def test_mapping_split_file(self):
bhm.main("transaction_test_split", "./test/test_simple_batch.json", "./test/configuration_to_build_matrix.json")
bhm.main(
"transaction_test_combined", "./test/test_all_file_batch.json", "./test/configuration_to_build_matrix.json"
)
f5s1 = h5py.File("./test/transaction_test_split_1.hdf5", "r")
f5s2 = h5py.File("./test/transaction_test_split_2.hdf5", "r")
f5s3 = h5py.File("./test/transaction_test_split_3.hdf5", "r")
f5c = h5py.File("./test/transaction_test_combined_1.hdf5", "r")
lab_category_count_1 = f5s1["/independent/classes/lab/category/core_array"][...]
lab_category_count_2 = f5s2["/independent/classes/lab/category/core_array"][...]
lab_category_count_3 = f5s3["/independent/classes/lab/category/core_array"][...]
lab_category_count_c = f5c["/independent/classes/lab/category/core_array"][...]
# Concatenate to create a single file
lab_category_count_cs = np.concatenate((lab_category_count_1, lab_category_count_2, lab_category_count_3))
self.assertEquals(lab_category_count_c.shape, lab_category_count_cs.shape)
self.assertEquals(np.sum(lab_category_count_c), np.sum(lab_category_count_cs))
# Read the combined file
f5ca = h5py.File("./test/transaction_test_split_combined.hdf5", "r")
lab_category_count_ca = f5ca["/independent/classes/lab/category/core_array"][...]
self.assertEqual(lab_category_count_c.shape, lab_category_count_ca.shape)
self.assertEquals(np.sum(lab_category_count_c), np.sum(lab_category_count_ca))
def test_mapping_single_file(self):
if os.path.exists("./test/transaction_test_1.hdf5"):
os.remove("./test/transaction_test_1.hdf5")
bhm.main("transaction_test", "./test/test_single_file_batch.json", "./test/configuration_to_build_matrix.json")
f5 = h5py.File("./test/transaction_test_1.hdf5", "r")
dca = f5["/independent/classes/discharge/core_array"][...]
self.assertEquals(dca.shape, (2, 5))
dcac = f5["/independent/classes/discharge/column_annotations"][...]
self.assertEquals(dcac.shape, (3, 5))
lab_count = f5["/independent/classes/lab/count/core_array"][...]
lab_count_c = f5["/independent/classes/lab/count/column_annotations"][...]
self.assertEqual(lab_count.tolist(), [[4.0, 0.0], [4.0, 1.0]])
self.assertEqual(lab_count_c.tolist(), [["BUN", "Troponin"], ["value", "value"], ["count", "count"]])
lab_category_count = f5["/independent/classes/lab/category/core_array"][...]
lab_category_count_c = f5["/independent/classes/lab/category/column_annotations"][...]
self.assertEqual(lab_category_count.tolist(), [[4.0, 0.0, 0.0, 0.0], [1.0, 1.0, 2.0, 1.0]])
self.assertEqual(
lab_category_count_c.tolist(),
[["BUN", "BUN", "BUN", "Troponin"], ["high", "low", "normal", "extreme"], ["", "", "", ""]],
)
self.assertEquals(np.sum(lab_category_count), np.sum(lab_count), "Sums should be equal")
diagnosis1 = f5["/independent/classes/diagnosis/core_array"][...]
diagnosis_c = f5["/independent/classes/diagnosis/column_annotations"]
self.assertTrue(np.sum(diagnosis1) > 0)
def test_filter_and_first_set(self):
directory = "./test/"
bhm.main("filter_test", "./test/test_single_file_batch.json", "./test/configuration_to_build_matrix.json")
f5 = h5py.File("./test/filter_test_1.hdf5", "r")
lab_first = f5["/independent/classes/lab/first/core_array"][...]
lab_first_day = f5["/independent/classes/lab/first_day/core_array"][...]
self.assertNotEqual(lab_first.tolist(), lab_first_day.tolist())
def test_mapping_single_compressed_file(self):
if os.path.exists("./test/transaction_test_1.hdf5"):
os.remove("./test/transaction_test_1.hdf5")
bhm.main("transaction_test", "./test/test_all_file_batch_gz.json", "./test/configuration_to_build_matrix.json")