class OWMAPlot(OWWidget):
settingsList = ["appendZScore", "appendRIValues"]
contextHandlers = {
"": DomainContextHandler("", ["selectedGroup", "zCutoff"])
}
CENTER_METHODS = [("Average", obiExpression.MA_center_average),
("Lowess (fast - interpolated)",
obiExpression.MA_center_lowess_fast),
("Lowess", obiExpression.MA_center_lowess)]
MERGE_METHODS = [("Average", numpy.ma.average),
("Median", numpy.ma.median),
("Geometric mean", obiExpression.geometric_mean)]
def __init__(self,
parent=None,
signalManager=None,
name="Normalize Expression Array"):
OWWidget.__init__(self, parent, signalManager, name, wantGraph=True)
self.inputs = [("Expression array", ExampleTable, self.setData)]
self.outputs = [("Normalized expression array", ExampleTable, Default),
("Filtered expression array", ExampleTable)]
self.selectedGroup = 0
self.selectedCenterMethod = 0
self.selectedMergeMethod = 0
self.zCutoff = 1.96
self.appendZScore = False
self.appendRIValues = False
self.autoCommit = False
self.loadSettings()
## GUI
self.infoBox = OWGUI.widgetLabel(
OWGUI.widgetBox(self.controlArea, "Info", addSpace=True),
"No data on input.")
box = OWGUI.widgetBox(self.controlArea, "Split by", addSpace=True)
self.groupCombo = OWGUI.comboBox(box,
self,
"selectedGroup",
callback=self.onGroupSelection)
self.centerCombo = OWGUI.comboBox(
self.controlArea,
self,
"selectedCenterMethod",
box="Center Fold-change Using",
items=[name for name, _ in self.CENTER_METHODS],
callback=self.onCenterMethodChange,
addSpace=True)
self.mergeCombo = OWGUI.comboBox(
self.controlArea,
self,
"selectedMergeMethod",
box="Merge Replicates",
items=[name for name, _ in self.MERGE_METHODS],
tooltip="Select the method for replicate merging",
callback=self.onMergeMethodChange,
addSpace=True)
box = OWGUI.doubleSpin(self.controlArea,
self,
"zCutoff",
0.0,
3.0,
0.01,
box="Z-Score Cutoff",
callback=[self.replotMA, self.commitIf])
OWGUI.separator(self.controlArea)
box = OWGUI.widgetBox(self.controlArea, "Ouput")
OWGUI.checkBox(box,
self,
"appendZScore",
"Append Z-Scores",
tooltip="Append calculated Z-Scores to output",
callback=self.commitIf)
OWGUI.checkBox(
box,
self,
"appendRIValues",
"Append Log Ratio and Intensity values",
tooltip=
"Append calculated Log Ratio and Intensity values to output data",
callback=self.commitIf)
cb = OWGUI.checkBox(box,
self,
"autoCommit",
"Commit on change",
tooltip="Commit data on any change",
callback=self.commitIf)
b = OWGUI.button(box, self, "Commit", callback=self.commit)
OWGUI.setStopper(self, b, cb, "changedFlag", callback=self.commit)
self.connect(self.graphButton, SIGNAL("clicked()"), self.saveGraph)
OWGUI.rubber(self.controlArea)
self.graph = OWGraph(self.mainArea)
self.graph.setAxisTitle(QwtPlot.xBottom,
"Intensity: log<sub>10</sub>(R*G)")
self.graph.setAxisTitle(QwtPlot.yLeft,
"Log ratio: log<sub>2</sub>(R/G)")
self.graph.showFilledSymbols = True
self.mainArea.layout().addWidget(self.graph)
self.groups = []
self.split_data = None, None
self.merged_splits = None, None
self.centered = None, None
self.changedFlag = False
self.data = None
self.resize(800, 600)
def onFinished(self, status):
self.setEnabled(True)
def onUnhandledException(self, ex_info):
self.setEnabled(True)
print >> sys.stderr, "Unhandled exception in non GUI thread"
ex_type, ex_val, tb = ex_info
if ex_type == numpy.linalg.LinAlgError and False:
self.error(0, "Linear algebra error: %s" % repr(ex_val))
else:
sys.excepthook(*ex_info)
def onGroupSelection(self):
if self.data:
self.updateInfoBox()
self.splitData()
self.runNormalization()
def onCenterMethodChange(self):
