Updated go-workflow.

workflows/go_workflow.py

@@ -26,6 +26,7 @@ DISTANCE_METRICS = [('resnik', 'Resnik'),
 GO_DATA_DIR = '/home/einarr/data'
 
 evidence = None
+go = None
 
 class GoTermView (gtk.Frame):
 
@@ -137,11 +138,14 @@ class LoadGOFunction(workflow.Function):
 class LoadTextDatasetFunction(workflow.Function):
 
     def __init__(self):
-        workflow.Function.__init__(self, 'load-text-ds', 'Load text dataset')
+        workflow.Function.__init__(self, 'load-text-ds', 'Load GO Evidence')
+
     def run(self):
         f = open('/home/einarr/data/goa-condensed.ftsv')
-        return [dataset.read_ftsv(f)]
-        
+        global evidence
+        evidence = dataset.read_ftsv(f)
+        return [evidence]
+
 
 class LoadAnnotationsFunction(workflow.Function):
 
@@ -253,17 +257,54 @@ class GoDistanceDialog(gtk.Dialog):
         self._metric_frame.set_sensitive(editable)
 
 
+class NumericDict(dict):
+    def __init__(self):
+        dict.__init__(self)
+
+    def __getitem__(self, key):
+        retval = 0
+        try:
+            retval = dict.__getitem__(self, key)
+        except:
+            retval = 0.0
+        return retval
+
 class GoDistanceFunction(workflow.Function):
     def __init__(self):
         workflow.Function.__init__(self, 'go-dist', 'GO term distance matrix')
         self.options = GoDistanceOptions()
         
-    def run(self, selection):
+    def run(self, x, selection):
+        global evidence, go
         self.options = self.show_gui(self.options)
         if not selection.has_key('go-terms') or len(selection['go-terms']) == 0:
             logger.log('warning', 'No GO terms selected. Cannot make distance matrix.')
 
+        codes = [c for c, d in EVIDENCE_CODES if self.options[c]]
+        ev_indices = evidence.get_indices('evidence', codes)
+        ann_count_matrix = evidence._array[:, ev_indices].sum(1)
+        total_ann = ann_count_matrix.sum(0)
 
+        annotations = NumericDict()
+        ic = NumericDict()
+
+        # 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']])
+
+        # 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
+#        for t1 in selection['go-terms']:
+#            for t2 in selection['go-terms']:
+                
+        
+        
     def show_gui(self, options, edit=True):
         dialog = GoDistanceDialog()
         dialog.set_options(self.options)