Tralala ...

workflows/smokers.py

@@ -1,5 +1,6 @@
 import sys,os
 import webbrowser
+import cPickle
 
 from fluents import logger, plots,workflow,dataset,main
 from fluents.lib import blmfuncs,nx_utils,validation,engines,cx_stats,cx_utils
@@ -57,6 +58,8 @@ class SmallTestWorkflow(workflow.Workflow):
         go.add_function(gobrowser.PlotDagFunction())
         go.add_function(GoEnrichment())
         go.add_function(GoEnrichmentCond())
+        go.add_function(MapGO2Gene())
+        go.add_function(MapGene2GO())
         self.add_stage(go)
 
         # EXTRA PLOTS
@@ -354,7 +357,7 @@ class GoEnrichment(workflow.Function):
         # Get universe
         # Here, we are using a defined dataset to represent the universe
         if not 'gene_ids' in data:
-            logger.log('notice', 'No dimension called [gene_ids] in dataset: %s', data.get_name())
+            logger.log('notice', 'No dimension called [gene_ids] in dataset: %s' %data.get_name())
             return
         universe = list(data.get_identifiers('gene_ids'))
         logger.log('notice', 'Universe consists of %s gene ids from %s' %(len(universe), data.get_name()))
@@ -417,7 +420,7 @@ class GoEnrichmentCond(workflow.Function):
             return
         
         # Hypergeometric parameter object
-        pval_cutoff = 0.9999
+        pval_cutoff = 1
         cond = True
         test_direction = 'over'
         params = rpy.r.new("GOHyperGParams",
@@ -443,3 +446,65 @@ class GoEnrichmentCond(workflow.Function):
                                name='P values (enrichment)')
         return [xout]
 
+
+class MapGene2GO(workflow.Function):
+    def __init__(self, ont='bp', gene_id_name='gene_ids'):
+        self._ont = ont
+        self._gene_id_name = gene_id_name
+        workflow.Function.__init__(self, 'gene2go', 'gene->GO')
+        # load data at init
+        try:
+            fname = "/home/flatberg/fluents/data/gene2go.pcl"
+            self._gene2go = cPickle.load(open(fname))
+        except:
+            logger.log("notice", "could not load mapping")
+        
+    def run(self):
+        selection = main.project.get_selection()
+        if not selection.has_key(self._gene_id_name):
+            logger.log("notice", "Expected gene ids: %s, but got. %s" %(self._gene_id_name, selection.keys()))
+            return None
+        if len(selection[self._gene_id_name])==0:
+            logger.log("notice", "No selected genes to query")
+            return None
+
+        gene_ids = selection[self._gene_id_name]
+        go_ids = set()
+        for gene in gene_ids:
+            go_ids_new = self._gene2go.get(gene, [])
+            if not go_ids_new:
+                logger.log("notice", "Could not find any goterms for %s" %gene)
+            go_ids.update(self._gene2go.get(gene, []))
+        main.project.set_selection('go-terms', go_ids)
+        logger.log("notice", "GO terms updated")
+
+
+class MapGO2Gene(workflow.Function):
+    def __init__(self, ont='bp', gene_id_name='go-terms'):
+        self._ont = ont
+        self._gene_id_name = gene_id_name
+        workflow.Function.__init__(self, 'go2gene', 'GO->gene')
+        # load data at init
+        try:
+            fname = "/home/flatberg/fluents/data/go2gene.pcl"
+            self._go2gene = cPickle.load(open(fname))
+        except:
+            logger.log("notice", "could not load mapping")
+        
+    def run(self):
+        selection = main.project.get_selection()
+        if not selection.has_key(self._gene_id_name):
+            logger.log("notice", "Expected gene ids: %s, but got. %s" %(self._gene_id_name, selection.keys()))
+            return None
+        if len(selection[self._gene_id_name])==0:
+            logger.log("notice", "No selected genes to query")
+            return None
+
+        go_ids = selection[self._gene_id_name]
+        gene_ids = set()
+        for go in go_ids:
+            if not self._go2gene.get(go,[]):
+                logger.log("notice", "Could not find any gene ids for %s" %go)
+            gene_ids.update(self._go2gene.get(go,[]))
+        main.project.set_selection('gene_ids', gene_ids)
+        logger.log("notice", "GO terms updated")