if self.data:
self.runNormalization()
def onMergeMethodChange(self):
if self.data:
self.splitData()
self.runNormalization()
def proposeGroups(self, data):
col_labels = [
attr.attributes.items() for attr in data.domain.attributes
]
col_labels = sorted(reduce(set.union, col_labels, set()))
col_labels = [(key, value, 1) for key, value in col_labels]
attrs = [attr for attr in data.domain.variables + data.domain.getmetas().values() \
if attr.varType == orange.VarTypes.Discrete]
row_labels = [(attr.name, value, 0) for attr in attrs
for value in attr.values]
def filterSingleValues(labels):
ret = []
for name, value, axis in labels:
match = [(n, v, a) for n, v, a in labels if n == name]
if len(match) > 1:
ret.append((name, value, axis))
return ret
col_labels = filterSingleValues(col_labels)
row_labels = filterSingleValues(row_labels)
return col_labels + row_labels
def setData(self, data):
self.closeContext("")
self.data = data
self.error([0, 1])
if data is not None:
self.infoBox.setText("%i genes on input" % len(data))
self.groups = self.proposeGroups(data)
self.groupCombo.clear()
self.groupCombo.addItems(
["%s: %s" % (key, value) for key, value, axis in self.groups])
if not self.groups:
self.error(
1,
"Input data has no class attribute or attribute labels!")
self.clear()
return
self.openContext("", data)
self.selectedGroup = min(self.selectedGroup, len(self.groups) - 1)
self.updateInfoBox()
self.splitData()
self.runNormalization()
else:
self.clear()
def clear(self):
self.groups = []
self.data = None
self.centered = None, None
self.split_data = None, None
self.merged_splits = None, None
self.graph.removeDrawingCurves()
self.infoBox.setText("No data on input")
self.send("Normalized expression array", None)
self.send("Filtered expression array", None)
def updateInfoBox(self):
genes = self.getGeneNames()
self.infoBox.setText("%i genes on input" % len(self.data))
def getSelectedGroup(self):
return self.groups[self.selectedGroup]
def getSelectedGroupSplit(self):
key, value, axis = self.getSelectedGroup()
other_values = [
v for k, v, a in self.groups
if k == key and a == axis and v != value
]
return [(key, value), (key, other_values)], axis
def getGeneNames(self):
key, value, axis = self.getSelectedGroup()
if axis == 0:
genes = [str(ex[key]) for ex in self.data]
else:
genes = [attr.name for attr in self.data.domain.attributes]
return genes
def splitData(self):
groups, axis = self.getSelectedGroupSplit()
self.split_ind = [
obiExpression.select_indices(self.data, key, value, axis)
for key, value in groups
]
self.split_data = obiExpression.split_data(self.data, groups, axis)
def getMerged(self):
split1, split2 = self.split_data
(array1, _, _), (array2, _, _) = split1.toNumpyMA(), split2.toNumpyMA()
_, _, axis = self.getSelectedGroup()
merge_function = self.MERGE_METHODS[self.selectedMergeMethod][1]
merged1 = obiExpression.merge_replicates(array1,
axis,
merge_function=merge_function)
merged2 = obiExpression.merge_replicates(array2,
axis,
merge_function=merge_function)
self.merged_splits = merged1, merged2
return self.merged_splits
def runNormalization(self):
self.progressBarInit()
self.progressBarSet(0.0)
G, R = self.getMerged()
self.progressBarSet(5.0)
center_method = self.CENTER_METHODS[self.selectedCenterMethod][1]
# TODO: progess bar , lowess can take a long time
if self.selectedCenterMethod in [1, 2]: #Lowess
Gc, Rc = center_method(G,
R,
f=1. / min(500.,
len(G) / 100),
iter=1)
else:
Gc, Rc = center_method(G, R)
self.progressBarSet(70.0)
self.centered = Gc, Rc
self.z_scores = obiExpression.MA_zscore(Gc, Rc, 1. / 3.)
