add OH,CH3 subs; avoid duplicate substituents; add to tests

automol/graph/_3super_func_group.py

@@ -78,6 +78,7 @@ def classify_species(gra):
     "C5,OH-M",
     "A1,OH-M",
     "A1,OH,OH-M",
+    "A1,OH,CH3-M",
     "A1,OH,CHO-M",
     "A1,OH,OCH3-M",
     "A1,CHO-M",
@@ -111,28 +112,36 @@ def assign_grps(self, gra):
         # assign base groups
         for key, fct in BASE_GRP_DCT.items():
             self.sup_grps[key] = self.grp_fct_dct[fct]
-
+        base_grps_0 = list(itertools.chain(*[grp for grp in self.sup_grps.values() if len(grp) > 0]))
         # assign substituents
         subs_fct_dct = {}
         for key, fct in SUBSTITUENTS_GRP_DCT.items():
             subs_fct_dct[key] = self.grp_fct_dct[fct]
 
         # CH3CK C6H5C2H2, C6H5C2H4!!
         # assign composite
+        heavy_atms = list(implicit(gra)[0].keys())
         for comp_grp in COMPOSITE_GRP_LIST:
             base_and_subs, base_type = comp_grp.split("-")
             base, subs = (
                 base_and_subs.split(",")[0] + "-" + base_type,
                 base_and_subs.split(",")[1:],
             )
+
             base_grps = self.sup_grps[base]  # base groups to search substituents in
             for sub in subs:
                 sub_grps = subs_fct_dct[sub]
-                # intersection becomes the new base_grps;
+                sub_grps_eff = ()
+
+                # if the atoms of the substituent are part of (any) base group: skip
+                for grp in sub_grps:
+                    if not any(all(atm in basei for atm in grp if atm in heavy_atms) for basei in base_grps_0):
+                        sub_grps_eff += (grp,)
+                # intersection base+sub becomes the new base_grps;
                 # filter by bond type, e.g., C-C, C-O..
                 # with bonded_grps only: fails for OCH3
                 # (CH2-O bonded to an aromatic would work too)
-                base_grps = bonded_grps_checksymb(gra, base_grps, sub_grps, "C", sub[0])
+                base_grps = bonded_grps_checksymb(gra, base_grps, sub_grps_eff, "C", sub[0])
             # add to dct
             self.sup_grps[comp_grp] = base_grps
 
@@ -179,6 +188,7 @@ def bonded_grps_checksymb(gra, grps1, grps2, symb1, symb2):
     heavy_atms = list(implicit(gra)[0].keys())
     correct_bonds = bonds_of_type(gra, symb1, symb2)
     grps = ()
+    assigned_grps2 = () # make sure that each substituent is assigned to one group only
     if len(grps1) > 0 and len(grps2) > 0 and len(correct_bonds) > 0:
         for grp1 in grps1:
             # keep only heavy atoms
@@ -187,13 +197,14 @@ def bonded_grps_checksymb(gra, grps1, grps2, symb1, symb2):
                 grp2 = tuple(
                     atm for atm in grp2 if atm in heavy_atms and atm not in grp1
                 )
-                possible_bonds = list(itertools.product(grp1, grp2))
-                effective_bonds = (
+                possible_bonds = list(itertools.product(grp1, grp2)) + list(itertools.product(grp2, grp1))
+                effective_bonds = list(
                     bond for bond in possible_bonds if frozenset(bond) in gra[1].keys()
                 )
                 if len(tuple(set(effective_bonds).intersection(correct_bonds))) > 0:
                     grp = grp1 + grp2
-                    if sorted(grp) not in [sorted(grpi) for grpi in grps]:
+                    if sorted(grp) not in [sorted(grpi) for grpi in grps] and grp2 not in assigned_grps2:
                         grps += (grp,)
-
+                        assigned_grps2 += (grp2,)
+
     return grps

automol/tests/test_super_func_group.py

@@ -37,34 +37,42 @@
         automol.smiles.chi(smi)))
 
