Resnik distances between GO terms work now.

workflows/geneontology.py

@@ -60,6 +60,31 @@ class GeneOntology(networkx.XDiGraph):
         """Returns the root node of the molecular_function tree"""
         return self.by_id['GO:0003674']
 
+    def subsumer(self, t1, t2):
+        if t1 == t2:
+#            print "t1 == t2"
+            return t1
+
+        go_undir = self.to_undirected()
+#        print go_undir.nodes()
+        path = networkx.shortest_path(go_undir, t1, t2)
+        if not path:
+            print "Woah, path not found."
+            return None
+
+        if path == [1]:
+            print "This shouldn't happen"
+            return t1
+
+#        print t1['id'], t2['id'], path
+#        print "path:", path
+        for t in path:
+            if networkx.shortest_path(self, t, t1) and \
+                networkx.shortest_path(self, t, t2): 
+#                print "   ", t1, t2, t
+                return t
+
+        print "GeneOntology.subsumer: should not reach this point"
 
 def _split_obo_line(line):
     """Splits a line from an obo file in its three constituent parts.

workflows/go_workflow.py

@@ -118,6 +118,7 @@ class GoWorkflow (workflow.Workflow):
         self.add_stage(load)
         
         go = workflow.Stage('go', 'Gene Ontology')
+        go.add_function(SelectGoTermsFunction(self))
         go.add_function(GoDistanceFunction())
         self.add_stage(go)
 
@@ -269,11 +270,36 @@ class NumericDict(dict):
             retval = 0.0
         return retval
 
+
+class SelectGoTermsFunction(workflow.Function):
+    def __init__(self, wf):
+        workflow.Function.__init__(self, 'go-select', 'Select GO Terms')
+        self.wf = wf
+
+    def run(self):
+        self.wf.project.set_selection('go-terms', set(['GO:0007582', 'GO:0008150', 'GO:0051704']))
+
+
 class GoDistanceFunction(workflow.Function):
     def __init__(self):
         workflow.Function.__init__(self, 'go-dist', 'GO term distance matrix')
         self.options = GoDistanceOptions()
         
+    def resnik_distance_matrix(self, selection, ic):
+        size = len(selection['go-terms'])
+        m = zeros((size, size))
+        # Create resnik distance matrix
+        ids = list(selection['go-terms'])
+        for i, t1 in enumerate(ids):
+            for j, t2 in enumerate(ids):
+                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[subsumer['id']] - ic[t1] + ic[subsumer['id']] - ic[t2]
+        ds = dataset.Dataset(m, (('go-terms', ids), ('_go-terms', ids)), 'Resnik')
+        return ds
+
     def run(self, x, selection):
         global evidence, go
         self.options = self.show_gui(self.options)
@@ -288,22 +314,21 @@ class GoDistanceFunction(workflow.Function):
         annotations = NumericDict()
         ic = NumericDict()
 
+        # Insert annotations into dict
+        for i, v in enumerate(evidence.get_identifiers('go-terms')):
+            annotations[v] = ann_count_matrix[i]
+
         # Accumulate annotations
         for term in reversed(networkx.topological_sort(go)):
             for parent in go.in_neighbors(term):
                 annotations[parent['id']] += annotations[term['id']]
-                print "%s -> %s (%s)" % (term['id'], parent['id'], annotations[parent['id']])
+#                print "%s -> %s (%s)" % (term['id'], parent['id'], annotations[parent['id']])
 
         # Create information content dictionary
         for i, v in enumerate(evidence.get_identifiers('go-terms')):
-            annotations[v] = ann_count_matrix[i]
             ic[v] = -log(ann_count_matrix[i] / total_ann)
 
-#        # Create resnik distance matrix
+        return [self.resnik_distance_matrix(selection, ic)]
-#        for t1 in selection['go-terms']:
-#            for t2 in selection['go-terms']:
-                
-        
         
     def show_gui(self, options, edit=True):
         dialog = GoDistanceDialog()