Deprecate fix_CO in the from_xyz workflow and add it to fix_mol

rdmc/fix.py

@@ -17,48 +17,52 @@
     rdChemReactions.ReactionFromSmarts(
         "[O+0-0v2X1:1]=[C+0-0v2X1:2]>>[O+1v3X1:1]#[C-1v3X1:2]"
     ),
-    # Remedy 2 - Oxygen Molecule: O=O to [O]-[O]
+    # Remedy 2 - Carbon monoxide: [C]=O to [C-]#[O+]
+    rdChemReactions.ReactionFromSmarts(
+        "[O+0-0v3X1:1]#[C+0-0v3X1:2]>>[O+1v3X1:1]#[C-1v3X1:2]"
+    ),
+    # Remedy 3 - Oxygen Molecule: O=O to [O]-[O]
     rdChemReactions.ReactionFromSmarts(
         "[O+0-0v2X1:1]=[O+0-0v2X1:2]>>[O+0-0v1X1:1]-[O+0-0v1X1:2]"
     ),
-    # Remedy 3 - isocyanide: R[N]#[C] to R[N+]#[C-]
+    # Remedy 4 - isocyanide: R[N]#[C] to R[N+]#[C-]
     rdChemReactions.ReactionFromSmarts(
         "[N+0-0v4X2:1]#[C+0-0v3X1:2]>>[N+v4X2:1]#[C-v3X1:2]"
     ),
-    # Remedy 4 - azide: RN=N=[N] to RN=[N+]=[N-]
+    # Remedy 5 - azide: RN=N=[N] to RN=[N+]=[N-]
     rdChemReactions.ReactionFromSmarts(
         "[N+0-0v3X2:1]=[N+0-0v4X2:2]=[N+0-0v2X1:3]>>[N+0-0v3X2:1]=[N+1v4X2:2]=[N-1v2X1:3]"
     ),
-    # Remedy 5 - amine oxide: RN(R)(R)-O to R[N+](R)(R)-[O-]
+    # Remedy 6 - amine oxide: RN(R)(R)-O to R[N+](R)(R)-[O-]
     rdChemReactions.ReactionFromSmarts(
         "[N+0-0v4X4:1]-[O+0-0v1X1:2]>>[N+1v4X4:1]-[O-1v1X1:2]"
     ),
-    # Remedy 6 - amine radical: R[C](R)-N(R)(R)R to R[C-](R)-[N+](R)(R)R
+    # Remedy 7 - amine radical: R[C](R)-N(R)(R)R to R[C-](R)-[N+](R)(R)R
     rdChemReactions.ReactionFromSmarts(
         "[N+0-0v4X4:1]-[C+0-0v3X3:2]>>[N+1v4X4:1]-[C-1v3X3:2]"
     ),
-    # Remedy 7 - amine radical: RN(R)=C to RN(R)-[C]
+    # Remedy 8 - amine radical: RN(R)=C to RN(R)-[C]
     rdChemReactions.ReactionFromSmarts(
         "[N+0-0v4X3:1]=[C+0-0v4X3:2]>>[N+0-0v3X3:1]-[C+0-0v3X3:2]"
     ),
-    # Remedy 8 - quintuple C bond, usually due to RC(=O)=O: R=C(R)=O to R=C(R)-[O]
+    # Remedy 9 - quintuple C bond, usually due to RC(=O)=O: R=C(R)=O to R=C(R)-[O]
     rdChemReactions.ReactionFromSmarts(
         "[C+0-0v5X3:1]=[O+0-0v2X1:2]>>[C+0-0v4X3:1]-[O+0-0v1X1:2]"
     ),
-    # Remedy 9 - sulphuric bi-radicals: R[S](R)(-[O])-[O] to R[S](R)(=O)(=O)
+    # Remedy 10 - sulphuric bi-radicals: R[S](R)(-[O])-[O] to R[S](R)(=O)(=O)
     rdChemReactions.ReactionFromSmarts(
         "[S+0-0v4X4:1](-[O+0-0v1X1:2])-[O+0-0v1X1:3]>>[S+0-0v6X4:1](=[O+0-0v2X1:2])=[O+0-0v2X1:3]"
     ),
-    # Remedy 10 - Triazinane: C1=N=C=N=C=N=1 to c1ncncn1
+    # Remedy 11 - Triazinane: C1=N=C=N=C=N=1 to c1ncncn1
     rdChemReactions.ReactionFromSmarts(
         "[C+0-0v5X3:1]1=[N+0-0v4X2:2]=[C+0-0v5X3:3]=[N+0-0v4X2:4]=[C+0-0v5X3:5]=[N+0-0v4X2:6]=1"
         ">>[C+0-0v5X3:1]1[N+0-0v4X2:2]=[C+0-0v5X3:3][N+0-0v4X2:4]=[C+0-0v5X3:5][N+0-0v4X2:6]=1"
     ),
-    # Remedy 11 - peroxide biradical: R[C](R)O[O] to R[C+](R)O[O-]
+    # Remedy 12 - peroxide biradical: R[C](R)O[O] to R[C+](R)O[O-]
     rdChemReactions.ReactionFromSmarts(
         "[C+0-0v3X3:1]-[O+0-0v2X2:2]-[O+0-0v1X1:3]>>[C+1v3X3:1]-[O+0-0v2X2:2]-[O-1v1X1:3]"
     ),
-    # Remedy 12 - conjugate peroxide biradical: [C]-C=C(R)O[O] to C=C-[C+](R)O[O-]
+    # Remedy 13 - conjugate peroxide biradical: [C]-C=C(R)O[O] to C=C-[C+](R)O[O-]
     rdChemReactions.ReactionFromSmarts(
         "[C+0-0v3X3:1]-[C:2]=[C+0-0v4X3:3]-[O+0-0v2X2:4]-[O+0-0v1X1:5]>>[C+0-0v4X3:1]=[C:2]-[C+1v3X3:3]-[O+0-0v2X2:4]-[O-1v1X1:5]"
     ),