self.progressBarSet(100.0)
self.plotMA(Gc, Rc, self.z_scores, self.zCutoff)
self.progressBarFinished()
def runNormalizationAsync(self):
""" Run MA centering and z_score estimation in a separate thread
"""
self.error(0)
self.progressBarInit()
self.progressBarSet(0.0)
G, R = self.getMerged()
# self.progressBarSet(5.0)
center_method = self.CENTER_METHODS[self.selectedCenterMethod][1]
use_lowess = self.selectedCenterMethod in [1, 2]
def run(progressCallback=lambda value: None
): # the function to run in a thread
# progressCallback(5.0)
if use_lowess:
Gc, Rc = center_method(
G,
R,
f=2. / 3.,
iter=1,
progressCallback=lambda val: progressCallback(val / 2))
else:
Gc, Rc = center_method(G, R)
progressCallback(50)
z_scores = obiExpression.MA_zscore(
Gc,
Rc,
1. / 3.,
progressCallback=lambda val: progressCallback(50 + val / 2))
return Gc, Rc, z_scores
self.progressDiscard = ProgressBarDiscard(self, self)
async = self.asyncCall(run,
name="Normalization",
onResult=self.onResults,
onError=self.onUnhandledException)
self.connect(async, SIGNAL("progressChanged(float)"),
self.progressDiscard.progressBarSet, Qt.QueuedConnection)
self.setEnabled(False)
async .__call__(progressCallback=async .emitProgressChanged)
## comment out this line if threading creates any problems
runNormalization = runNormalizationAsync
def onResults(self, (Gc, Rc, z_scores)):
""" Handle the results of centering and z-scoring
"""
assert (QThread.currentThread() is self.thread())
self.setEnabled(True)
qApp.processEvents()
self.progressBarFinished()
self.centered = Gc, Rc
self.z_scores = z_scores
self.plotMA(Gc, Rc, z_scores, self.zCutoff)
self.commit()
class OWMAPlot(OWWidget):
settingsList = ["appendZScore", "appendRIValues"]
contextHandlers = {"": DomainContextHandler("", ["selectedGroup", "zCutoff"])}
CENTER_METHODS = [("Average", obiExpression.MA_center_average),
("Lowess (fast - interpolated)", obiExpression.MA_center_lowess_fast),
("Lowess", obiExpression.MA_center_lowess)]
MERGE_METHODS = [("Average", numpy.ma.average),
("Median", numpy.ma.median),
("Geometric mean", obiExpression.geometric_mean)]
def __init__(self, parent=None, signalManager=None, name="Normalize Expression Array"):
OWWidget.__init__(self, parent, signalManager, name, wantGraph=True)
self.inputs = [("Expression array", ExampleTable, self.setData)]
self.outputs = [("Normalized expression array", ExampleTable, Default), ("Filtered expression array", ExampleTable)]
self.selectedGroup = 0
self.selectedCenterMethod = 0
self.selectedMergeMethod = 0
self.zCutoff = 1.96
self.appendZScore = False
self.appendRIValues = False
self.autoCommit = False
self.loadSettings()
## GUI
self.infoBox = OWGUI.widgetLabel(OWGUI.widgetBox(self.controlArea, "Info", addSpace=True),
"No data on input.")
