Add function to check whether a compound SMILES has atom-mapping info (…

…#25)

src/rxn/chemutils/miscellaneous.py

@@ -317,3 +317,28 @@ def merge_reactions(*reactions: ReactionEquation) -> ReactionEquation:
         agents.extend(reaction.agents)
         products.extend(reaction.products)
     return ReactionEquation(reactants=reactants, agents=agents, products=products)
+
+
+def mol_has_atom_mapping(mol: Mol) -> bool:
+    """
+    Whether at least one atom of an RDKit Mol contains an atom map number.
+
+    Args:
+        mol: RDKit Mol.
+    """
+    atom: Atom
+    for atom in mol.GetAtoms():
+        if atom.GetAtomMapNum() != 0:
+            return True
+    return False
+
+
+def smiles_has_atom_mapping(smiles: str) -> bool:
+    """
+    Whether at least one atom of a compound SMILES contains an atom map number.
+
+    Args:
+        smiles: compound SMILES.
+    """
+    mol = smiles_to_mol(smiles, sanitize=False)
+    return mol_has_atom_mapping(mol)

tests/test_miscellaneous.py

@@ -15,6 +15,7 @@
     merge_reactions,
     remove_chiral_centers,
     remove_double_bond_stereochemistry,
+    smiles_has_atom_mapping,
     sort_any,
 )
 from rxn.chemutils.reaction_equation import ReactionEquation
@@ -260,3 +261,20 @@ def test_merge_reactions() -> None:
     assert merge_reactions(r1, r2, r3, r4) == ReactionEquation(
         ["A", "B", "F"], ["C", "G", "H"], ["D", "I"]
     )
+
+
+def test_smiles_has_atom_mapping() -> None:
+    assert smiles_has_atom_mapping("[CH3:9][CH:8]([CH3:10])c1ccccc1")
+    assert smiles_has_atom_mapping("CCO[CH3:9]")
+    assert smiles_has_atom_mapping("C[CH3:9].O")
+
+    assert not smiles_has_atom_mapping("CC")
+    assert not smiles_has_atom_mapping("C[CH3]")
+    assert not smiles_has_atom_mapping("C[12CH3]")
+
+    # Note: does not care about valence, but SMILES must be valid
+    # invalid valence -> call succeeds:
+    assert smiles_has_atom_mapping("CF(C)CC[C:9]")
+    # invalid SMILES -> call raises:
+    with pytest.raises(InvalidSmiles):
+        _ = canonicalize_any("[CH3:9]C(")