diff options
| -rw-r--r-- | wqflask/base/data_set.py | 113 | ||||
| -rw-r--r-- | wqflask/base/webqtlCaseData.py | 41 | ||||
| -rw-r--r-- | wqflask/tests/base/data.py | 110 | ||||
| -rw-r--r-- | wqflask/tests/base/test_data_set.py | 135 | ||||
| -rw-r--r-- | wqflask/tests/base/test_webqtl_case_data.py | 39 |
5 files changed, 336 insertions, 102 deletions
diff --git a/wqflask/base/data_set.py b/wqflask/base/data_set.py index 5d562871..cfba9104 100644 --- a/wqflask/base/data_set.py +++ b/wqflask/base/data_set.py @@ -93,7 +93,7 @@ Publish or ProbeSet. E.g. """ self.redis_instance = redis_instance self.datasets = {} - data = redis_instance.get("dataset_structure") + data = self.redis_instance.get("dataset_structure") if data: self.datasets = json.loads(data) else: # ZS: I don't think this should ever run unless Redis is emptied @@ -115,73 +115,58 @@ Publish or ProbeSet. E.g. except: pass - redis_instance.set("dataset_structure", json.dumps(self.datasets)) + self.redis_instance.set("dataset_structure", json.dumps(self.datasets)) - # Set LOG_LEVEL_DEBUG=5 to see the following: - logger.debugf(5, "datasets", self.datasets) + def set_dataset_key(self, t, name): + """If name is not in the object's dataset dictionary, set it, and update + dataset_structure in Redis + + args: + t: Type of dataset structure which can be: 'mrna_expr', 'pheno', + 'other_pheno', 'geno' + name: The name of the key to inserted in the datasets dictionary + + """ + sql_query_mapping = { + 'mrna_expr': ("""SELECT ProbeSetFreeze.Id FROM """ + + """ProbeSetFreeze WHERE ProbeSetFreeze.Name = "{}" """), + 'pheno': ("""SELECT InfoFiles.GN_AccesionId """ + + """FROM InfoFiles, PublishFreeze, InbredSet """ + + """WHERE InbredSet.Name = '{}' AND """ + + """PublishFreeze.InbredSetId = InbredSet.Id AND """ + + """InfoFiles.InfoPageName = PublishFreeze.Name"""), + 'other_pheno': ("""SELECT PublishFreeze.Name """ + + """FROM PublishFreeze, InbredSet """ + + """WHERE InbredSet.Name = '{}' AND """ + + """PublishFreeze.InbredSetId = InbredSet.Id"""), + 'geno': ("""SELECT GenoFreeze.Id FROM GenoFreeze WHERE """ + + """GenoFreeze.Name = "{}" """) + } + + dataset_name_mapping = { + "mrna_expr": "ProbeSet", + "pheno": "Publish", + "other_pheno": "Publish", + "geno": "Geno", + } + + if t in ['pheno', 'other_pheno']: + name = name.replace("Publish", "") + if bool(len(g.db.execute(sql_query_mapping[t].format(name)))): + self.datasets[name] = dataset_name_mapping[t] + self.redis_instance.set("dataset_structure", json.dumps(self.datasets)) + return True + + return None def __call__(self, name): + if name not in self.datasets: - mrna_expr_query = """ - SELECT - ProbeSetFreeze.Id - FROM - ProbeSetFreeze - WHERE - ProbeSetFreeze.Name = "{0}" - """.format(name) - - results = g.db.execute(mrna_expr_query).fetchall() - if len(results): - self.datasets[name] = "ProbeSet" - redis_instance.set("dataset_structure", json.dumps(self.datasets)) - return self.datasets[name] - - group_name = name.replace("Publish", "") - - pheno_query = """SELECT InfoFiles.GN_AccesionId - FROM InfoFiles, PublishFreeze, InbredSet - WHERE InbredSet.Name = '{0}' AND - PublishFreeze.InbredSetId = InbredSet.Id AND - InfoFiles.InfoPageName = PublishFreeze.Name""".format(group_name) - - results = g.db.execute(pheno_query).fetchall() - if len(results): - self.datasets[name] = "Publish" - redis_instance.set("dataset_structure", json.dumps(self.datasets)) - return self.datasets[name] - - # ZS: For when there isn't an InfoFiles ID; not sure if this and the preceding query are both necessary - other_pheno_query = """SELECT PublishFreeze.Name - FROM PublishFreeze, InbredSet - WHERE InbredSet.Name = '{}' AND - PublishFreeze.InbredSetId = InbredSet.Id""".format(group_name) - - results = g.db.execute(other_pheno_query).fetchall() - if len(results): - self.datasets[name] = "Publish" - redis_instance.set("dataset_structure", json.dumps(self.datasets)) - return self.datasets[name] - - geno_query = """ - SELECT - GenoFreeze.Id - FROM - GenoFreeze - WHERE - GenoFreeze.Name = "{0}" - """.format(name) - - results = g.db.execute(geno_query).fetchall() - if len(results): - self.datasets[name] = "Geno" - self.redis_instance.set("dataset_structure", json.dumps(self.datasets)) - return self.datasets[name] - - # ZS: It shouldn't ever reach this - return None - else: - return self.datasets[name] + for t in ["mrna_expr", "pheno", "other_pheno", "geno"]: + # This has side-effects, with the end result being a truth-y value + if(self.set_dataset_key(t, name)): + break + return self.datasets.get(name, None) # Return None if name has not been set # Do the intensive work at startup one time only diff --git a/wqflask/base/webqtlCaseData.py b/wqflask/base/webqtlCaseData.py index d8487f01..2844cedd 100644 --- a/wqflask/base/webqtlCaseData.py +++ b/wqflask/base/webqtlCaseData.py @@ -19,8 +19,7 @@ # This module is used by GeneNetwork project (www.genenetwork.org) # # Created by GeneNetwork Core Team 2010/08/10 -# -# Last updated by GeneNetwork Core Team 2010/10/20 + from utility.logger import getLogger logger = getLogger(__name__) @@ -29,7 +28,7 @@ import utility.tools utility.tools.show_settings() -class webqtlCaseData(object): +class webqtlCaseData: """one case data in one trait""" def __init__(self, name, value=None, variance=None, num_cases=None, name2=None): @@ -43,44 +42,40 @@ class webqtlCaseData(object): self.outlier = None # Not set to True/False until later def __repr__(self): - str = "<webqtlCaseData> " - if self.value != None: - str += "value=%2.3f" % self.value - if self.variance != None: - str += " variance=%2.3f" % self.variance + case_data_string = "<webqtlCaseData> " + if self.value is not None: + case_data_string += "value=%2.3f" % self.value + if self.variance is not None: + case_data_string += " variance=%2.3f" % self.variance if self.num_cases: - str += " ndata=%s" % self.num_cases + case_data_string += " ndata=%s" % self.num_cases if self.name: - str += " name=%s" % self.name + case_data_string += " name=%s" % self.name if self.name2: - str += " name2=%s" % self.name2 - return str + case_data_string += " name2=%s" % self.name2 + return case_data_string @property def class_outlier(self): """Template helper""" if self.outlier: return "outlier" - else: - return "" + return "" @property def display_value(self): - if self.value != None: + if self.value is not None: return "%2.3f" % self.value - else: - return "x" + return "x" @property def display_variance(self): - if self.variance != None: + if self.variance is not None: return "%2.3f" % self.variance - else: - return "x" + return "x" @property def display_num_cases(self): - if self.num_cases != None: + if self.num_cases is not None: return "%s" % self.num_cases - else: - return "x" + return "x" diff --git a/wqflask/tests/base/data.py b/wqflask/tests/base/data.py new file mode 100644 index 00000000..06a5a989 --- /dev/null +++ b/wqflask/tests/base/data.py @@ -0,0 +1,110 @@ +gen_menu_json = """ +{ + "datasets": { + "human": { + "HLC": { + "Liver mRNA": [ + [ + "320", + "HLC_0311", + "GSE9588 Human Liver Normal (Mar11) Both Sexes" + ] + ], + "Phenotypes": [ + [ + "635", + "HLCPublish", + "HLC Published Phenotypes" + ] + ] + } + }, + "mouse": { + "BXD": { + "Genotypes": [ + [ + "600", + "BXDGeno", + "BXD Genotypes" + ] + ], + "Hippocampus mRNA": [ + [ + "112", + "HC_M2_0606_P", + "Hippocampus Consortium M430v2 (Jun06) PDNN" + ] + ], + "Phenotypes": [ + [ + "602", + "BXDPublish", + "BXD Published Phenotypes" + ] + ] + } + } + }, + "groups": { + "human": [ + [ + "HLC", + "Liver: Normal Gene Expression with Genotypes (Merck)", + "Family:None" + ] + ], + "mouse": [ + [ + "BXD", + "BXD", + "Family:None" + ] + ] + }, + "species": [ + [ + "human", + "Human" + ], + [ + "mouse", + "Mouse" + ] + ], + "types": { + "human": { + "HLC": [ + [ + "Phenotypes", + "Traits and Cofactors", + "Phenotypes" + ], + [ + "Liver mRNA", + "Liver