PFASGroups 3.2.2__tar.gz → 3.4.0__tar.gz

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/HalogenGroupModel.py

@@ -154,11 +154,11 @@ class HalogenGroup():
                 # No metadata available to infer component SMARTS
                 return
             # prepare halogens (accepts list, str or None)
-            if isinstance(self.componentHalogens,list) and len(set(self.componentHalogens).intersection(['F','Cl','Br','I'])) == 0:
+            if isinstance(self.componentHalogens,list) and len(set(self.componentHalogens).intersection(['F','Cl','Br','I','H'])) == 0:
                 raise ValueError(f"Invalid componentHalogens for HalogenGroup '{self.name}' (ID: {self.id})")
             if self.componentHalogens is None:
                 self.componentHalogens = ['F','Cl','Br','I']
-            elif isinstance(self.componentHalogens, str) and self.componentHalogens in ['F','Cl','Br','I']:
+            elif isinstance(self.componentHalogens, str) and self.componentHalogens in ['F','Cl','Br','I','H']:
                 self.componentHalogens = [self.componentHalogens]
             if self.componentSaturation is None:
                 self.componentSaturation = ['per','poly']
@@ -226,9 +226,9 @@ class HalogenGroup():
         constraints : dict
             Component constraints with optional keys:
 
-            * ``'gte'`` – ``{element: min_count}``; component must have at least
+            * ``'gte'`` - ``{element: min_count}``; component must have at least
               *min_count* atoms of *element*.
-            * ``'exclude'`` – list of element symbols that must be absent from
+            * ``'exclude'`` - list of element symbols that must be absent from
               the component.
 
         Returns
@@ -578,13 +578,12 @@ class HalogenGroup():
         if len(augmented_matched_components) == 0:
             return 0, [], 0, []
 
-        # Get all component sizes from all path types
-        all_components = list(set([comp for comps in augmented_matched_components for comp in comps]))
-        component_sizes = [len(x) for x in all_components]
+        # Get all component sizes from matched component dicts
+        component_sizes = [comp.get('size', 0) for comp in augmented_matched_components]
 
         self.all_matches = []  # Clear matches after use
         self.component_specific_extra_atoms = []
-        return max([0] + component_sizes), component_sizes, len(all_components), augmented_matched_components
+        return max([0] + component_sizes), component_sizes, len(augmented_matched_components), augmented_matched_components
 
     def find_aryl_components(self,mol, component_solver=None, **kwargs):
         """Find aryl components in a molecule with comprehensive metrics."""

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/PFASEmbeddings.py

@@ -198,7 +198,7 @@ def _ansi(text: str, *codes: str) -> str:
 
 
 # ---------------------------------------------------------------------------
-# Molecule-highlight colour palettes (RGB float triples, 0–1)
+# Molecule-highlight colour palettes (RGB float triples, 0-1)
 # ---------------------------------------------------------------------------
 
 # Highlight colours by halogen element — mapped to the project colour palette
@@ -311,7 +311,7 @@ def _get_group_info() -> Dict[int, Dict[str, str]]:
 
 # Groups always excluded from the non-OECD category label (perhalogenated /
 # polyhalogenated alkyl catch-alls that add no structural specificity).
-_CLASSIFY_EXCLUDED_IDS: frozenset = frozenset({51, 52})
+_CLASSIFY_EXCLUDED_IDS: frozenset = frozenset({34,35})  # "Perhalogenated alkyl" and "Polyhalogenated alkyl"
 
 # Name subsumption for non-OECD classification: when a more-specific group
 # name (key) is present, every name in its list is suppressed.
@@ -642,11 +642,19 @@ class MatchView(dict):
 
     @property
     def group_id(self) -> Optional[int]:
-        return self.get("id") if self.is_group else None
+        # Keep legacy semantics for HalogenGroup while also exposing wildcard
+        # ids through the same convenience accessor.
+        if self.is_group or self.get("type") == "WildcardGroup":
+            return self.get("id")
+        return None
 
     @property
     def group_name(self) -> Optional[str]:
-        return self.get("group_name") if self.is_group else None
+        # Keep legacy semantics for HalogenGroup while also exposing wildcard
+        # group names through the same convenience accessor.
+        if self.is_group or self.get("type") == "WildcardGroup":
+            return self.get("group_name")
+        return None
 
