class OWBioMart(widget.OWWidget):
name = "BioMart"
description = "Query BioMart service"
icon = "../widgets/icons/BioMart.svg"
priority = 2010
outputs = [("Data", Orange.data.Table)]
SHOW_FILTERS = True
selectedService = settings.Setting(MartServices[0][1])
selectedDataset = settings.Setting(0)
def __init__(self, parent=None):
super().__init__(parent)
self.selectedDatabase = 0
self.uniqueRows = True
gui.button(gui.widgetBox(self.controlArea, "Cache", addSpace=True),
self, "Clear cache",
tooltip="Clear saved query results",
callback=self.clearCache)
self.serviceindex = 0
self.serviceCombo = gui.comboBox(
self.controlArea, self, "serviceindex", "Mart Service",
callback=self._setServiceUrl
)
for name, url in MartServices:
self.serviceCombo.addItem(name, userData=url)
idx = self.serviceCombo.findData(self.selectedService, Qt.UserRole)
self.serviceCombo.setCurrentIndex(idx)
# self.selectedService = self.serviceCombo.itemData(self.serviceCombo.currentItem())
self.martsCombo = gui.comboBox(
self.controlArea, self, "selectedDatabase", "Database",
callback=self.setSelectedMart,
addSpace=True)
self.martsCombo.setMaximumWidth(250)
self.datasetsCombo = gui.comboBox(
self.controlArea, self, "selectedDataset", "Dataset",
callback=self.setSelectedDataset,
addSpace=True)
self.datasetsCombo.setMaximumWidth(250)
gui.rubber(self.controlArea)
box = gui.widgetBox(self.controlArea, "Results")
gui.checkBox(
box, self, "uniqueRows", "Unique results only",
tooltip="Return unique results only.",)
self.commitButton = gui.button(
box, self, "Get Results", callback=self.commit,
tooltip="Query the BioMart server and output the results",
autoDefault=True)
self.commitButton.setEnabled(False)
self.mainWidget = gui.widgetBox(
self.mainArea, orientation=QStackedLayout())
self.mainTab = QTabWidget()
self.mainWidget.layout().addWidget(self.mainTab)
self.attributesConfigurationBox = gui.createTabPage(self.mainTab, "Attributes")
if self.SHOW_FILTERS: # ??
self.filtersConfigurationBox = gui.createTabPage(self.mainTab, "Filters")
self.error(0)
self.setEnabled(False)
self._task = None
self._executor = concurrent.ThreadExecutor(
threadPool=QThreadPool(maxThreadCount=2)
)
service = self.selectedService
self._task = task = concurrent.Task(
function=partial(self._get_registry, url=service))
task.resultReady.connect(self.setBioMartRegistry)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
self._setServiceUrl()
self._afterInitQueue = []
try:
from Bio import SeqIO
self.hasBiopython = True
except ImportError:
self.warning(100, "Biopython package not found.\nTo retrieve FASTA sequence data from BioMart install Biopython.")
self.hasBiopython = False
def sizeHint(self):
return QSize(800, 600)
def _setServiceUrl(self):
service = self.serviceCombo.itemData(self.serviceCombo.currentIndex())
if service is not None:
self.selectedService = service
self._task = task = concurrent.Task(
function=partial(self._get_registry, url=service))
task.resultReady.connect(self.setBioMartRegistry)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
@staticmethod
def _get_registry(url=None, precache=True):
if url is None:
url = MartServices[0][1]
con = biomart.BioMartConnection(address=url, timeout=30)
reg = biomart.BioMartRegistry(con)
if precache:
_ = reg.marts()
return reg
@Slot(Exception)
def _handleException(self, exception):
assert(QThread.currentThread() is self.thread())
print("Task failed with:", exception, file=sys.stderr)
import logging
log = logging.getLogger(__name__)
log.exception("Error:", exc_info=exception)
self.error(0, str(exception))
self.setEnabled(True)
@Slot(object)
def setBioMartRegistry(self, registry):
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
self.registry = registry
self.marts = [mart for mart in self.registry.marts()
if getattr(mart, "visible", "0") != "0"]
self.martsCombo.clear()
for mart in self.marts:
self.martsCombo.addItem(mart.displayName)
def setSelectedMart(self):
self.mart = self.marts[self.selectedDatabase]
self.error(0)
self.setEnabled(False)
self._task = task = concurrent.Task(function=self.mart.datasets)
task.resultReady.connect(self.setBioMartDatasets)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
@Slot(object)
def setBioMartDatasets(self, datasets):
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
self.datasets = [data for data in datasets if
getattr(data, "visible", "0") != "0"]
self.datasetsCombo.clear()
self.datasetsCombo.addItems([data.displayName for data in self.datasets])
def setSelectedDataset(self):
self.dataset = self.datasets[self.selectedDataset]
self.error(0)
self.setEnabled(False)
def get_configuration(dataset):
connection = dataset.connection
stream = connection.configuration(
dataset=dataset.internalName,
virtualSchema=dataset.serverVirtualSchema)
response = stream.read()
return response
self._task = task = concurrent.Task(
function=partial(get_configuration, self.dataset))
task.resultReady.connect(self.setBioMartConfiguration)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
@Slot(object)
def setBioMartConfiguration(self, configuration):