 SPCS_GRPS = {
+    'BZFUR': {'FUR-M': ((4, 5, 8, 7, 6),) ,
+        'A1-M': ((0, 1, 3, 7, 6, 2),) ,},
+    'AMN': {'A1-M': ((5, 6, 9, 10, 7, 8), (1, 2, 4, 10, 9, 3)) ,
+            'A1,CH3-M': ((5, 6, 9, 10, 7, 8, 0),)},
     'C6H5': {'A1-R': ((0, 1, 3, 5, 4, 2),)},
     'INDENYL': {
         'C5-RSR': ((2, 5, 7, 8, 6),) ,
      },
     'C12H8': {
         'A1-M': ((0, 2, 8, 11, 9, 4), (1, 3, 8, 11, 10, 5)) ,
+        'A1,C2H3-M': ((0, 2, 8, 11, 9, 4, 6, 7),),
      },
     'C9H7O': {'C5O-RSR': ((4, 5, 8, 7, 6, 9),) ,},
     'C5H4O': {'C5O-M': ((0, 1, 3, 4, 2, 5),) ,},
     'OC6H4CH3': {'A1O-RSR': ((1, 2, 4, 6, 5, 3, 7),) ,
             },
     'HOC6H4CH3': {
         'A1-M': ((1, 2, 5, 4, 6, 3),) ,
-        'A1,OH-M': ((1, 2, 5, 4, 6, 3, 7),)
-        },        # da aggiungere altri},
+        'A1,CH3-M':  ((1, 2, 5, 4, 6, 3, 0),),
+        'A1,OH-M': ((1, 2, 5, 4, 6, 3, 7),),
+        'A1,OH,CH3-M': ((1, 2, 5, 4, 6, 3, 7, 0),),
+        },     
     'C6H5CH2OOH': {'A1-M': ((0, 1, 3, 6, 4, 2),) ,
         },
-    'BZFUR': {'FUR-M': ((4, 5, 8, 7, 6),) ,
-        'A1-M': ((0, 1, 3, 7, 6, 2),) ,},
     'C10H7CH2': {'A1CH2-RSR': ((3, 4, 8, 10, 9, 6, 0),) ,},
     'C6H5C2H3': {'A1-M': ((2, 3, 5, 7, 6, 4),) ,
+                 'A1,C2H3-M': ((2, 3, 5, 7, 6, 4, 0, 1),),
         },
     'C10H9': {'A1CH2-RSR': ((0, 1, 5, 9, 8, 4, 7), (0, 1, 5, 9, 8, 4, 6)) ,
               },
     'CH3C6H4': {'A1-R': ((1, 2, 4, 6, 5, 3),) ,
+                'A1,CH3-R': ((1, 2, 4, 6, 5, 3, 0),),
         },
     'C6H5CCC6H5': {
         'A1-M': ((1, 4, 8, 13, 9, 5), (0, 2, 6, 12, 7, 3)) ,
+        'A1,C2H-M': ((1, 4, 8, 13, 9, 5, 10, 11),),
         },
     'C6H5C3H3-A': {'A1-M': ((2, 3, 6, 8, 7, 4),) ,
         'A1,C3.DD-M': ((2, 3, 6, 8, 7, 4, 0, 1, 5),)},
@@ -73,6 +81,7 @@
         },
     'CYC5H7': {'C5H2-RSR': ((0, 1, 3, 4, 2),) ,},
     'MEINDENYL': {'C5-RSR': ((5, 6, 8, 9, 7),) ,
+                  'C5,CH3-RSR': ((5, 6, 8, 9, 7, 0),)
         },
     'BENZOFLUORENE': {'C5-M': ((10, 12, 15, 16, 13),) ,
         'A1-M': ((0, 2, 6, 14, 11, 4), (8, 9, 13, 16, 14, 11), (1, 3, 7, 15, 12, 5)) ,
@@ -85,17 +94,22 @@
     'SALICALD': {
         'A1-M': ((0, 1, 3, 6, 5, 2),) ,
         'A1,OH-M': ((0, 1, 3, 6, 5, 2, 8),) ,
+        'A1,CHO-M': ((0, 1, 3, 6, 5, 2, 4, 7),), 
+        'A1,OH,CHO-M': ((0, 1, 3, 6, 5, 2, 8, 4, 7),), 
         },
 }
 
 def test_super_functional_group_dct():
 
     for SPC, DCT in SPCS_GRPS.items():
+        print(SPC)
         gra = SPCS_CHECKS_SMI[SPC]
         fgrps = automol.graph.SuperFunctionalGroup()
         fgrps.assign_grps(gra)
-        for key, val in DCT.items():
-            assert val == fgrps.sup_grps[key]
+        for key, val in fgrps.sup_grps.items():
+            if len(val) > 0:
+                assert val == DCT[key]
 
+
 if __name__ == '__main__':
     test_super_functional_group_dct()