     @property
     def components(self) -> List[ComponentView]:
@@ -2798,11 +2806,11 @@ class PFASEmbeddingSet(list):
         Parameters
         ----------
         color_by : None | ``'top_group'`` | list of str
-            * ``None`` – return ``(None, None, None)``; callers fall back to a
+            * ``None`` - return ``(None, None, None)``; callers fall back to a
               single default colour.
-            * ``'top_group'`` – derive one label per molecule from the
+            * ``'top_group'`` - derive one label per molecule from the
               highest ``match_count`` ``HalogenGroup`` match.
-            * list of str – use directly as per-molecule labels (must be the
+            * list of str - use directly as per-molecule labels (must be the
               same length as ``self``).
         """
         import matplotlib.pyplot as plt

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/__init__.py

@@ -3,7 +3,7 @@ from .HalogenGroupModel import HalogenGroup
 from .PFASDefinitionModel import PFASDefinition
 from .ComponentsSolverModel import ComponentsSolver
 from .core import rdkit_disable_log, HALOGEN_GROUPS_FILE
-from .parser import parse_smiles, parse_mols, parse_mol, parse_groups_in_mol, parse_from_database, setup_halogen_groups_database, load_HalogenGroups
+from .parser import parse_smiles, parse_mols, parse_mol, parse_groups_in_mol, parse_from_database, setup_halogen_groups_database, load_HalogenGroups, _match_wildcard_groups_in_mol
 # PFASFingerprint: convenience alias for parse_smiles — returns a PFASEmbeddingSet
 PFASFingerprint = parse_smiles
 from .draw_mols import plot_mol, plot_mols, plot_HalogenGroups
@@ -16,6 +16,6 @@ from .fragmentation import generate_degradation_products
 from .PFASEmbeddings import PFASEmbedding, PFASEmbeddingSet, EmbeddingArray, ResultsModel, MoleculeResult, generate_fingerprint
 from .prioritise import prioritise_molecules, prioritize_molecules, get_priority_statistics
 __version__ = "3.2.0"
-__all__ = ['HalogenGroup', 'PFASDefinition', 'parse_smiles', 'parse_mols','parse_mol', 'parse_groups_in_mol', 'parse_from_database', 'setup_halogen_groups_database', 'plot_HalogenGroups', 'plot_mol','plot_mols', 'FINGERPRINT_PRESETS', 'PFASFingerprint', 'generate_fingerprint', 'get_compiled_componentSMARTSs', 'get_componentSMARTSs', 'get_HalogenGroups', 'get_compiled_HalogenGroups', 'get_compiled_PFASGroups', 'get_PFASDefinitions' ,'ComponentsSolver', 'generate_homologues', 'generate_degradation_products',"rdkit_disable_log","load_HalogenGroups", "HALOGEN_GROUPS_FILE"]
+__all__ = ['HalogenGroup', 'PFASDefinition', 'parse_smiles', 'parse_mols','parse_mol', 'parse_groups_in_mol', 'parse_from_database', 'setup_halogen_groups_database', 'plot_HalogenGroups', 'plot_mol','plot_mols', 'FINGERPRINT_PRESETS', 'PFASFingerprint', 'generate_fingerprint', 'get_compiled_componentSMARTSs', 'get_componentSMARTSs', 'get_HalogenGroups', 'get_compiled_HalogenGroups', 'get_compiled_PFASGroups', 'get_PFASDefinitions' ,'ComponentsSolver', 'generate_homologues', 'generate_degradation_products',"rdkit_disable_log","load_HalogenGroups", "HALOGEN_GROUPS_FILE", "_match_wildcard_groups_in_mol"]
 __all__.extend(['PFASEmbedding', 'PFASEmbeddingSet', 'EmbeddingArray', 'ResultsModel', 'MoleculeResult', 'prioritise_molecules', 'prioritize_molecules', 'get_priority_statistics'])
 __all__.extend(['HomologueSeries', 'HomologueEntry'])

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/cli.py

@@ -47,7 +47,7 @@ Examples:
   # List available path types
   PFASGroups list-paths
 