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
# parse the xml in the main thread (a long time ago this step was
# done in a thread but would frequently cause `expat` to segfault.
doc = biomart.parseXML(io.BytesIO(configuration))
config = list(doc.elements("DatasetConfig"))[0]
configuration = biomart.DatasetConfig(
self.registry, config.tag, config.attributes, config.children)
self.clearConfiguration()
self.configuration = configuration
def hidden(tree):
return getattr(tree, "hidden", "false") != "false" or \
getattr(tree, "hideDisplay", "false") != "false"
self.attributePagesTabWidget = tabs = gui.tabWidget(self.attributesConfigurationBox)
for page in configuration.elements("AttributePage"):
if not hidden(page):
page_widget = PageWidget(page, self.dataset, self)
gui.createTabPage(tabs, getattr(page, "displayName", ""),
widgetToAdd=page_widget, canScroll=True)
if self.SHOW_FILTERS:
self.filterPagesTabWidget = tabs = gui.tabWidget(self.filtersConfigurationBox)
for page in configuration.elements("FilterPage"):
if not hidden(page):
page_widget = PageWidget(page, self.dataset, self)
gui.createTabPage(tabs, getattr(page, "displayName", ""),
widgetToAdd=page_widget, canScroll=True)
self.afterInit()
self.commitButton.setEnabled(True)
def clearConfiguration(self):
self.mainTab.deleteLater()
self.mainTab = QTabWidget()
self.mainWidget.layout().addWidget(self.mainTab)
self.mainWidget.layout().setCurrentWidget(self.mainTab)
self.attributesConfigurationBox = gui.createTabPage(self.mainTab, "Attributes")
if self.SHOW_FILTERS:
self.filtersConfigurationBox = gui.createTabPage(self.mainTab, "Filters")
def commit(self):
pageconf = self.attributePagesTabWidget.currentWidget().widget()
format = pageconf.outFormats
self.error(100)
if not self.hasBiopython and format.lower() == "fasta":
self.error(100, "Cannot parse FASTA format")
return
query = pageconf.query()
bydatasets = defaultdict(lambda: ([], []))
for conftype, tree, val in query:
dataset = self.dataset
if conftype == "Attribute":
bydatasets[dataset][0].append(tree.internalName)
elif conftype == "Filter":
bydatasets[dataset][1].append((tree.internalName, val))
if self.SHOW_FILTERS:
pageconf = self.filterPagesTabWidget.currentWidget().widget()
query = pageconf.query()
for conftype, tree, val in query:
dataset = self.dataset
if conftype == "Attribute":
bydatasets[dataset][0].append(tree.internalName)
elif conftype == "Filter":
bydatasets[dataset][1].append((tree.internalName, val))
query = self.registry.query(
format="TSV" if "tsv" in format.lower() else format.upper(),
uniqueRows=self.uniqueRows,
virtualSchema=dataset.virtualSchema,
serverVirtualSchema=dataset.serverVirtualSchema
)
for dataset, (attributes, filters) in bydatasets.items():
query.set_dataset(dataset if dataset else self.dataset)
for attr in attributes:
query.add_attribute(attr)
for filter, value in filters:
query.add_filter(filter, value)
self.error(0)
self.setEnabled(False)
self._task = task = concurrent.Task(function=query.get_table)
task.resultReady.connect(self.dataReady)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
def dataReady(self, data):
self.setEnabled(True)
self.send("Data", data)
def pushAction(self, action):
ref = action.ref
ref_widget = self.findChild(QWidget, ref)
if hasattr(ref_widget, "setOptions"):
ref_widget.setOptions(action.subelements_top("Option"))
def registerDelayedCall(self, call):
self._afterInitQueue.append(call)
def afterInit(self):
while self._afterInitQueue:
call = self._afterInitQueue.pop(0)
call()
def clearCache(self):
self.registry.connection.clear_cache()
class OWBioMart(widget.OWWidget):
name = "BioMart"
description = "Query BioMart service"
icon = "../widgets/icons/BioMart.svg"
priority = 2010
outputs = [("Data", Orange.data.Table)]
SHOW_FILTERS = True
selectedService = settings.Setting(MartServices[0][1])
selectedDataset = settings.Setting(0)
def __init__(self, parent=None):
super().__init__(parent)
self.selectedDatabase = 0
self.uniqueRows = True
gui.button(gui.widgetBox(self.controlArea, "Cache", addSpace=True),
self, "Clear cache",
tooltip="Clear saved query results",
callback=self.clearCache)
self.serviceindex = 0
self.serviceCombo = gui.comboBox(
self.controlArea, self, "serviceindex", "Mart Service",
callback=self._setServiceUrl
)
for name, url in MartServices:
self.serviceCombo.addItem(name, userData=url)