mRNA", + "Molecular Trait Datasets" + ] + ] + }, + "mouse": { + "BXD": [ + [ + "Phenotypes", + "Traits and Cofactors", + "Phenotypes" + ], + [ + "Genotypes", + "DNA Markers and SNPs", + "Genotypes" + ], + [ + "Hippocampus mRNA", + "Hippocampus mRNA", + "Molecular Trait Datasets" + ] + ] + } + } +} +""" diff --git a/wqflask/tests/base/test_data_set.py b/wqflask/tests/base/test_data_set.py index 44a54c7e..94780a5d 100644 --- a/wqflask/tests/base/test_data_set.py +++ b/wqflask/tests/base/test_data_set.py @@ -1,25 +1,18 @@ +"""Tests for wqflask/base/data_set.py""" + import unittest import mock from wqflask import app - +from data import gen_menu_json from base.data_set import DatasetType - -class TestDataSetTypes(unittest.TestCase): - def setUp(self): - self.app_context = app.app_context() - self.app_context.push() - def tearDown(self): - self.app_context.pop() +class TestDataSetTypes(unittest.TestCase): + """Tests for the DataSetType class""" - @mock.patch('base.data_set.g') - def test_data_set_type(self, db_mock): - with app.app_context(): - db_mock.get = mock.Mock() - r = mock.Mock() - r.get.return_value = """ + def setUp(self): + self.test_dataset = """ { "AD-cases-controls-MyersGeno": "Geno", "AD-cases-controls-MyersPublish": "Publish", @@ -32,4 +25,116 @@ class TestDataSetTypes(unittest.TestCase): "B139_K_1206_R": "ProbeSet" } """ - self.assertEqual(DatasetType(r)("All Phenotypes"), "Publish") + self.app_context = app.app_context() + self.app_context.push() + + def tearDown(self): + self.app_context.pop() + + @mock.patch('base.data_set.g') + def test_data_set_type(self, db_mock): + """Test that DatasetType returns correctly if the Redis Instance is not empty + and the name variable exists in the dictionary + + """ + with app.app_context(): + db_mock.get = mock.Mock() + redis_mock = mock.Mock() + redis_mock.get.return_value = self.test_dataset + self.assertEqual(DatasetType(redis_mock) + ("All Phenotypes"), "Publish") + redis_mock.get.assert_called_once_with("dataset_structure") + + @mock.patch('base.data_set.requests.get') + def test_data_set_type_with_empty_redis(self, request_mock): + """Test that DatasetType returns correctly if the Redis Instance is empty and + the name variable exists in the dictionary + + """ + with app.app_context(): + request_mock.return_value.content = gen_menu_json + redis_mock = mock.Mock() + redis_mock.get.return_value = None + data_set = DatasetType(redis_mock) + self.assertEqual(data_set("BXDGeno"), "Geno") + self.assertEqual(data_set("BXDPublish"), "Publish") + self.assertEqual(data_set("HLC_0311"), "ProbeSet") + redis_mock.set.assert_called_once_with( + "dataset_structure", + '{"BXDGeno": "Geno", "BXDPublish": "Publish", "HLCPublish": "Publish", "HLC_0311": "ProbeSet", "HC_M2_0606_P": "ProbeSet"}') + + @mock.patch('base.data_set.g') + def test_set_dataset_key_mrna(self, db_mock): + with app.app_context(): + db_mock.db.execute.return_value = [1, 2, 3] + redis_mock = mock.Mock() + redis_mock.get.return_value = self.test_dataset + data_set = DatasetType(redis_mock) + data_set.set_dataset_key("mrna_expr", "Test") + self.assertEqual(data_set("Test"), "ProbeSet") + redis_mock.set.assert_called_once_with( + "dataset_structure", + '{"Aging-Brain-UCIPublish": "Publish", "AKXDGeno": "Geno", "B139_K_1206_M": "ProbeSet", "AD-cases-controls-MyersGeno": "Geno", "AD-cases-controls-MyersPublish": "Publish", "All Phenotypes": "Publish", "Test": "ProbeSet", "AXBXAPublish": "Publish", "B139_K_1206_R": "ProbeSet", "AXBXAGeno": "Geno"}') + expected_db_call = """""" + db_mock.db.execute.assert_called_with( + ("SELECT ProbeSetFreeze.Id FROM ProbeSetFreeze " + + "WHERE ProbeSetFreeze.Name = \"Test\" ") + ) + + @mock.patch('base.data_set.g') + def test_set_dataset_key_pheno(self, db_mock): + with app.app_context(): + db_mock.db.execute.return_value = [1, 2, 3] + redis_mock = mock.Mock() + redis_mock.get.return_value = self.test_dataset + data_set = DatasetType(redis_mock) + data_set.set_dataset_key("pheno", "Test") + self.assertEqual(data_set("Test"), "Publish") + redis_mock.set.assert_called_once_with( + "dataset_structure", + '{"Aging-Brain-UCIPublish": "Publish", "AKXDGeno": "Geno", "B139_K_1206_M": "ProbeSet", "AD-cases-controls-MyersGeno": "Geno", "AD-cases-controls-MyersPublish": "Publish", "All Phenotypes": "Publish", "Test": "Publish", "AXBXAPublish": "Publish", "B139_K_1206_R": "ProbeSet", "AXBXAGeno": "Geno"}') + expected_db_call = """""" + db_mock.db.execute.assert_called_with( + ("SELECT InfoFiles.GN_AccesionId " + + "FROM InfoFiles, PublishFreeze, InbredSet " + + "WHERE InbredSet.Name = 'Test' AND " + "PublishFreeze.InbredSetId = InbredSet.Id AND " + + "InfoFiles.InfoPageName = PublishFreeze.Name") + ) + + @mock.patch('base.data_set.g') + def test_set_dataset_other_pheno(self, db_mock): + with app.app_context(): + db_mock.db.execute.return_value = [1, 2, 3] + redis_mock = mock.Mock() + redis_mock.get.return_value = self.test_dataset + data_set = DatasetType(redis_mock) + data_set.set_dataset_key("other_pheno", "Test") + self.assertEqual(data_set("Test"), "Publish") + redis_mock.set.assert_called_once_with( + "dataset_structure", + '{"Aging-Brain-UCIPublish": "Publish", "AKXDGeno": "Geno", "B139_K_1206_M": "ProbeSet", "AD-cases-controls-MyersGeno": "Geno", "AD-cases-controls-MyersPublish": "Publish", "All Phenotypes": "Publish", "Test": "Publish", "AXBXAPublish": "Publish", "B139_K_1206_R": "ProbeSet", "AXBXAGeno": "Geno"}') + expected_db_call = """""" + db_mock.db.execute.assert_called_with( + ("SELECT PublishFreeze.Name " + + "FROM PublishFreeze, InbredSet " + + "WHERE InbredSet.Name = 'Test' AND " + "PublishFreeze.InbredSetId = InbredSet.Id") + ) + + @mock.patch('base.data_set.g') + def test_set_dataset_geno(self, db_mock): + with app.app_context(): + db_mock.db.execute.return_value = [1, 2, 3] + redis_mock = mock.Mock() + redis_mock.get.return_value = self.test_dataset + data_set = DatasetType(redis_mock) + data_set.set_dataset_key("geno", "Test") + self.assertEqual(data_set("Test"), "Geno") + redis_mock.set.assert_called_once_with( + "dataset_structure", + '{"Aging-Brain-UCIPublish": "Publish", "AKXDGeno": "Geno", "B139_K_1206_M": "ProbeSet", "AD-cases-controls-MyersGeno": "Geno", "AD-cases-controls-MyersPublish": "Publish", "All Phenotypes": "Publish", "Test": "Geno", "AXBXAPublish": "Publish", "B139_K_1206_R": "ProbeSet", "AXBXAGeno": "Geno"}') + expected_db_call = """""" + db_mock.db.execute.assert_called_with( + ("SELECT GenoFreeze.Id FROM GenoFreeze WHERE GenoFreeze.Name = \"Test\" ") + ) diff --git a/wqflask/tests/base/test_webqtl_case_data.py b/wqflask/tests/base/test_webqtl_case_data.py new file mode 100644 index 00000000..8e8ba482 --- /dev/null +++ b/wqflask/tests/base/test_webqtl_case_data.py @@ -0,0 +1,39 @@ +"""Tests for wqflask/base/webqtlCaseData.py""" +import unittest + +from wqflask import app # Required because of utility.tools in webqtlCaseData.py +from base.webqtlCaseData import webqtlCaseData + +class TestWebqtlCaseData(unittest.TestCase): + """Tests for WebqtlCaseData class""" + + def setUp(self): + self.w = webqtlCaseData(name="Test", + value=0, + variance=0.0, + num_cases=10, + name2="Test2") + + def test_webqtl_case_data_repr(self): + self.assertEqual( + repr(self.w), + "<webqtlCaseData> value=0.000 variance=0.000 ndata=10 name=Test name2=Test2" + ) + + def test_class_outlier(self): + self.assertEqual(self.w.class_outlier, "") + + def test_display_value(self): + self.assertEqual(self.w.display_value, "0.000") + self.w.value = None + self.assertEqual(self.w.display_value, "x") + + def test_display_variance(self): + self.assertEqual(self.w.display_variance, "0.000") + self.w.variance = None + self.assertEqual(self.w.display_variance, "x") + + def test_display_num_cases(self): + self.assertEqual(self.w.display_num_cases, "10") + self.w.num_cases = None + self.assertEqual(self.w.display_num_cases, "x") |