-Note: Use get_componentSMARTSs() and get_PFASGroups() in Python to extend defaults.
+Note: Use get_componentSMARTSs() and get_HalogenGroups() in Python to extend defaults.
         """
     )
 
@@ -186,7 +186,7 @@ Note: Use get_componentSMARTSs() and get_PFASGroups() in Python to extend defaul
     # List groups command
     list_parser = subparsers.add_parser(
         'list-groups',
-        help='List available PFAS groups (use in Python to extend with get_PFASGroups)'
+        help='List available PFAS groups (use in Python to extend with get_HalogenGroups)'
     )
     list_parser.add_argument(
         '-o', '--output',

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/core.py

@@ -204,11 +204,11 @@ def fragment_until_valence_is_correct(mol, frags, verbose=False, _events=None):
         When *True* the function returns a 2-tuple ``(frags, events)`` instead
         of just ``frags``.  *events* is a list of dicts, one per fragmentation
         step, each containing:
-        ``atom_idx`` – index of the offending atom,
-        ``error`` – the original :class:`~rdkit.Chem.AtomValenceException`
+        ``atom_idx`` - index of the offending atom,
+        ``error`` - the original :class:`~rdkit.Chem.AtomValenceException`
         message,
-        ``n_fragments`` – how many pieces the bond-cut produced,
-        ``smiles`` – SMILES of the fragment being cut (may be ``None`` if
+        ``n_fragments`` - how many pieces the bond-cut produced,
+        ``smiles`` - SMILES of the fragment being cut (may be ``None`` if
         RDKit cannot generate one for an unsanitized mol).
     _events : list or None
         Internal accumulator used during recursion; do **not** pass explicitly.
@@ -370,6 +370,7 @@ def add_componentSmarts(filename = COMPONENTS_FILE):
             # Normalize filters to lists
             if halogens is not None:
                 halogens = [halogens] if isinstance(halogens, str) else list(halogens)
+                halogens = [h for h in halogens if h != '*']
             if form is not None:
                 form = [form] if isinstance(form, str) else list(form)
             if saturation is not None:

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/data/Halogen_groups_smarts.json