rdmc/mol.py

@@ -605,7 +605,6 @@ def FromXYZ(
         xyz: str,
         backend: str = "openbabel",
         header: bool = True,
-        correctCO: bool = True,
         sanitize: bool = True,
         embed_chiral: bool = False,
         **kwargs,
@@ -619,7 +618,6 @@ def FromXYZ(
                            Currently, we only support ``'openbabel'`` and ``'jensen'``.
             header (bool, optional): If lines of the number of atoms and title are included.
                                      Defaults to ``True.``
-            correctCO (bool, optional): Whether to correct the CO bond as "[C-]#[O+]". Defaults to ``True``.
             sanitize (bool): Sanitize the RDKit molecule during conversion. Helpful to set it to ``False``
                              when reading in TSs. Defaults to ``True``.
             embed_chiral: ``True`` to embed chiral information. Defaults to ``True``.
@@ -642,7 +640,7 @@ def FromXYZ(
 
         # Openbabel support read xyz and perceive atom connectivities
         if backend.lower() == "openbabel":
-            obmol = parse_xyz_by_openbabel(xyz, correct_CO=correctCO)
+            obmol = parse_xyz_by_openbabel(xyz)
             rdmol = cls.FromOBMol(obmol, sanitize=sanitize)
             if embed_chiral:
                 rdmol.AssignStereochemistryFrom3D()
@@ -652,7 +650,7 @@ def FromXYZ(
         # provides an approach to convert xyz to mol
         elif backend.lower() == "jensen":
             mol = parse_xyz_by_jensen(
-                xyz, correct_CO=correctCO, embed_chiral=embed_chiral, **kwargs
+                xyz, embed_chiral=embed_chiral, **kwargs
             )
             return cls(mol)
 
@@ -690,7 +688,6 @@ def FromFile(
         path: str,
         backend: str = "openbabel",
         header: bool = True,
-        correctCO: bool = True,
         removeHs: bool = False,
         sanitize: bool = True,
         sameMol: bool = False,
@@ -723,7 +720,6 @@ def FromFile(
                 xyz,
                 backend=backend,
                 header=header,
-                correctCO=correctCO,
                 sanitize=sanitize,
                 **kwargs,
             )

rdmc/utils.py

@@ -56,11 +56,6 @@
     "[!$(*#*)&!D1!H3]-&!@[!$(*#*)&!D1&!H3]"
 )
 
-# When perceiving molecules, openbabel will always perceive carbon monoxide as [C]=O
-# Needs to correct it by [C-]#[O+]
-CO_OPENBABEL_PATTERN = ob.OBSmartsPattern()
-CO_OPENBABEL_PATTERN.Init("[Cv2X1]=[OX1]")
-
 # Carbene, nitrene, and atomic oxygen templates. RDKit and Openbabel have difficulty
 # distinguish their multiplicity when input as SMILES or XYZ
 CARBENE_PATTERN = Chem.MolFromSmarts("[Cv0,Cv1,Cv2,Nv0,Nv1,Ov0]")
@@ -457,38 +452,12 @@ def set_obmol_coords(obmol: ob.OBMol, coords: np.array):
         atom.SetVector(ob.vector3(*coords[atom_idx].tolist()))
 
