def test_empty_mtx_file(self):
"""Ingest Pipeline should fail gracefully when an empty file is given
"""
args = [
"--study-id",
"5d276a50421aa9117c982845",
"--study-file-id",
"5dd5ae25421aa910a723a337",
"ingest_expression",
"--taxon-name",
"H**o sapiens",
"--taxon-common-name",
"human",
"--ncbi-taxid",
"9606",
"--genome-assembly-accession",
"GCA_000001405.15",
"--genome-annotation",
"Ensembl 94",
"--matrix-file",
"../tests/data/mtx/AB_toy_data_toy.matrix.mtx",
"--matrix-file-type",
"mtx",
"--gene-file",
"../tests/data/empty_file.txt",
"--barcode-file",
"../tests/data/mtx/AB_toy_data_toy.barcodes.tsv",
]
ingest, arguments, status, status_cell_metadata = self.execute_ingest(
args)
with self.assertRaises(SystemExit) as cm:
exit_pipeline(ingest, status, status_cell_metadata, arguments)
self.assertEqual(cm.exception.code, 1)
def test_bad_cluster_missing_coordinate_file(self):
"""Ingest Pipeline should fail for missing coordinate in cluster file
"""
args = [
"--study-id",
"5d276a50421aa9117c982845",
"--study-file-id",
"5dd5ae25421aa910a723a337",
"ingest_cluster",
"--cluster-file",
"../tests/data/cluster_bad_missing_coordinate.txt",
"--ingest-cluster",
"--name",
"cluster1",
]
ingest, arguments, status, status_cell_metadata = self.execute_ingest(
args)
with self.assertRaises(SystemExit) as cm:
exit_pipeline(ingest, status, status_cell_metadata, arguments)
self.assertEqual(cm.exception.code, 1)
def test_bad_metadata_file(self):
"""Ingest Pipeline should not succeed for misformatted metadata file
"""
args = [
"--study-id",
"5d276a50421aa9117c982845",
"--study-file-id",
"5dd5ae25421aa910a723a337",
"ingest_cell_metadata",
"--cell-metadata-file",
"../tests/data/metadata_bad.txt",
"--study-accession",
"SCP123",
"--ingest-cell-metadata",
]
ingest, arguments, status, status_cell_metadata = self.execute_ingest(
args)
with self.assertRaises(SystemExit) as cm:
exit_pipeline(ingest, status, status_cell_metadata, arguments)
self.assertEqual(cm.exception.code, 1)
def test_subsample_no_cell_intersection(self, mock_load_subsample):
"""When cell values in cluster are not present in cell metadata file ingest should fail.
"""
args = [
"--study-id",
"5d276a50421aa9117c982845",
"--study-file-id",
"5dd5ae25421aa910a723a337",
"ingest_subsample",
"--cluster-file",
"../tests/data/good_subsample_cluster.csv",
"--name",
"custer1",
"--cell-metadata-file",
"../tests/data/test_cell_metadata.csv",
"--subsample",
]
ingest, arguments, status, status_cell_metadata = self.execute_ingest(
args)
with self.assertRaises(SystemExit) as cm, self.assertRaises(
ValueError):
exit_pipeline(ingest, status, status_cell_metadata, arguments)
self.assertEqual(cm.exception.code, 1)
def test_exponential_back_off_expression_file(self,
mock_check_unique_cells,
mock_load):
"""Ingest Pipeline should not succeed if mongo cannot connect after 5
reconnection tries.
"""
args = [
"--study-id",
"5d276a50421aa9117c982845",
"--study-file-id",
"5dd5ae25421aa910a723a337",
"ingest_expression",
"--taxon-name",
"H**o sapiens",
"--taxon-common-name",
"human",
"--ncbi-taxid",
"9606",
"--genome-assembly-accession",
"GCA_000001405.15",
"--genome-annotation",
"Ensembl 94",
"--matrix-file",
"gs://fake-bucket/tests/data/AB_toy_data_toy.matrix.mtx",
"--matrix-file-type",
"mtx",
"--gene-file",
"gs://fake-bucket/tests/data/AB_toy_data_toy.genes.tsv",
"--barcode-file",
"gs://fake-bucket/tests/data/AB_toy_data_toy.barcodes.tsv",
]
ingest, arguments, status, status_cell_metadata = self.execute_ingest(
args)
with self.assertRaises(SystemExit) as cm:
exit_pipeline(ingest, status, status_cell_metadata, arguments)
self.assertEqual(cm.exception.code, 1)