def test_mongo(self):
self.assertIsInstance(Sample.mongo(), pymongo.collection.Collection)
cursor=Sample.mongo().find()
warn("cursor: got %d records" % cursor.count())
self.assertTrue(cursor.count() > 1)
record=cursor.next()
self.assertTrue('geo_id' in record)
self.assertTrue('_id' in record)
def test_exp_data(self):
sample = Sample('GSM15718').populate() # GSM15718 comes from GSE994
(id_type, data) = sample.expression_data('probe')
self.assertEqual(id_type, 'probe')
self.assertEqual(len(data), 22215)
self.assertEqual(data['200003_s_at'], 2483.300)
self.assertEqual(data['1438_at'], 78.500)
self.assertEqual(data['200707_at'], 159.800)
self.assertEqual(data['200773_x_at'], 2505.100)
self.assertEqual(data['200981_x_at'], 2842.700)
self.assertEqual(data['91682_at'], 65.000)
def test_exp_data(self):
sample=Sample('GSM15718').populate() # GSM15718 comes from GSE994
(id_type,data)=sample.expression_data('probe')
self.assertEqual(id_type, 'probe')
self.assertEqual(len(data), 22215)
self.assertEqual(data['200003_s_at'], 2483.300)
self.assertEqual(data['1438_at'], 78.500)
self.assertEqual(data['200707_at'], 159.800)
self.assertEqual(data['200773_x_at'], 2505.100)
self.assertEqual(data['200981_x_at'], 2842.700)
self.assertEqual(data['91682_at'], 65.000)
def test_init_geo_id(self):
geo_id='GSM30429'
s1=Sample(geo_id);
self.assertIs(s1.geo_id, geo_id)
self.assertFalse(hasattr(s1, 'dataset_ids'))
self.assertFalse(hasattr(s1, 'subset_ids'))
self.assertFalse(hasattr(s1, 'series_ids'))
s1.populate()
self.assertTrue(hasattr(s1, 'dataset_ids'))
self.assertTrue(hasattr(s1, 'subset_ids'))
self.assertTrue(hasattr(s1, 'series_ids'))
def test_init_geo_id(self):
geo_id = 'GSM30429'
s1 = Sample(geo_id)
self.assertIs(s1.geo_id, geo_id)
self.assertFalse(hasattr(s1, 'dataset_ids'))
self.assertFalse(hasattr(s1, 'subset_ids'))
self.assertFalse(hasattr(s1, 'series_ids'))
s1.populate()
self.assertTrue(hasattr(s1, 'dataset_ids'))
self.assertTrue(hasattr(s1, 'subset_ids'))
self.assertTrue(hasattr(s1, 'series_ids'))
def test_all_ids_with_pheno(self):
phenos=[
"normal",
"adenocarcinoma",
"squamous cell carcinoma",
"asthma",
"chronic obstructive pulmonary disease",
"large cell lung carcinoma"
]
for pheno in phenos:
ids=Sample.all_ids_with_pheno(pheno)
self.assertIsInstance(ids, list)
warn("got %d '%s' samples" % (len(ids), pheno))
self.assertTrue(len(ids) > 0)
ids_with_data=[x for x in ids if os.access(Sample.data_path_of(geo_id=x), os.R_OK)]
warn("got %d '%s' samples with data" % (len(ids_with_data), pheno))
def test_all_ids_with_pheno(self):
phenos = [
"normal", "adenocarcinoma", "squamous cell carcinoma", "asthma",
"chronic obstructive pulmonary disease",
"large cell lung carcinoma"
]
for pheno in phenos:
ids = Sample.all_ids_with_pheno(pheno)
self.assertIsInstance(ids, list)
warn("got %d '%s' samples" % (len(ids), pheno))
self.assertTrue(len(ids) > 0)
ids_with_data = [
x for x in ids
if os.access(Sample.data_path_of(geo_id=x), os.R_OK)
]
warn("got %d '%s' samples with data" % (len(ids_with_data), pheno))
def test_all_ids_with_data(self):
ids=Sample.all_ids_with_data(id_type='probe')
warn("len(ids)=%d" % (len(ids)))
self.assertTrue(len(ids) > 1000)
def test_all_with_data_gene(self):
samples = Sample.all_with_data(id_type='gene')
warn("got %d 'gene' samples" % (len(samples)))
self.assertTrue(len(samples) > 100)
def test_all_with_data_probe(self):
samples = Sample.all_with_data(id_type='probe', ids_only=True)
warn("got %d 'probe' samples" % (len(samples)))
self.assertTrue(len(samples) > 100)
def test_all_with_data_gene(self):
samples=Sample.all_with_data(id_type='gene')
warn("got %d 'gene' samples" % (len(samples)))
self.assertTrue(len(samples)>100)
def test_all_with_data_probe(self):
samples=Sample.all_with_data(id_type='probe', ids_only=True)
warn("got %d 'probe' samples" % (len(samples)))
self.assertTrue(len(samples)>100)
def test_init_dict(self):
geo_id = 'GSM30429'
s1 = Sample({'geo_id': geo_id, 'name': 'fred'})
self.assertEqual(s1.geo_id, geo_id)
self.assertEqual(s1.name, 'fred')