ -130,3 +130,5 @@ class AssemblyParts:
     name_token_spans: list[dict] = field(default_factory=list)
     name_rewrite_history: list[dict] = field(default_factory=list)
     stereo_audit_issues: list[str] = field(default_factory=list)
+    reconstruction_audit_status: str | None = None
+    reconstruction_audit_issues: list[str] = field(default_factory=list)

src/openclatura/component_namer.py

@@ -26,6 +26,7 @@
     filter_component_groups_to_parent,
     partition_principal_and_prefix_groups,
 )
+from .reconstruction_audit import audit_component_reconstruction
 from .retained_fused_production import production_retained_fused_parent
 from .special_cases import (
     single_atom_component_name,
@@ -611,6 +612,9 @@ def name_component_again(next_mol: Molecule, next_atoms: set[int], is_substituen
 
     refresh_name_atom_bindings(parts)
     parts.stereo_audit_issues = list(audit_stereochemistry(mol, parts).issues)
+    reconstruction = audit_component_reconstruction(mol, parts)
+    parts.reconstruction_audit_status = reconstruction.status
+    parts.reconstruction_audit_issues = list(reconstruction.issues)
     assert_component_fully_named(mol, state.component_atoms, parts, "<component>")
     name = assemble_parent_name(mol, parts, numbered_path, get_loc, emit_metadata=emit_metadata)
     if emit_metadata:
@@ -629,6 +633,10 @@ def name_component_again(next_mol: Molecule, next_atoms: set[int], is_substituen
             "substituent_count": len(parts.substituents),
             "unsaturation_count": len(parts.unsaturations),
             "stereo_audit_issues": parts.stereo_audit_issues,
+            "reconstruction_audit": {
+                "status": parts.reconstruction_audit_status,
+                "issues": parts.reconstruction_audit_issues,
+            },
             "name_atom_bindings": binding_trace_data(parts.name_atom_bindings, include_emitted_tokens=token_debug),
             "name_token_spans": parts.name_token_spans if token_debug else [],
             "name_rewrite_history": parts.name_rewrite_history,

src/openclatura/graph_io.py

@@ -1,6 +1,7 @@
 """SMILES input and graph component helpers."""
 
 from rdkit import Chem
+from rdkit.Chem import rdCIPLabeler
 
 from .molecule import Molecule
 
@@ -29,31 +30,53 @@ def read_smiles(smiles: str) -> Molecule:
         atom_metadata = _atom_metadata(rdmol)
 
     Chem.AssignStereochemistry(rdmol, force=True, cleanIt=True)
-    chiral_centers = dict(Chem.FindMolChiralCenters(rdmol, includeUnassigned=False))
+    legacy_centers = dict(Chem.FindMolChiralCenters(rdmol, includeUnassigned=False))
+    # The legacy labeler decides *which* centers carry an absolute descriptor:
+    # centers it leaves unassigned are ring-symmetry-dependent ones that the
+    # scoped small-ring / cis-trans fallbacks must handle (OPSIN rejects
+    # absolute R/S there). But its CIP *values* mis-rank ligands in
+    # deep-sphere comparisons (e.g. ring-closure duplicate atoms) and invert
+    # 3-coordinate sulfur centers, so the label itself comes from
+    # rdCIPLabeler, which overwrites the _CIPCode properties.
+    try:
+        rdCIPLabeler.AssignCIPLabels(rdmol)
+    except Exception:
+        pass  # unsanitized input; keep the legacy labels
+    chiral_centers = {}
+    for idx, legacy_code in legacy_centers.items():
+        atom = rdmol.GetAtomWithIdx(idx)
+        code = atom.GetProp("_CIPCode") if atom.HasProp("_CIPCode") else legacy_code
+        if code in ("R", "S"):
+            chiral_centers[idx] = code
 
     for atom in rdmol.GetAtoms():
         stereo = chiral_centers.get(atom.GetIdx())
-        if stereo and atom.GetSymbol() == "S" and atom.GetTotalDegree() == 3:
-            stereo = "R" if stereo == "S" else "S"
         raw_stereo = _raw_tetrahedral_stereo(atom) if not stereo else None
         mol.add_atom(
             symbol=atom.GetSymbol(),
             idx=atom.GetIdx(),
             charge=atom.GetFormalCharge(),
             stereo=stereo,
             raw_stereo=raw_stereo,
+            cip_code=atom.GetProp("_CIPCode") if atom.HasProp("_CIPCode") else None,
             is_aromatic=atom_metadata[atom.GetIdx()]["is_aromatic"],
             explicit_h_count=atom_metadata[atom.GetIdx()]["explicit_h_count"],
             total_h_count=atom_metadata[atom.GetIdx()]["total_h_count"],
         )
 
     for bond in rdmol.GetBonds():
-        stereo = None
-        st = bond.GetStereo()
-        if st == Chem.rdchem.BondStereo.STEREOE:
-            stereo = "E"
-        elif st == Chem.rdchem.BondStereo.STEREOZ:
-            stereo = "Z"
+        # rdCIPLabeler rewrites bond enums to STEREOCIS/STEREOTRANS but stamps
+        # the authoritative E/Z label as _CIPCode; fall back to the enums when
+        # the labeler did not run.
+        stereo = bond.GetPropsAsDict().get("_CIPCode")
+        if stereo not in ("E", "Z"):
+            st = bond.GetStereo()
+            if st == Chem.rdchem.BondStereo.STEREOE:
+                stereo = "E"
+            elif st == Chem.rdchem.BondStereo.STEREOZ:
+                stereo = "Z"
+            else:
+                stereo = None
 
         in_small_ring = any(bond.IsInRingSize(i) for i in range(3, 8))
 

src/openclatura/molecule.py

@@ -13,6 +13,7 @@ class Atom:
     isotope: int | None = None
     stereo: str | None = None  # 'R' or 'S'
     raw_stereo: str | None = None  # RDKit tetrahedral tag when CIP is unavailable: 'CW' or 'CCW'
+    cip_code: str | None = None  # full-molecule rdCIPLabeler code ('R'/'S'/'r'/'s'), also set on dependent centers
     is_aromatic: bool = False
     explicit_h_count: int = 0
     total_h_count: int = 0
@@ -179,6 +180,7 @@ def add_atom(
         stereo: str | None = None,
         *,
         raw_stereo: str | None = None,
+        cip_code: str | None = None,
         is_aromatic: bool = False,
         explicit_h_count: int = 0,
         total_h_count: int = 0,
@@ -193,6 +195,7 @@ def add_atom(
             charge=charge,
             stereo=stereo,
             raw_stereo=raw_stereo,
+            cip_code=cip_code,
             is_aromatic=is_aromatic,
             explicit_h_count=explicit_h_count,
             total_h_count=total_h_count,