Merge pull request #33 from gruenewald-lab/maint

Maint

cgsmiles/graph_utils.py

@@ -175,7 +175,9 @@ def set_atom_names_atomistic(molecule, meta_graph=None):
             assert len(fragids) == 1
             fraglist[fragids[0]].append(node)
 
-    for fragnodes in fraglist.values():
+    for meta_node, fragnodes in fraglist.items():
         for idx, node in enumerate(fragnodes):
             atomname = molecule.nodes[node]['element'] + str(idx)
             molecule.nodes[node]['atomname'] = atomname
+            if meta_graph:
+                meta_graph.nodes[meta_node]['graph'].nodes[node]['atomname'] = atomname

cgsmiles/pysmiles_utils.py

@@ -76,9 +76,21 @@ def rebuild_h_atoms(mol_graph, keep_bonding=False):
         raise SyntaxError(msg)
     nx.set_node_attributes(mol_graph, 0, 'hcount')
 
+    # first we need to figure out the correct hcounts on each node
+    # this also corrects for simple aromatic problems like in thiophene
     pysmiles.smiles_helper.fill_valence(mol_graph, respect_hcount=False)
+
+    # optionally we adjust the hcount by the number of bonding operators
+    if keep_bonding:
+        bonding_nodes = nx.get_node_attributes(mol_graph, 'bonding')
+        for node, bond_ops in bonding_nodes.items():
+            mol_graph.nodes[node]['hcount'] -= sum([int(bond[-1]) for bond in bond_ops])
+
+    # now we add the hydrogen atoms
     pysmiles.smiles_helper.add_explicit_hydrogens(mol_graph)
 
+    # if we are having single hydrogen fragments we need to
+    # make sure the fragid and fragname is keept
     for node in mol_graph.nodes:
         if mol_graph.nodes[node].get("element", "*") == "H" and\
         not mol_graph.nodes[node].get("single_h_frag", False):

cgsmiles/read_fragments.py

@@ -126,6 +126,7 @@ def strip_bonding_descriptors(fragment_string):
     node_count = 0
     prev_node = 0
     current_order = None
+    anchor = []
     for token in smile_iter:
         if token == '[':
             peek = next(smile_iter)
@@ -157,15 +158,15 @@ def strip_bonding_descriptors(fragment_string):
                     else:
                         atom += peek
                     peek = next(smile_iter)
-
                 smile = smile + atom + "]"
                 prev_node = node_count
                 node_count += 1
+                current_order = None
         elif token == '(':
-            anchor = prev_node
+            anchor.append(prev_node)
             smile += token
         elif token == ')':
-            prev_node = anchor
+            prev_node = anchor.pop()
             smile += token
         elif token in bond_to_order:
             current_order = bond_to_order[token]

cgsmiles/resolve.py

@@ -382,18 +382,16 @@ def resolve(self):
             mark_chiral_atoms(self.molecule)
             # assign rs isomerism
             annotate_ez_isomers(self.molecule)
-            # in all-atom MD there are common naming conventions
-            # that might be expected and hence we set them here
-            set_atom_names_atomistic(self.molecule, self.meta_graph)
 
         # and redo the meta molecule
         self.meta_graph = annotate_fragments(self.meta_graph,
                                              self.molecule)
 
-        # in all-atom MD there are common naming conventions
-        # that might be expected and hence we set them here
         if all_atom:
-            set_atom_names_atomistic(self.molecule, self.meta_graph)
+            # in all-atom MD there are common naming conventions
+            # that might be expected and hence we set them here
+            set_atom_names_atomistic(self.molecule,
+                                     self.meta_graph)
 
         # increment the resolution counter
         self.resolution_counter += 1

cgsmiles/tests/test_cgsmile_parsing.py

@@ -269,6 +269,36 @@ def test_read_cgsmiles(smile, nodes, charges, edges, orders):
                         {0: ["$1"], 2: ["$1"]},
                         None,
                         None),
+                        # smiple symmetric bonding after branch
+                        ("[$]CC(CC)[$]",
+                         "CC(CC)",
+                        {0: ["$1"], 1: ["$1"]},
+                        None,
+                        None),
+                        # smiple symmetric bonding after ring
+                        ("[$]CC1[$]CCC1",
+                         "CC1CCC1",
+                        {0: ["$1"], 1: ["$1"]},
+                        None,
+                        None),
+                        # clear order symbol
+                        ("[CH][$a]=[CH][$c]",
+                         "[CH]=[CH]",
+                        {0: ["$a1"], 1: ["$c1"]},
+                        None,
+                        None),
+                        # multiple non-one bonding l
+                        ("CC=[$a]=[$b]CC",
+                         "CCCC",
+                        {1: ["$a2", "$b2"]},
+                        None,
+                        None),
+                        # multiple non-one bonding l
+                        ("CC[$a]=[$b]CC",
+                         "CCCC",
+                        {1: ["$a1", "$b2"]},
+                        None,
+                        None),
                         # smiple symmetric bonding with more than one name
                         ("[$1A]COC[$1A]",
                          "COC",

cgsmiles/tests/test_utils.py

@@ -0,0 +1,32 @@
+import re
+import pytest
+import cgsmiles
+
+err_msg_rebuild_h = ("Likely you are writing an aromatic molecule that does not "
+                     "show delocalization-induced molecular equivalency and thus "
+                     "is not considered aromatic. For example, 4-methyl imidazole "
+                     "is often written as [nH]1cc(nc1)C, but should be written as "
+                     "[NH]1C=C(N=C1)C. A corresponding CGSmiles string would be "
+                     "{[#A]1[#B][#C]1}.{#A=[>][<]N,#B=[$]N=C[>],#C=[$]C(C)=C[<]}")
+
+@pytest.mark.parametrize('frag_str, hatoms_ref, error_type, err_msg', (
+                        ('{#A=[$]CCC[$]}', 6, None, None),
+                        ('{#A=CCC}', 8, None, None),
+                        ('{#A=C[!]CC}', 7, None, None),
+                        ('{#A=[$]=CCC=[$]}', 4, None, None),
+                        ('{#A=[$]cccc}',5, None, None),
+                        ('{#A=[$]ccc}', 0, SyntaxError, err_msg_rebuild_h),
+))
+def test_rebuild_hatoms(frag_str, hatoms_ref, error_type, err_msg):
+    frag_dict = cgsmiles.read_fragments(frag_str)
+    frag_graph = frag_dict['A']
+    if error_type:
+        with pytest.raises(error_type, match=re.escape(err_msg)):
+            cgsmiles.pysmiles_utils.rebuild_h_atoms(frag_graph, keep_bonding=True)
+    else:
+        cgsmiles.pysmiles_utils.rebuild_h_atoms(frag_graph, keep_bonding=True)
+        hatoms = 0
+        for node, ele in frag_graph.nodes(data='element'):
+            if ele == 'H':
+                hatoms += 1
+        assert hatoms == hatoms_ref