 
-def fix_CO_openbabel(obmol: "Openbabel.OBMol", correct_CO: bool = True):
-    """
-    Fix the CO perception issue for openbabel molecule.
-
-    Args:
-        obmol (Openbabel.OBMol): The Openbabel molecule instance.
-        correct_CO (bool, optional): Whether to fix this issue. Defaults to True.
-    """
-    if not correct_CO:
-        return
-    CO_OPENBABEL_PATTERN.Match(obmol)
-    for pair in CO_OPENBABEL_PATTERN.GetUMapList():
-        obmol.GetBond(*pair).SetBondOrder(3)
-        for idx in pair:
-            atom = obmol.GetAtom(idx)
-            if atom.GetAtomicNum() == 6:
-                atom.SetSpinMultiplicity(0)
-                atom.SetFormalCharge(-1)
-            elif atom.GetAtomicNum() == 8:
-                atom.SetSpinMultiplicity(0)
-                atom.SetFormalCharge(+1)
-
-
-def parse_xyz_by_openbabel(xyz: str, correct_CO: bool = True):
+def parse_xyz_by_openbabel(xyz: str):
     """
     Perceive a xyz str using openbabel and generate the corresponding OBMol.
 
     Args:
         xyz (str): A str in xyz format containing atom positions.
-        correctCO (bool, optional): It is known that openbabel will parse carbon monoxide
-                                    as [C]=O instead of [C-]#[O+]. This function contains
-                                    a patch to correct that. Defaults to ``True``.
 
     Returns:
         ob.OBMol: An openbabel molecule from the xyz
@@ -522,9 +491,6 @@ def parse_xyz_by_openbabel(xyz: str, correct_CO: bool = True):
         ):
             obatom.SetSpinMultiplicity(2)
 
-    # Correct [C]=O to [C-]#[O+]
-    fix_CO_openbabel(obmol, correct_CO=correct_CO)
-
     return obmol
 
 
@@ -620,7 +586,6 @@ def parse_xyz_by_jensen(
     allow_charged_fragments: bool = False,
     use_huckel: bool = False,
     embed_chiral: bool = True,
-    correct_CO: bool = True,
     use_atom_maps: bool = False,
     force_rdmc: bool = False,
     **kwargs,
@@ -634,10 +599,6 @@ def parse_xyz_by_jensen(
         allow_charged_fragments: ``True`` for charged fragment, ``False`` for radical. Defaults to False.
         use_huckel: ``True`` to use extended Huckel bond orders to locate bonds. Defaults to False.
         embed_chiral: ``True`` to embed chiral information. Defaults to True.
-        correctCO (bool, optional): Defaults to ``True``.
-                                    In order to get correct RDKit molecule for carbon monoxide
-                                    ([C-]#[O+]), allow_charged_fragments should be forced to ``True``.
-                                    This function contains a patch to correct that.
         use_atom_maps(bool, optional): ``True`` to set atom map numbers to the molecule. Defaults to ``False``.
         force_rdmc (bool, optional): Defaults to ``False``. In rare case, we may hope to use a tailored
                                      version of the Jensen XYZ parser, other than the one available in RDKit.
@@ -657,7 +618,6 @@ def parse_xyz_by_jensen(
             use_huckel=use_huckel,
             embed_chiral=embed_chiral,
             use_atom_maps=use_atom_maps,
-            correct_CO=correct_CO,
         )
 
     # Version >= 2022.09.1
@@ -684,13 +644,6 @@ def parse_xyz_by_jensen(
         useHueckel=use_huckel,
         charge=charge,
     )
-    # A force correction for CO
-    if (
-        correct_CO
-        and mol.GetNumAtoms() == 2
-        and {atom.GetAtomicNum() for atom in mol.GetAtoms()} == {6, 8}
-    ):
-        allow_charged_fragments = True
     rdDetermineBonds.DetermineBondOrders(
         mol,
         charge=charge,

test/test_fix.py

@@ -28,6 +28,7 @@ def test_fix_sanitize_ok(smi, exp_smi):
 @pytest.mark.parametrize(
     "smi, exp_smi",
     [
+        ("[C]#[O]", "[C-]#[O+]"),
         ("[NH3][O]", "[NH3+][O-]"),
         ("[CH2][NH3]", "[CH2-][NH3+]"),
         ("[C]#[NH]", "[C-]#[NH+]"),

test/test_utils.py

@@ -22,10 +22,6 @@
 
 
 @pytest.fixture(params=[
-    ('[C-]#[O+]', """2
-
-C      0.559061    0.000000    0.000000
-O     -0.559061    0.000000    0.000000"""),
     ('C', """5
 
 C      0.005119   -0.010620    0.006014