def test_2_no_seqid(self):
csv_path = self.csv_files_dir.joinpath("test_2_no_seqid.csv")
csv_content = pathtools.get_file_content_stringio(csv_path)
with self.assertRaises(Exception) as err:
_ = msa.MetadataCSV(csv_content, csv_path)
self.assertEqual(f"No \'seqid\' column in metadata csv.",
str(err.exception))
def setUp(self):
metadata_path = Path(__file__).parent.joinpath(
"../seq_metadata.csv").resolve()
self.metadatacsv = msa.MetadataCSV(
pathtools.get_file_content_stringio(metadata_path), metadata_path)
self.po_files_dir = Path(__file__).parent.joinpath(
"po_files").resolve()
def test_3_empty_file(self):
csv_path = self.csv_files_dir.joinpath("test_3_empty_file.csv")
csv_content = pathtools.get_file_content_stringio(csv_path)
with self.assertRaises(Exception) as err:
_ = msa.MetadataCSV(csv_content, csv_path)
self.assertEqual(f"Empty csv file.", str(err.exception))
def setUp(self):
metadata_path = Path(__file__).parent.joinpath(
"../seq_metadata.csv").resolve()
self.metadatacsv = msa.MetadataCSV(
pathtools.get_file_content_stringio(metadata_path), metadata_path)
self.maf_files_dir = Path(__file__).parent.joinpath(
"maf_files_with_gaps").resolve()
self.missing_n = missings.MissingBase()
def test_7_not_unique_seqids(self):
csv_path = self.csv_files_dir.joinpath("test_7_not_unique_seqids.csv")
csv_content = pathtools.get_file_content_stringio(csv_path)
with self.assertRaises(Exception) as err:
_ = msa.MetadataCSV(csv_content, csv_path)
self.assertEqual(
"Repeated values in seqid column in metadata file. Make them unique.",
str(err.exception))
def setUp(self):
metadata_path = Path(__file__).parent.joinpath(
"../seq_metadata.csv").resolve()
self.metadatacsv = msa.MetadataCSV(
pathtools.get_file_content_stringio(metadata_path), metadata_path)
self.maf_files_dir = Path(__file__).parent.joinpath(
"maf_files_with_gaps").resolve()
self.fasta_provider = DAGMaf2PoagraphFakeFastaProviderTests.FakeFastaProvider(
)
def test_6_incorrect_commas_number(self):
csv_path = self.csv_files_dir.joinpath(
"test_6_incorrect_commas_number.csv")
csv_content = pathtools.get_file_content_stringio(csv_path)
with self.assertRaises(Exception) as err:
_ = msa.MetadataCSV(csv_content, csv_path)
self.assertEqual(
"CSV metadata error. Different number of columns in line 0 than in header line.",
str(err.exception))
def test_9_get_seqids(self):
metadata_path = self.csv_files_dir.joinpath("test_1_correct.csv")
csv_content = pathtools.get_file_content_stringio(metadata_path)
expected_seqids = [
msa.SequenceID('s1'),
msa.SequenceID('s2'),
msa.SequenceID('s3')
]
m = msa.MetadataCSV(csv_content, metadata_path)
actual_seqids = m.get_all_sequences_ids()
self.assertEqual(expected_seqids, actual_seqids)
def test_10_metadata_feed_to_alignment_from_csv(self, test_name, maf_name,
csv_name, po_name,
expected_metadata):
maf_path = self.alignment_files_dir.joinpath(maf_name)
csv_path = self.csv_files_dir.joinpath(csv_name)
po_path = self.alignment_files_dir.joinpath(po_name)
poagraph, _ = builder.build_from_dagmaf(
msa.Maf(pathtools.get_file_content_stringio(maf_path), maf_path),
self.fasta_provider,
msa.MetadataCSV(pathtools.get_file_content_stringio(csv_path),
csv_path))
actual_metadata = {
seq_id: seq.seqmetadata
for seq_id, seq in poagraph.sequences.items()
}
self.assertEqual(expected_metadata, actual_metadata)
poagraph = builder.build_from_maf(
msa.Maf(pathtools.get_file_content_stringio(maf_path), maf_path),
msa.MetadataCSV(pathtools.get_file_content_stringio(csv_path),
csv_path))
actual_metadata = {
seq_id: seq.seqmetadata
for seq_id, seq in poagraph.sequences.items()
}
self.assertEqual(expected_metadata, actual_metadata)
poagraph = builder.build_from_po(
msa.Po(pathtools.get_file_content_stringio(po_path), maf_path),
msa.MetadataCSV(pathtools.get_file_content_stringio(csv_path),
csv_path))
actual_metadata = {
seq_id: seq.seqmetadata
for seq_id, seq in poagraph.sequences.items()
}
self.assertEqual(expected_metadata, actual_metadata)
def test_4_seqid_is_last(self):
metadata_path = self.csv_files_dir.joinpath("test_4_seqid_is_last.csv")
csv_content = pathtools.get_file_content_stringio(metadata_path)
expected_metadata = {
msa.SequenceID('s1'): {
'name': 'sequence1',
'group': 'A'
},
msa.SequenceID('s2'): {
'name': 'sequence2',
'group': 'B'
},
msa.SequenceID('s3'): {
'name': 'sequence3',
'group': 'B'
}
}
m = msa.MetadataCSV(csv_content, metadata_path)
actual_metadata = m.metadata
self.assertEqual(expected_metadata, actual_metadata)
def setUp(self):
metadata_path = Path(__file__).parent.joinpath("../seq_metadata.csv").resolve()
self.metadatacsv = msa.MetadataCSV(pathtools.get_file_content_stringio(metadata_path), metadata_path)
self.maf_files_dir = Path(__file__).parent.joinpath("maf_files_with_cycles_or_reversion").resolve()
self.fasta_provider = missings.ConstBaseProvider(missings.MissingBase())