idx = self.serviceCombo.findData(self.selectedService, Qt.UserRole)
self.serviceCombo.setCurrentIndex(idx)
# self.selectedService = self.serviceCombo.itemData(self.serviceCombo.currentItem())
self.martsCombo = gui.comboBox(
self.controlArea, self, "selectedDatabase", "Database",
callback=self.setSelectedMart,
addSpace=True)
self.martsCombo.setMaximumWidth(250)
self.datasetsCombo = gui.comboBox(
self.controlArea, self, "selectedDataset", "Dataset",
callback=self.setSelectedDataset,
addSpace=True)
self.datasetsCombo.setMaximumWidth(250)
gui.rubber(self.controlArea)
box = gui.widgetBox(self.controlArea, "Results")
gui.checkBox(
box, self, "uniqueRows", "Unique results only",
tooltip="Return unique results only.",)
self.commitButton = gui.button(
box, self, "Get Results", callback=self.commit,
tooltip="Query the BioMart server and output the results",
autoDefault=True)
self.commitButton.setEnabled(False)
self.mainWidget = gui.widgetBox(
self.mainArea, orientation=QStackedLayout())
self.mainTab = QTabWidget()
self.mainWidget.layout().addWidget(self.mainTab)
self.attributesConfigurationBox = gui.createTabPage(self.mainTab, "Attributes")
if self.SHOW_FILTERS: # ??
self.filtersConfigurationBox = gui.createTabPage(self.mainTab, "Filters")
self.error(0)
self.setEnabled(False)
self._task = None
self._executor = concurrent.ThreadExecutor(
threadPool=QThreadPool(maxThreadCount=2)
)
service = self.selectedService
self._task = task = concurrent.Task(
function=partial(self._get_registry, url=service))
task.resultReady.connect(self.setBioMartRegistry)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
self._setServiceUrl()
self._afterInitQueue = []
try:
from Bio import SeqIO
self.hasBiopython = True
except ImportError:
self.warning(100, "Biopython package not found.\nTo retrieve FASTA sequence data from BioMart install Biopython.")
self.hasBiopython = False
def sizeHint(self):
return QSize(800, 600)
def _setServiceUrl(self):
service = self.serviceCombo.itemData(self.serviceCombo.currentIndex())
if service is not None:
self.selectedService = service
self._task = task = concurrent.Task(
function=partial(self._get_registry, url=service))
task.resultReady.connect(self.setBioMartRegistry)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
@staticmethod
def _get_registry(url=None, precache=True):
if url is None:
url = MartServices[0][1]
con = biomart.BioMartConnection(address=url, timeout=30)
reg = biomart.BioMartRegistry(con)
if precache:
_ = reg.marts()
return reg
@Slot(Exception)
def _handleException(self, exception):
assert(QThread.currentThread() is self.thread())
print("Task failed with:", exception, file=sys.stderr)
import logging
log = logging.getLogger(__name__)
log.exception("Error:", exc_info=exception)
self.error(0, str(exception))
self.setEnabled(True)
@Slot(object)
def setBioMartRegistry(self, registry):
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
self.registry = registry
self.marts = [mart for mart in self.registry.marts()
if getattr(mart, "visible", "0") != "0"]
self.martsCombo.clear()
for mart in self.marts:
self.martsCombo.addItem(mart.displayName)
def setSelectedMart(self):
self.mart = self.marts[self.selectedDatabase]
self.error(0)
self.setEnabled(False)
self._task = task = concurrent.Task(function=self.mart.datasets)
task.resultReady.connect(self.setBioMartDatasets)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
@Slot(object)
def setBioMartDatasets(self, datasets):
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
self.datasets = [data for data in datasets if
getattr(data, "visible", "0") != "0"]
self.datasetsCombo.clear()
self.datasetsCombo.addItems([data.displayName for data in self.datasets])
def setSelectedDataset(self):
self.dataset = self.datasets[self.selectedDataset]
self.error(0)
self.setEnabled(False)
def get_configuration(dataset):
connection = dataset.connection
stream = connection.configuration(
dataset=dataset.internalName,
virtualSchema=dataset.serverVirtualSchema)
response = stream.read()
return response
self._task = task = concurrent.Task(
function=partial(get_configuration, self.dataset))
task.resultReady.connect(self.setBioMartConfiguration)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
@Slot(object)
def setBioMartConfiguration(self, configuration):
assert(QThread.currentThread() is self.thread())
self.setEnabled(True)