box = OWGUI.widgetBox(self.controlArea, "Split by", addSpace=True)
self.groupCombo = OWGUI.comboBox(box, self, "selectedGroup",
callback=self.onGroupSelection
)
self.centerCombo = OWGUI.comboBox(self.controlArea, self, "selectedCenterMethod",
box="Center Fold-change Using",
items=[name for name, _ in self.CENTER_METHODS],
callback=self.onCenterMethodChange,
addSpace=True
)
self.mergeCombo = OWGUI.comboBox(self.controlArea, self, "selectedMergeMethod",
box="Merge Replicates",
items=[name for name, _ in self.MERGE_METHODS],
tooltip="Select the method for replicate merging",
callback=self.onMergeMethodChange,
addSpace=True
)
box = OWGUI.doubleSpin(self.controlArea, self, "zCutoff", 0.0, 3.0, 0.01,
box="Z-Score Cutoff",
callback=[self.replotMA, self.commitIf])
OWGUI.separator(self.controlArea)
box = OWGUI.widgetBox(self.controlArea, "Ouput")
OWGUI.checkBox(box, self, "appendZScore", "Append Z-Scores",
tooltip="Append calculated Z-Scores to output",
callback=self.commitIf
)
OWGUI.checkBox(box, self, "appendRIValues", "Append Log Ratio and Intensity values",
tooltip="Append calculated Log Ratio and Intensity values to output data",
callback=self.commitIf
)
cb = OWGUI.checkBox(box, self, "autoCommit", "Commit on change",
tooltip="Commit data on any change",
callback=self.commitIf
)
b = OWGUI.button(box, self, "Commit", callback=self.commit)
OWGUI.setStopper(self, b, cb, "changedFlag", callback=self.commit)
self.connect(self.graphButton, SIGNAL("clicked()"), self.saveGraph)
OWGUI.rubber(self.controlArea)
self.graph = OWGraph(self.mainArea)
self.graph.setAxisTitle(QwtPlot.xBottom, "Intensity: log<sub>10</sub>(R*G)")
self.graph.setAxisTitle(QwtPlot.yLeft, "Log ratio: log<sub>2</sub>(R/G)")
self.graph.showFilledSymbols = True
self.mainArea.layout().addWidget(self.graph)
self.groups = []
self.split_data = None, None
self.merged_splits = None, None
self.centered = None, None
self.changedFlag = False
self.data = None
self.resize(800, 600)
def onFinished(self, status):
self.setEnabled(True)
def onUnhandledException(self, ex_info):
self.setEnabled(True)
print >> sys.stderr, "Unhandled exception in non GUI thread"
ex_type, ex_val, tb = ex_info
if ex_type == numpy.linalg.LinAlgError and False:
self.error(0, "Linear algebra error: %s" % repr(ex_val))
else:
sys.excepthook(*ex_info)
def onGroupSelection(self):
if self.data:
self.updateInfoBox()
self.splitData()
self.runNormalization()
def onCenterMethodChange(self):
if self.data:
self.runNormalization()
def onMergeMethodChange(self):
if self.data:
self.splitData()
self.runNormalization()
def proposeGroups(self, data):
col_labels = [attr.attributes.items() for attr in data.domain.attributes]
col_labels = sorted(reduce(set.union, col_labels, set()))
col_labels = [(key, value, 1) for key, value in col_labels]
attrs = [attr for attr in data.domain.variables + data.domain.getmetas().values() \
if attr.varType == orange.VarTypes.Discrete]
row_labels = [(attr.name, value, 0) for attr in attrs for value in attr.values]
def filterSingleValues(labels):
ret = []
for name, value, axis in labels:
match = [(n, v, a) for n, v, a in labels if n == name]
if len(match) > 1:
ret.append((name, value, axis))
return ret
col_labels = filterSingleValues(col_labels)
row_labels = filterSingleValues(row_labels)
return col_labels + row_labels
def setData(self, data):
self.closeContext("")
self.data = data
self.error([0 ,1])
if data is not None:
self.infoBox.setText("%i genes on input" % len(data))
self.groups = self.proposeGroups(data)
self.groupCombo.clear()
self.groupCombo.addItems(["%s: %s" % (key, value) for key, value, axis in self.groups])
if not self.groups:
self.error(1, "Input data has no class attribute or attribute labels!")
