fix rare issue in subgraph_isomorphism due to greedy handling of matches

build/lib/glycowork/motif/graph.py

@@ -274,7 +274,7 @@ def subgraph_isomorphism(glycan, motif, termini_list = [], count = False, wildca
       return (0, []) if return_matches else 0 if count else False
     motif_comp = min_process_glycans([motif, glycan])
     if wildcards_ptm:
-      glycan, motif = [re.sub(r"(?<=[a-zA-Z])\d+(?=[a-zA-Z])", 'O', glycan).replace('NeuOAc', 'Neu5Ac').replace('NeuOGc', 'Neu5Gc') for glycan in [glycan, motif]]
+      glycan, motif = [re.sub(r"(?<=[a-zA-Z])\d+(?=[a-zA-Z])", 'O', g).replace('NeuOAc', 'Neu5Ac').replace('NeuOGc', 'Neu5Gc') for g in [glycan, motif]]
     g1 = glycan_to_nxGraph(glycan, termini = 'calc') if termini_list else glycan_to_nxGraph(glycan)
     g2 = glycan_to_nxGraph(motif, termini = 'provided', termini_list = termini_list) if termini_list else glycan_to_nxGraph(motif)
   else:
@@ -322,11 +322,11 @@ def subgraph_isomorphism(glycan, motif, termini_list = [], count = False, wildca
     return counts if not return_matches else (counts, mappings)
   else:
     if graph_pair.subgraph_is_isomorphic():
-      mapping = graph_pair.mapping
-      mapping = {v: k for k, v in mapping.items()}
-      res = all(mapping[node] < mapping[neighbor] for node, neighbor in g2.edges())
-      return res if not return_matches else (int(res), mappings)
-    return False if not return_matches else (0, [])
+      for mapping in graph_pair.subgraph_isomorphisms_iter():
+        mapping = {v: k for k, v in mapping.items()}
+        if all(mapping[node] < mapping[neighbor] for node, neighbor in g2.edges()):
+          return True if not return_matches else (1, mappings)
+  return False if not return_matches else (0, [])
 
 
 def generate_graph_features(glycan, glycan_graph = True, label = 'network'):

glycowork/motif/graph.py

@@ -274,7 +274,7 @@ def subgraph_isomorphism(glycan, motif, termini_list = [], count = False, wildca
       return (0, []) if return_matches else 0 if count else False
     motif_comp = min_process_glycans([motif, glycan])
     if wildcards_ptm:
-      glycan, motif = [re.sub(r"(?<=[a-zA-Z])\d+(?=[a-zA-Z])", 'O', glycan).replace('NeuOAc', 'Neu5Ac').replace('NeuOGc', 'Neu5Gc') for glycan in [glycan, motif]]
+      glycan, motif = [re.sub(r"(?<=[a-zA-Z])\d+(?=[a-zA-Z])", 'O', g).replace('NeuOAc', 'Neu5Ac').replace('NeuOGc', 'Neu5Gc') for g in [glycan, motif]]
     g1 = glycan_to_nxGraph(glycan, termini = 'calc') if termini_list else glycan_to_nxGraph(glycan)
     g2 = glycan_to_nxGraph(motif, termini = 'provided', termini_list = termini_list) if termini_list else glycan_to_nxGraph(motif)
   else:
@@ -322,11 +322,11 @@ def subgraph_isomorphism(glycan, motif, termini_list = [], count = False, wildca
     return counts if not return_matches else (counts, mappings)
   else:
     if graph_pair.subgraph_is_isomorphic():
-      mapping = graph_pair.mapping
-      mapping = {v: k for k, v in mapping.items()}
-      res = all(mapping[node] < mapping[neighbor] for node, neighbor in g2.edges())
-      return res if not return_matches else (int(res), mappings)
-    return False if not return_matches else (0, [])
+      for mapping in graph_pair.subgraph_isomorphisms_iter():
+        mapping = {v: k for k, v in mapping.items()}
+        if all(mapping[node] < mapping[neighbor] for node, neighbor in g2.edges()):
+          return True if not return_matches else (1, mappings)
+  return False if not return_matches else (0, [])
 
 
 def generate_graph_features(glycan, glycan_graph = True, label = 'network'):

glycowork/motif/regex.py

@@ -113,6 +113,8 @@ def convert_pattern_component(pattern_component):
   | Returns a string for simple components and a dict of form string : occurrence for complex components
   """
   if not any([k in pattern_component for k in ['[', '{', '*', '+', '=', '<!', '?!']]):
+    if pattern_component[-1].isdigit() or pattern_component[-1] == '?':
+      pattern_component += '-'
     return specify_linkages(replace_patterns(pattern_component))
   pattern, occurrence = None, None
   if '[' in pattern_component:
@@ -592,7 +594,7 @@ def filter_dealbreakers(lists, ggraph, pattern):
     return lists2
 
 
-def compile(pattern):
+def compile_pattern(pattern):
   """pre-compiles glyco-regular expression for faster processing\n
   | Arguments:
   | :-
@@ -610,7 +612,7 @@ def get_match(pattern, glycan, return_matches = True):
   | Arguments:
   | :-
   | pattern (string): glyco-regular expression in the form of "Hex-HexNAc-([Hex|Fuc]){1,2}-HexNAc"; accepts pre-compiled pattern
-  | glycan (string): glycan sequence in IUPAC-condensed
+  | glycan (string or networkx): glycan sequence in IUPAC-condensed or as networkx graph
   | return_matches (bool): whether to return True/False or return the matches as a list of strings; default:True\n
   | Returns:
   | :-
@@ -644,7 +646,7 @@ def get_match_batch(pattern, glycan_list, return_matches = True):
   | Arguments:
   | :-
   | pattern (string): glyco-regular expression in the form of "Hex-HexNAc-([Hex|Fuc]){1,2}-HexNAc"; accepts pre-compiled pattern
-  | glycan_list (list of strings): list of glycan sequence in IUPAC-condensed
+  | glycan_list (list of strings or networkx): list of glycan sequence in IUPAC-condensed or as networkx graph
   | return_matches (bool): whether to return True/False or return the matches as a list of strings; default:True\n
   | Returns:
   | :-