Small updates.

workflows/geneontology.py

@@ -61,7 +61,29 @@ class GeneOntology(networkx.XDiGraph):
         """Returns the root node of the molecular_function tree"""
         return self.by_id['GO:0003674']
 
+    def _subsumer(self, t1, t2, heap):
+        while heap != []:
+            t = heap[0]
+            heap = heap[1:]
+
+            p1 = networkx.shortest_path(self, t, t1)
+            p2 = networkx.shortest_path(self, t, t2)
+            if p1 and p2:
+                return t
+
+            heap += self.in_neighbors(t)
+        return None
+
     def subsumer(self, t1, t2):
+        if t1 == t2:
+            return t1
+        if networkx.shortest_path(self, t1, t2):
+            return t1
+        elif networkx.shortest_path(self, t2, t1):
+            return t2
+        return self._subsumer(t1, t2, self.in_neighbors(t1))
+
+    def old_subsumer(self, t1, t2):
         if t1 == t2:
             return t1
 
@@ -83,6 +105,8 @@ class GeneOntology(networkx.XDiGraph):
                 return t
 
         print "GeneOntology.subsumer: should not reach this point"
+        print "path is now: %s" % path
+        print "ids are: %s " % [x['id'] for x in path]
 
 def _split_obo_line(line):
     """Splits a line from an obo file in its three constituent parts.

workflows/go_workflow.py

@@ -277,8 +277,11 @@ class SelectGoTermsFunction(workflow.Function):
         workflow.Function.__init__(self, 'go-select', 'Select GO Terms')
         self.wf = wf
 
-    def run(self):
+    def run(self, ds):
-        self.wf.project.set_selection('go-terms', set(['GO:0007582', 'GO:0008150', 'GO:0051704', 'GO:0044419']))
+        terms = [x['id'] for x in networkx.paths.bfs(go, go.get_bp())]
+        
+        self.wf.project.set_selection('go-terms', set(terms[:100]))
+#        self.wf.project.set_selection('go-terms', set(['GO:0007582', 'GO:0008150', 'GO:0051704', 'GO:0044419']))
 
 
 class GoDistanceFunction(workflow.Function):
@@ -296,8 +299,12 @@ class GoDistanceFunction(workflow.Function):
                 term1 = go.by_id[t1]
                 term2 = go.by_id[t2]
                 subsumer = go.subsumer(term1, term2)
-                print "%s - %s - %s" % (t1, subsumer['id'], t2)
+                
-                m[i, j] = ic[t1] + ic[t2] - 2.0 * ic[subsumer['id']]
+                if subsumer == None:
+                    m[i, j] = 1000
+                else:
+#                    print "%s - %s - %s" % (t1, subsumer['id'], t2)
+                    m[i, j] = ic[t1] + ic[t2] - 2.0 * ic[subsumer['id']]
         ds = dataset.Dataset(m, (('go-terms', ids), ('_go-terms', ids)), 'Resnik')
         return ds