# parse the xml in the main thread (a long time ago this step was
# done in a thread but would frequently cause `expat` to segfault.
doc = biomart.parseXML(io.BytesIO(configuration))
config = list(doc.elements("DatasetConfig"))[0]
configuration = biomart.DatasetConfig(
self.registry, config.tag, config.attributes, config.children)
self.clearConfiguration()
self.configuration = configuration
def hidden(tree):
return getattr(tree, "hidden", "false") != "false" or \
getattr(tree, "hideDisplay", "false") != "false"
self.attributePagesTabWidget = tabs = gui.tabWidget(self.attributesConfigurationBox)
for page in configuration.elements("AttributePage"):
if not hidden(page):
page_widget = PageWidget(page, self.dataset, self)
gui.createTabPage(tabs, getattr(page, "displayName", ""),
widgetToAdd=page_widget, canScroll=True)
if self.SHOW_FILTERS:
self.filterPagesTabWidget = tabs = gui.tabWidget(self.filtersConfigurationBox)
for page in configuration.elements("FilterPage"):
if not hidden(page):
page_widget = PageWidget(page, self.dataset, self)
gui.createTabPage(tabs, getattr(page, "displayName", ""),
widgetToAdd=page_widget, canScroll=True)
self.afterInit()
self.commitButton.setEnabled(True)
def clearConfiguration(self):
self.mainTab.deleteLater()
self.mainTab = QTabWidget()
self.mainWidget.layout().addWidget(self.mainTab)
self.mainWidget.layout().setCurrentWidget(self.mainTab)
self.attributesConfigurationBox = gui.createTabPage(self.mainTab, "Attributes")
if self.SHOW_FILTERS:
self.filtersConfigurationBox = gui.createTabPage(self.mainTab, "Filters")
def commit(self):
pageconf = self.attributePagesTabWidget.currentWidget().widget()
format = pageconf.outFormats
self.error(100)
if not self.hasBiopython and format.lower() == "fasta":
self.error(100, "Cannot parse FASTA format")
return
query = pageconf.query()
bydatasets = defaultdict(lambda: ([], []))
for conftype, tree, val in query:
dataset = self.dataset
if conftype == "Attribute":
bydatasets[dataset][0].append(tree.internalName)
elif conftype == "Filter":
bydatasets[dataset][1].append((tree.internalName, val))
if self.SHOW_FILTERS:
pageconf = self.filterPagesTabWidget.currentWidget().widget()
query = pageconf.query()
for conftype, tree, val in query:
dataset = self.dataset
if conftype == "Attribute":
bydatasets[dataset][0].append(tree.internalName)
elif conftype == "Filter":
bydatasets[dataset][1].append((tree.internalName, val))
query = self.registry.query(
format="TSV" if "tsv" in format.lower() else format.upper(),
uniqueRows=self.uniqueRows,
virtualSchema=dataset.virtualSchema,
serverVirtualSchema=dataset.serverVirtualSchema
)
for dataset, (attributes, filters) in bydatasets.items():
query.set_dataset(dataset if dataset else self.dataset)
for attr in attributes:
query.add_attribute(attr)
for filter, value in filters:
query.add_filter(filter, value)
self.error(0)
self.setEnabled(False)
self._task = task = concurrent.Task(function=query.get_table)
task.resultReady.connect(self.dataReady)
task.exceptionReady.connect(self._handleException)
self._executor.submit(task)
def dataReady(self, data):
self.setEnabled(True)
self.send("Data", data)
def pushAction(self, action):
ref = action.ref
ref_widget = self.findChild(QWidget, ref)
if hasattr(ref_widget, "setOptions"):
ref_widget.setOptions(action.subelements_top("Option"))
def registerDelayedCall(self, call):
self._afterInitQueue.append(call)
def afterInit(self):
while self._afterInitQueue:
call = self._afterInitQueue.pop(0)
call()
def clearCache(self):
self.registry.connection.clear_cache()