@@ -4,6 +4,7 @@
     "name": "OECD perfluoroalkyl alcohols",
     "alias": "Perfluoroalkyl alcohols",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "only": [
@@ -57,6 +58,7 @@
     "name": "OECD telomer alcohols",
     "alias": "telomer alcohols",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "max_dist_from_comp": 0,
     "constraints": {
@@ -132,6 +134,7 @@
     "name": "OECD perfluoroalkane",
     "alias": "Perfluoroalkane",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "only": [
@@ -181,6 +184,7 @@
     "name": "OECD semi-fluorinated alkanes",
     "alias": "SFAs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -236,6 +240,7 @@
     "name": "OECD perfluoroalkene",
     "alias": "Perfluoroalkene",
     "componentSmarts": null,
+    "componentHalogen": "F",
     "max_dist_from_comp": 16,
     "linker_smarts": null,
     "constraints": {
@@ -287,6 +292,7 @@
     "name": "OECD perfluoroalkyl carboxylic acids",
     "alias": "PFCAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -343,6 +349,7 @@
     "name": "OECD polyfluoroalkyl carboxylic acid",
     "alias": "PolyFCAs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -404,6 +411,7 @@
     "name": "OECD perfluoroalkyl dicarboxylic acids",
     "alias": "PFdiCAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "max_dist_from_comp": 1,
     "linker_smarts": null,
     "constraints": {
@@ -462,6 +470,7 @@
     "name": "OECD perfluoroalkylether carboxylic acids",
     "alias": "PFECAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -518,6 +527,7 @@
     "name": "OECD polyfluoroalkylether carboxylic acid",
     "alias": "PolyFECAs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -582,6 +592,7 @@
     "name": "OECD hydrofluoroethers",
     "alias": "HFEs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -637,6 +648,7 @@
     "name": "OECD hydrofluorocarbons",
     "alias": "HFCs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -689,6 +701,7 @@
     "name": "OECD perfluoroalkyl iodides",
     "alias": "PFAIs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -735,6 +748,7 @@
     "name": "OECD perfluoroalkyl ketones",
     "alias": "Perfluoroalkyl ketones",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "only": [
@@ -787,6 +801,7 @@
     "name": "OECD semi-fluoroalkyl ketones",
     "alias": "Semi-fluoroalkyl ketones",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -842,6 +857,7 @@
     "name": "OECD hydrofluoroolefins",
     "alias": "HFOs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "max_dist_from_comp": 16,
     "linker_smarts": "[CH2]",
     "constraints": {
@@ -895,6 +911,7 @@
     "name": "OECD perfluoropolyethers",
     "alias": "PFPEs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "only": [
@@ -948,6 +965,7 @@
     "name": "OECD side-chain fluorinated aromatics",
     "alias": "Side-chain fluorinated aromatics",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "gte": {
@@ -985,6 +1003,7 @@
     "name": "OECD perfluoroalkylether sulfonic acids",
     "alias": "PFESAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1049,6 +1068,7 @@
     "name": "OECD polyfluoroalkylether sulfonic acid",
     "alias": "PolyFESAs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1121,6 +1141,7 @@
     "name": "OECD perfluoroalkyl sulfinic acids",
     "alias": "PFSiAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1178,6 +1199,7 @@
     "name": "OECD perfluoroalkyl sulfonic acids",
     "alias": "PFSAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1236,6 +1258,7 @@
     "name": "OECD polyfluoroalkyl sulfonic acid",
     "alias": "PolyFSAs",
     "componentSmarts": "Polyfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1302,6 +1325,7 @@
     "name": "OECD perfluoroalkyl disulfonic acids",
     "alias": "PFdiSAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "max_dist_from_comp": 1,
     "linker_smarts": null,
     "constraints": {
@@ -1364,6 +1388,7 @@
     "name": "OECD perfluoroalkane sulfonyl fluorides",
     "alias": "PASFs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "max_dist_from_comp": 0,
     "linker_smarts": null,
     "constraints": {
@@ -1411,6 +1436,7 @@
     "name": "OECD perfluoroalkyl phosphinic acids",
     "alias": "PFPiAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1468,6 +1494,7 @@
     "name": "OECD perfluoroalkyl phosphonic acids",
     "alias": "PFPAs",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "eq": {
@@ -1525,6 +1552,7 @@
     "name": "OECD perfluoroalkyl-tert-amines",
     "alias": "Perfluoroalkyl-tert-amines",
     "componentSmarts": "Perfluoroalkyl",
+    "componentHalogen": "F",
     "linker_smarts": null,
     "constraints": {
       "only": [
@@ -4258,8 +4286,8 @@
   },
   {
     "id": 65,
-    "name": "thia keto propanoic acid",
-    "alias": "Thia keto propanoic acid",
+    "name": "thioester keto dicarboxylic acid",
+    "alias": "Thioester keto dicarboxylic acid",
     "componentSmarts": null,
     "linker_smarts": null,
     "max_dist_from_comp": 0,
@@ -4331,7 +4359,7 @@
     "sorting": {
       "category": "generic",
       "elements": "SCO",
-      "group": "thia keto propanoic acids",
+      "group": "thioester keto dicarboxylic acids",
       "saturation": null,
       "aggregated": false
     }
@@ -7871,8 +7899,8 @@
   },
   {
     "id": 106,
-    "name": "telomer thia keto propanoic acid",
-    "alias": "telomer thia keto propanoic acid",
+    "name": "telomer thioester keto dicarboxylic acid",
+    "alias": "telomer thioester keto dicarboxylic acid",
     "componentSmarts": null,
     "componentSaturation": "per",
     "linker_smarts": "[CH2X4]",
@@ -7960,7 +7988,7 @@
     "sorting": {
       "category": "telomer",
       "elements": "SCO",
-      "group": "thia keto propanoic acids",
+      "group": "thioester keto dicarboxylic acids",
       "saturation": null,
       "aggregated": false
     }

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/data/component_smarts_halogens.json

@@ -138,5 +138,14 @@
                 "component": "[#6aR$([#6aR][I,#6aR$([#6aR][#53])]),#6aR$([#6aR]([#9,#17,#35,#53])[#6aR]I)]"
             }
         }
+    },
+    "H": {
+        "alkyl": {
+            "per": {
+                "name": "Alkyl",
+                "component": "[CH2]",
+                "end": "[CH3]"
+            }
+        }
     }
 }

