def test_transform_fn_batch(self, mock_load):
"""
Assures transform function batches data array creation correctly and
"""
GeneExpression.DATA_ARRAY_BATCH_SIZE = 7
expression_matrix = MTXIngestor(
"../tests/data/mtx/AB_toy_data_toy.matrix.mtx",
"5dd5ae25421aa910a723a447",
"5f2f58eba0845f8c4cf1dc12",
gene_file="../tests/data/mtx/AB_toy_data_toy.genes.tsv",
barcode_file="../tests/data/mtx/AB_toy_data_toy.barcodes.tsv",
)
expression_matrix.is_raw_count = False
with patch(
"expression_files.expression_files.GeneExpression.DATA_ARRAY_BATCH_SIZE",
new_callable=PropertyMock,
return_value=7,
):
expression_matrix.extract_feature_barcode_matrices()
expression_matrix.transform()
def test_transform_fn(self):
"""
Assures transform function creates gene data model correctly
"""
expression_matrix = MTXIngestor(
"../tests/data/mtx/AB_toy_data_toy.matrix.mtx",
"5d276a50421aa9117c982845",
"5dd5ae25421aa910a723a337",
gene_file="../tests/data/mtx/AB_toy_data_toy.genes.tsv",
barcode_file="../tests/data/mtx/AB_toy_data_toy.barcodes.tsv",
)
expression_matrix.test_models = None
expression_matrix.models_processed = 0
expression_matrix.extract_feature_barcode_matrices()
expression_matrix.is_raw_count = False
expression_matrix.transform()
amount_of_models = len(
expression_matrix.test_models["data_arrays"].keys()) + len(
expression_matrix.test_models["gene_models"].keys())
self.assertEqual(expression_matrix.models_processed, amount_of_models)
# MTX with a single column feature file
expression_matrix = MTXIngestor(
"../tests/data/mtx/one_column_feature_file_data_models.mtx",
"5f2f58eba0845f8c4cf1dc12",
"5dd5ae25421aa910a723a447",
gene_file=
"../tests/data/mtx/one_column_feature_file_data_models.genes.tsv",
barcode_file=
"../tests/data/mtx/one_column_feature_file_data_models.barcodes.tsv",
)
expression_matrix.test_models = None
expression_matrix.models_processed = 0
expression_matrix.extract_feature_barcode_matrices()
expression_matrix.is_raw_count = False
expression_matrix.transform()
amount_of_models = len(
expression_matrix.test_models["data_arrays"]) + len(
expression_matrix.test_models["gene_models"])
self.assertEqual(expression_matrix.models_processed, amount_of_models)
# Checks models for unsorted MTX file
with patch(
"expression_files.expression_files.GeneExpression.check_unique_cells",
return_value=True,
), patch(
"expression_files.expression_files.GeneExpression.is_raw_count_file",
return_value=False,
):
expression_matrix = MTXIngestor(
"../tests/data/mtx/AB_toy_data_toy.unsorted_mtx.mtx",
"5d276a50421aa9117c982845",
"5dd5ae25421aa910a723a337",
gene_file="../tests/data/mtx/AB_toy_data_toy.genes.tsv",
barcode_file="../tests/data/mtx/AB_toy_data_toy.barcodes.tsv",
)
expression_matrix.test_models = None
expression_matrix.models_processed = 0
expression_matrix.execute_ingest()
amount_of_models = len(
expression_matrix.test_models["data_arrays"]) + len(
expression_matrix.test_models["gene_models"])
self.assertEqual(expression_matrix.models_processed,
amount_of_models)
# Checks models for raw count matrices
expression_matrix = MTXIngestor(
"../tests/data/mtx/raw_count.sparse.mtx",
"5d276a50421aa9117c982845",
"5dd5ae25421aa910a723a337",
gene_file="../tests/data/mtx/raw_count.genes.tsv",
barcode_file="../tests/data/mtx/raw_count.barcodes.tsv",
)
expression_matrix.is_raw_count = True
expression_matrix.test_models = None
expression_matrix.models_processed = 0
expression_matrix.extract_feature_barcode_matrices()
expression_matrix.transform()
amount_of_models = len(expression_matrix.test_models["data_arrays"])
self.assertEqual(expression_matrix.models_processed, amount_of_models)