self.clear()
return
self.openContext("", data)
self.selectedGroup = min(self.selectedGroup, len(self.groups) - 1)
self.updateInfoBox()
self.splitData()
self.runNormalization()
else:
self.clear()
def clear(self):
self.groups = []
self.data = None
self.centered = None, None
self.split_data = None, None
self.merged_splits = None, None
self.graph.removeDrawingCurves()
self.infoBox.setText("No data on input")
self.send("Normalized expression array", None)
self.send("Filtered expression array", None)
def updateInfoBox(self):
genes = self.getGeneNames()
self.infoBox.setText("%i genes on input" % len(self.data))
def getSelectedGroup(self):
return self.groups[self.selectedGroup]
def getSelectedGroupSplit(self):
key, value, axis = self.getSelectedGroup()
other_values = [v for k, v, a in self.groups if k == key and a == axis and v != value]
return [(key, value), (key, other_values)], axis
def getGeneNames(self):
key, value, axis = self.getSelectedGroup()
if axis == 0:
genes = [str(ex[key]) for ex in self.data]
else:
genes = [attr.name for attr in self.data.domain.attributes]
return genes
def splitData(self):
groups, axis = self.getSelectedGroupSplit()
self.split_ind = [obiExpression.select_indices(self.data, key, value, axis) for key, value in groups]
self.split_data = obiExpression.split_data(self.data, groups, axis)
def getMerged(self):
split1, split2 = self.split_data
(array1, _, _), (array2, _, _) = split1.toNumpyMA(), split2.toNumpyMA()
_, _, axis = self.getSelectedGroup()
merge_function = self.MERGE_METHODS[self.selectedMergeMethod][1]
merged1 = obiExpression.merge_replicates(array1, axis, merge_function=merge_function)
merged2 = obiExpression.merge_replicates(array2, axis, merge_function=merge_function)
self.merged_splits = merged1, merged2
return self.merged_splits
def runNormalization(self):
self.progressBarInit()
self.progressBarSet(0.0)
G, R = self.getMerged()
self.progressBarSet(5.0)
center_method = self.CENTER_METHODS[self.selectedCenterMethod][1]
# TODO: progess bar , lowess can take a long time
if self.selectedCenterMethod in [1, 2]: #Lowess
Gc, Rc = center_method(G, R, f = 1./min(500., len(G)/100), iter=1)
else:
Gc, Rc = center_method(G, R)
self.progressBarSet(70.0)
self.centered = Gc, Rc
self.z_scores = obiExpression.MA_zscore(Gc, Rc, 1./3.)
self.progressBarSet(100.0)
self.plotMA(Gc, Rc, self.z_scores, self.zCutoff)
self.progressBarFinished()
def runNormalizationAsync(self):
""" Run MA centering and z_score estimation in a separate thread
"""
self.error(0)
self.progressBarInit()
self.progressBarSet(0.0)
G, R = self.getMerged()
# self.progressBarSet(5.0)
center_method = self.CENTER_METHODS[self.selectedCenterMethod][1]
use_lowess = self.selectedCenterMethod in [1, 2]
def run(progressCallback = lambda value: None): # the function to run in a thread
# progressCallback(5.0)
if use_lowess:
Gc, Rc = center_method(G, R, f=2./3., iter=1, progressCallback=lambda val: progressCallback(val/2))
else:
Gc, Rc = center_method(G, R)
progressCallback(50)
z_scores = obiExpression.MA_zscore(Gc, Rc, 1./3., progressCallback= lambda val: progressCallback(50 + val/2))
return Gc, Rc, z_scores
self.progressDiscard = ProgressBarDiscard(self, self)
async = self.asyncCall(run, name="Normalization",
onResult=self.onResults,
onError=self.onUnhandledException)
self.connect(async, SIGNAL("progressChanged(float)"), self.progressDiscard.progressBarSet, Qt.QueuedConnection)
self.setEnabled(False)
async.__call__(progressCallback=async.emitProgressChanged)
## comment out this line if threading creates any problems
runNormalization = runNormalizationAsync
def onResults(self, (Gc, Rc, z_scores)):
""" Handle the results of centering and z-scoring
"""
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
qApp.processEvents()
self.progressBarFinished()
self.centered = Gc, Rc
self.z_scores = z_scores
self.plotMA(Gc, Rc, z_scores, self.zCutoff)
self.commit()