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/generate_homologues.py

@@ -44,7 +44,8 @@ def _mol_from_input(mol_or_string):
     )
 
 # Halogen element symbols and their atomic numbers used to build per-halogen SMARTS
-_HALOGEN_ATOMIC_NUM = {'F': 9, 'Cl': 17, 'Br': 35, 'I': 53}
+# 'H' is included to support CH2-based hydrocarbon homologue series
+_HALOGEN_ATOMIC_NUM = {'F': 9, 'Cl': 17, 'Br': 35, 'I': 53, 'H': 1}
 
 # Default componentSmartsName per halogen (per-halo, alkyl chain)
 _DEFAULT_COMPONENT_NAME = {
@@ -52,6 +53,7 @@ _DEFAULT_COMPONENT_NAME = {
     'Cl': 'Perchloroalkyl',
     'Br': 'Perbromoalkyl',
     'I': 'Periodoalkyl',
+    'H': 'Alkyl',
 }
 
 
@@ -82,8 +84,8 @@ def find_halogenated_components(mol, component_smarts, halogen='F'):
     list of dict
         Each entry describes one halogenated component:
 
-        - ``'component'``  (``frozenset[int]``) – atom indices of all component atoms.
-        - ``'cx2_carbons'`` (``list[int]``)      – backbone C atoms bearing ≥ 2 halogen
+        - ``'component'``  (``frozenset[int]``) - atom indices of all component atoms.
+        - ``'cx2_carbons'`` (``list[int]``)      - backbone C atoms bearing ≥ 2 halogen
           substituents (i.e., candidate ``CX2`` units for homologue generation).
     """
     component_atom_idxs = get_substruct(mol, component_smarts)
@@ -103,14 +105,24 @@ def find_halogenated_components(mol, component_smarts, halogen='F'):
         # Identify CX2 backbone carbons: C in component bearing *exactly* 2 halogen
         # neighbours.  Terminal CF3 (3 halogens) and mono-halo atoms are excluded —
         # only true -CX2- repeating units qualify.
-        cx2 = [
-            idx for idx in cc
-            if mol.GetAtomWithIdx(idx).GetSymbol() == 'C'
-            and sum(
-                1 for nb in mol.GetAtomWithIdx(idx).GetNeighbors()
-                if nb.GetSymbol() == halogen
-            ) == 2
-        ]
+        #
+        # For halogen='H', H atoms are implicit in RDKit molecules, so we use
+        # GetTotalNumHs() instead of iterating explicit neighbours.
+        if halogen == 'H':
+            cx2 = [
+                idx for idx in cc
+                if mol.GetAtomWithIdx(idx).GetSymbol() == 'C'
+                and mol.GetAtomWithIdx(idx).GetTotalNumHs() == 2
+            ]
+        else:
+            cx2 = [
+                idx for idx in cc
+                if mol.GetAtomWithIdx(idx).GetSymbol() == 'C'
+                and sum(
+                    1 for nb in mol.GetAtomWithIdx(idx).GetNeighbors()
+                    if nb.GetSymbol() == halogen
+                ) == 2
+            ]
         results.append({'component': frozenset(cc), 'cx2_carbons': cx2})
 
     return results
@@ -155,7 +167,7 @@ def generate_homologues(mol_input, componentSmartsName=None, componentSmartss=No
     Returns
     -------
     dict
-        ``{InChIKey: {formula: rdkit.Chem.Mol}}`` – all unique shorter
+        ``{InChIKey: {formula: rdkit.Chem.Mol}}`` - all unique shorter
         homologues, keyed first by InChIKey and then by molecular formula.
 
