Merge pull request #33 from dvklopfenstein/master

Added new GOTerm data member, depth, which tracks the depth of the longe...

goatools/obo_parser.py

@@ -111,14 +111,15 @@ def __init__(self):
         self._parents = []          # is_a basestring of parents
         self.parents = []           # parent records
         self.children = []          # children records
-        self.level = -1             # distance from root node
+        self.level = None           # shortest distance from root node
+        self.depth = None           # longest distance from root node
         self.is_obsolete = False    # is_obsolete
         self.alt_ids = []           # alternative identifiers
 
     def __str__(self):
         obsolete = "obsolete" if self.is_obsolete else ""
-        return "%s\tlevel-%02d\t%s [%s] %s" % (self.id, self.level, self.name,
-                                               self.namespace, obsolete)
+        return "%s\tlevel-%02d\tdepth-%02d\t%s [%s] %s" % (self.id, self.level, self.depth,
+                                               self.name, self.namespace, obsolete)
 
     def __repr__(self):
         return "GOTerm('%s')" % (self.id)
@@ -184,14 +185,22 @@ def load_obo_file(self, obo_file):
 
     def populate_terms(self):
 
-        def depth(rec):
-            if rec.level < 0:
+        def _init_level(rec):
+            if rec.level is None:
                 if not rec.parents:
                     rec.level = 0
                 else:
-                    rec.level = min(depth(rec) for rec in rec.parents) + 1
+                    rec.level = min(_init_level(rec) for rec in rec.parents) + 1
             return rec.level
 
+        def _init_depth(rec):
+            if rec.depth is None:
+                if not rec.parents:
+                    rec.depth = 0
+                else:
+                    rec.depth = max(_init_depth(rec) for rec in rec.parents) + 1
+            return rec.depth
+
         # make the parents references to the GO terms
         for rec in self.values():
             rec.parents = [self[x] for x in rec._parents]
@@ -201,8 +210,11 @@ def depth(rec):
             for p in rec.parents:
                 p.children.append(rec)
 
-            if rec.level < 0:
-                depth(rec)
+            if rec.level is None:
+                _init_level(rec)
+
+            if rec.depth is None:
+                _init_depth(rec)
 
     def write_dag(self, out=sys.stdout):
         for rec_id, rec in sorted(self.items()):

tests/test_get_paths.py

@@ -1,7 +1,18 @@
 
-from goatools.obo_parser import GODag
+
 import sys
 
+# Test local version of goatools
+sys.path.insert(0, '..')
+
+from goatools.obo_parser import GODag
+
+def prt_paths(paths, PRT=sys.stdout):
+  for path in paths:
+    PRT.write('\n')
+    for GO in path:
+      PRT.write('{}\n'.format(GO))
+
 def chk_results(actual_paths, expected_paths):
   for actual_path in actual_paths:
     # GOTerm -> list of Strings
@@ -10,14 +21,15 @@ def chk_results(actual_paths, expected_paths):
       raise Exception('ACTUAL {} NOT FOUND IN EXPECTED RESULTS\n'.format(actual))
 
 def test_paths_to_top():
+  #dag = GODag("./tests/data/mini_obo.obo")  
   dag = GODag("./data/mini_obo.obo")  
   expected_paths = [
     ['GO:0000001', 'GO:0000002', 'GO:0000005', 'GO:0000010'],
     ['GO:0000001', 'GO:0000003', 'GO:0000005', 'GO:0000010'],
     ['GO:0000001', 'GO:0000003', 'GO:0000006', 'GO:0000008', 'GO:0000010'] ]
   actual_paths = dag.paths_to_top("GO:0000010")
   chk_results(actual_paths, expected_paths)
-      
+  prt_paths(actual_paths)
 
 if __name__ == '__main__':
   test_paths_to_top()