Added pickling as a quick and dirty way of loading and saving data in einars workflow.

workflows/go_workflow.py

@@ -8,6 +8,7 @@ import dataset
 #import gostat
 import rpy
 from scipy import array,randn
+import cPickle
 
 class EinarsWorkflow (Workflow):
 
@@ -19,6 +20,7 @@ class EinarsWorkflow (Workflow):
         load.add_function(Function('load', 'Load Microarrays'))
         load.add_function(CelFileImportFunction())
         load.add_function(TestDataFunction())
+        load.add_function(DatasetLoadFunction())
         self.add_stage(load)
 
         preproc = Stage('preprocess', 'Preprocessing')
@@ -33,6 +35,10 @@ class EinarsWorkflow (Workflow):
         regression = Stage('regression', 'Regression')
         regression.add_function(Function('pls', 'PLS'))
         self.add_stage(regression)
+
+        save = Stage('save', 'Save Data')
+        save.add_function(DatasetSaveFunction())
+        self.add_stage(save)
         
         logger.log('debug', '\tEinar\'s workflow is now active')
 
@@ -102,6 +108,64 @@ class TestDataFunction(Function):
         return [X, plots.SinePlot(None)]
 
 
+class DatasetLoadFunction(Function):
+    """Loader for previously pickled Datasets."""
+    def __init__(self):
+        Function.__init__(self, 'load_data', 'Load Pickled Dataset')
+
+    def run(self, data):
+        chooser = gtk.FileChooserDialog(title="Select cel files...", parent=None,
+                                        action=gtk.FILE_CHOOSER_ACTION_OPEN,
+                                        buttons=(gtk.STOCK_CANCEL, gtk.RESPONSE_CANCEL,
+                                                 gtk.STOCK_OPEN, gtk.RESPONSE_OK))
+        pkl_filter = gtk.FileFilter()
+        pkl_filter.set_name("Python pickled data files (*.pkl)")
+        pkl_filter.add_pattern("*.[pP][kK][lL]")
+        all_filter = gtk.FileFilter()
+        all_filter.set_name("All Files (*.*)")
+        all_filter.add_pattern("*")
+        chooser.add_filter(pkl_filter)
+        chooser.add_filter(all_filter)
+
+        try:
+            if chooser.run() == gtk.RESPONSE_OK:
+                return [cPickle.load(open(chooser.get_filename()))]
+        finally:
+            chooser.destroy()
+
+
+class DatasetSaveFunction(Function):
+    """QND way to save data to file for later import to this program."""
+    def __init__(self):
+        Function.__init__(self, 'save_data', 'Save Pickled Dataset')
+
+    def run(self, data):
+        if not data:
+            logger.log("notice", "No data to save.")
+            return
+
+        chooser = gtk.FileChooserDialog(title="Save pickled data...", parent=None,
+                                        action=gtk.FILE_CHOOSER_ACTION_SAVE,
+                                        buttons=(gtk.STOCK_CANCEL, gtk.RESPONSE_CANCEL,
+                                                 gtk.STOCK_SAVE, gtk.RESPONSE_OK))
+        pkl_filter = gtk.FileFilter()
+        pkl_filter.set_name("Python pickled data files (*.pkl)")
+        pkl_filter.add_pattern("*.[pP][kK][lL]")
+        all_filter = gtk.FileFilter()
+        all_filter.set_name("All Files (*.*)")
+        all_filter.add_pattern("*")
+        chooser.add_filter(pkl_filter)
+        chooser.add_filter(all_filter)
+        chooser.set_current_name(data.get_name() + ".pkl")
+
+        try:
+            if chooser.run() == gtk.RESPONSE_OK:
+                cPickle.dump(data, open(chooser.get_filename(), "w"))
+                logger.log("notice", "Saved data to %r." % chooser.get_filename())
+        finally:
+            chooser.destroy()
+                
+
 class CelFileImportFunction(Function):
     """Loads AffyMetrix .CEL-files into matrix."""
     def __init__(self):