Create Resnik distances and save them as ftsv.

workflows/geneontology.py

@@ -32,6 +32,7 @@ class GeneOntology(networkx.XDiGraph):
     def __init__(self):
         networkx.XDiGraph.__init__(self)
         self.by_id = {}
+        self.undirected = None
 
     def add_term(self, term):
         self.add_node(term)
@@ -62,12 +63,12 @@ class GeneOntology(networkx.XDiGraph):
 
     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 self.undirected == None:
+            self.undirected = self.to_undirected()
+
+        path = networkx.shortest_path(self.undirected, t1, t2)
         if not path:
             print "Woah, path not found."
             return None
@@ -76,12 +77,9 @@ class GeneOntology(networkx.XDiGraph):
             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"

workflows/go_workflow.py

@@ -120,6 +120,7 @@ class GoWorkflow (workflow.Workflow):
         go = workflow.Stage('go', 'Gene Ontology')
         go.add_function(SelectGoTermsFunction(self))
         go.add_function(GoDistanceFunction())
+        go.add_function(SaveDistancesFunction())
         self.add_stage(go)
 
        
@@ -277,7 +278,7 @@ class SelectGoTermsFunction(workflow.Function):
         self.wf = wf
 
     def run(self):
-        self.wf.project.set_selection('go-terms', set(['GO:0007582', 'GO:0008150', 'GO:0051704']))
+        self.wf.project.set_selection('go-terms', set(['GO:0007582', 'GO:0008150', 'GO:0051704', 'GO:0044419']))
 
 
 class GoDistanceFunction(workflow.Function):
@@ -296,7 +297,7 @@ class GoDistanceFunction(workflow.Function):
                 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]
+                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
 
@@ -322,11 +323,10 @@ class GoDistanceFunction(workflow.Function):
         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']])
 
         # Create information content dictionary
-        for i, v in enumerate(evidence.get_identifiers('go-terms')):
-            ic[v] = -log(ann_count_matrix[i] / total_ann)
+        for term, count in annotations.items():
+            ic[term] = -log(count / total_ann)
 
         return [self.resnik_distance_matrix(selection, ic)]
         
@@ -343,6 +343,17 @@ class GoDistanceFunction(workflow.Function):
             return options
 
 
+class SaveDistancesFunction(workflow.Function):
+    def __init__(self):
+        workflow.Function.__init__(self, 'save-matrix', 'Save Matrix')
+    
+    def run(self, ds):
+        filename = '/home/einarr/data/output.ftsv'
+        fd = open(filename, 'w')
+        dataset.write_ftsv(fd, ds)
+        fd.close()
+
+
 class Options(dict):
     def __init__(self):
         dict.__init__(self)