     Raises
@@ -169,7 +181,7 @@ def generate_homologues(mol_input, componentSmartsName=None, componentSmartss=No
     --------
     >>> from rdkit import Chem
     >>> from HalogenGroups.generate_homologues import generate_homologues
-    >>> # PFOA – generate all shorter perfluoroalkyl chain homologues
+    >>> # PFOA - generate all shorter perfluoroalkyl chain homologues
     >>> pfoa = Chem.MolFromSmiles('OC(=O)' + 'C(F)(F)' * 7 + 'F')
     >>> homologues = generate_homologues(pfoa)
     >>> print(len(homologues))  # 6  (C2-C7)
@@ -198,7 +210,8 @@ def generate_homologues(mol_input, componentSmartsName=None, componentSmartss=No
 
     if halogen not in _HALOGEN_ATOMIC_NUM:
         raise ValueError(
-            f"halogen must be one of {list(_HALOGEN_ATOMIC_NUM)}, got {halogen!r}"
+            f"halogen must be one of {list(_HALOGEN_ATOMIC_NUM)}, got {halogen!r}. "
+            f"Use 'H' for hydrocarbon (CH2) backbone series."
         )
     if base_repeating is None:
         base_repeating = ['C']

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/generate_mol.py

@@ -646,7 +646,7 @@ def generate_random_mol(n, functional_groups=None, perfluorinated=True, cycle=Fa
             mol = append_functional_groups(mol, _fg, chain_n=n,
                                            **{k: v for k, v in kwargs.items() if k not in _skip_keys})
         if _max_defs is not None:
-            from PFASGroups import parse_mol as _parse_mol  # lazy import – avoids circular dependency
+            from PFASGroups import parse_mol as _parse_mol  # lazy import - avoids circular dependency
             _result = _parse_mol(mol, include_PFAS_definitions=True)
             _detected = {m['id'] for m in _result.get('matches', []) if m.get('type') == 'PFASdefinition'}
             if len(_detected) <= _max_defs:

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/getter.py

@@ -36,7 +36,7 @@ def get_compiled_HalogenGroups(**kwargs):
 
     Examples
     --------
-    >>> from PFASgroups import get_compiled_HalogenGroups, HalogenGroup, parse_smiles
+    >>> from PFASGroups import get_compiled_HalogenGroups, HalogenGroup, parse_smiles
     >>> groups = get_compiled_HalogenGroups()
     >>> groups.append(HalogenGroup(
     ...     id=200, name="Perfluoroalkyl nitrates",

{pfasgroups-3.2.2 → pfasgroups-3.4.0}/PFASGroups/homologue_series.py

@@ -1,4 +1,4 @@
-"""HomologueSeries – result container for :func:`~PFASGroups.generate_homologues`.
+"""HomologueSeries - result container for :func:`~PFASGroups.generate_homologues`.
 
 Backward-compatible with the raw ``{InChIKey: {formula: mol}}`` dict that
 ``generate_homologues`` previously returned.  All existing code that iterates
@@ -6,10 +6,10 @@ or indexes the dict continues to work unchanged.
 
 The class adds:
 
-- :meth:`summary` / :meth:`summarise` – coloured text summary
-- :meth:`show` / :meth:`plot` – grid image of all homologues
-- :meth:`svg` – export to an SVG file
-- :meth:`to_sql` – persist to SQLite or PostgreSQL
+- :meth:`summary` / :meth:`summarise` - coloured text summary
+- :meth:`show` / :meth:`plot` - grid image of all homologues
+- :meth:`svg` - export to an SVG file
+- :meth:`to_sql` - persist to SQLite or PostgreSQL
 
 Example
 -------
@@ -46,7 +46,7 @@ _HOMOLOGUE_COLOR = _HOMOLOGUE_COLOUR  # US alias
 
 
 # ---------------------------------------------------------------------------
-# HomologueEntry – lightweight descriptor for a single homologue
+# HomologueEntry - lightweight descriptor for a single homologue
 # ---------------------------------------------------------------------------
 
 @dataclass
@@ -86,7 +86,7 @@ class HomologueEntry:
 
 
 # ---------------------------------------------------------------------------
-# HomologueSeries – dict subclass (backward-compatible) with rich API
+# HomologueSeries - dict subclass (backward-compatible) with rich API
 # ---------------------------------------------------------------------------
 
 class HomologueSeries(dict):
@@ -116,7 +116,7 @@ class HomologueSeries(dict):
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
-        # Metadata – populated by generate_homologues after construction
+        # Metadata - populated by generate_homologues after construction
         self.parent_mol: Optional[Chem.Mol] = None
         self.parent_smiles: Optional[str] = None
         self.parent_formula: Optional[str] = None