def test_write_withallrefpos_matrix(self):
expected_files = ['TestContig_withallrefpos.tsv']
expected_lines = (
'\t'.join(write_matrix.get_header('withallrefpos', identifiers)) + '\n',
'TestContig::1 A C G T R Y K M S W B D H V N . a c g t r y k m s w b d h v n 2 0 1 -19/4 3/4 4/4 5 6 7 8 0 9 TestContig 1 True True YYYYNNNNNNNNNNNNYYYYNNNNNNNNNNN YYYYYYYYYYYYNNNNNNNNNNNN NNNNNNNNNNNNYYYYYYYYYYYY 1NNNNNNNNNNNNNNNNNNNNNNN\n',
'TestContig::2 A C G T R Y K M S W B D H V N . a c g t r y k m s w b d h v n 2 0 1 -19/4 3/4 4/4 5 6 7 8 0 9 TestContig 2 True True YYYYNNNNNNNNNNNNYYYYNNNNNNNNNNN YYYYYYYYYYYYNNNNNNNNNNNN NNNNNNNNNNNNYYYYYYYYYYYY 1NNNNNNNNNNNNNNNNNNNNNN1\n',
'TestContig::3 A C G T R Y K M S W B D H V N . a c g t r y k m s w b d h v n 2 0 1 -19/4 3/4 4/4 5 6 7 8 0 9 TestContig 3 True True YYYYNNNNNNNNNNNNYYYYNNNNNNNNNNN YYYYYYYYYYYYNNNNNNNNNNNN NNNNNNNNNNNNYYYYYYYYYYYY 1NNNNNNNNNNNNNNNNNNNNNN2\n'
)
with TemporaryDirectory() as tmpdir:
withallrefpos_dir = os.path.join(tmpdir, 'withallrefpos.tsv')
writer = write_matrix.write_withallrefpos_matrix(tmpdir, contig_name, identifiers)
writer.send(None)
for position in self.positions:
writer.send(position)
writer.close()
# The file was created.
self.assertListEqual(expected_files, os.listdir(withallrefpos_dir))
with open(os.path.join(withallrefpos_dir, expected_files[0])) as handle:
# The file contains all the expected rows.
for expected_line, line in zip(expected_lines, handle):
self.assertEqual(expected_line, line)
# The file does not contain any unexpected rows.
self.assertEqual([], handle.readlines())
# No other artifacts were created in the tmpdir.
self.assertListEqual(expected_files, os.listdir(withallrefpos_dir))
def test_write_bestsnp_vcf(self):
expected_files = ['TestContig_bestsnp.vcf']
expected_lines = (
self.metadata,
'\t'.join(write_matrix.get_header('vcf', identifiers)) + '\n',
'TestContig 3 . A C,G,T,R,Y,K,M,S,W,B,D,H,V,N,.,a,c,g,t,r,y,k,m,s,w,b,d,h,v,n . PASS AN=31;NS=3 GT:FT\n'
)
with TemporaryDirectory() as tmpdir:
writer = write_matrix.write_bestsnp_vcf(tmpdir, contig_name, identifiers, self.metadata)
writer.send(None)
for position in self.positions:
writer.send(position)
writer.close()
# The file was created.
self.assertListEqual(expected_files, os.listdir(tmpdir))
with open(os.path.join(tmpdir, expected_files[0])) as handle:
# The file contains all the expected rows.
for expected_line, line in itertools.zip_longest(expected_lines, handle):
self.assertEqual(expected_line, line)
# The file does not contain any unexpected rows.
self.assertEqual([], handle.readlines())
# No other artifacts were created in the tmpdir.
self.assertListEqual(expected_files, os.listdir(tmpdir))
def test_write_bestsnp_matrix(self):
identifiers = tuple(map(lambda sample: sample[0].name, sample_groups))
expected_files = ['TestContig_bestsnp.tsv']
expected_lines = (
'\t'.join(write_matrix.get_header('best_snp', identifiers)) + '\n',
'TestContig::3 A C G T R Y K M S W B D H V N . a c g t r y k m s w b d h v n 2 0 1 -19/4 3/4 4/4 5 6 7 8 0 9 TestContig 3 True True 1NNNNNNNNNNNNNNNNNNNNNN2\n',
)
with TemporaryDirectory() as tmpdir:
bestsnp_dir = os.path.join(tmpdir, 'bestsnp.tsv')
writer = write_matrix.write_bestsnp_matrix(tmpdir, contig_name, sample_groups)
writer.send(None)
for position in self.positions:
writer.send(position)
writer.close()
# The file was created.
self.assertListEqual(expected_files, os.listdir(bestsnp_dir))
with open(os.path.join(bestsnp_dir, expected_files[0])) as handle:
# The file contains all the expected rows.
for expected_line, line in itertools.zip_longest(expected_lines, handle):
self.assertEqual(expected_line, line)
# The file does not contain any unexpected rows.
# self.assertEqual([], handle.readlines())
# No other artifacts were created in the tmpdir.
self.assertListEqual(expected_files, os.listdir(bestsnp_dir))
def test_write_missingdata_matrix(self):
expected_files = ['TestContig_missingdata.tsv']
expected_lines = (
'\t'.join(write_matrix.get_header('missingdata', identifiers)) + '\n',
'TestContig::2 A C G T R Y K M S W B D H V N . a c g t r y k m s w b d h v n 2 0 1 7/30 3/30 4/30 5 6 7 8 0 9 TestContig 2 True True YYYYNNNNNNNNNNNNYYYYNNNNNNNNNNN YYYYYYYYYYYYNNNNNNNNNNNN NNNNNNNNNNNNYYYYYYYYYYYY 1NNNNNNNNNNNNNNNNNNNNNN1\n',
)
with TemporaryDirectory() as tmpdir:
missingdata_dir = os.path.join(tmpdir, 'missingdata.tsv')
writer = write_matrix.write_missingdata_matrix(tmpdir, contig_name, identifiers)
writer.send(None)
for position in self.positions:
writer.send(position)
writer.close()
# The file was created.
self.assertListEqual(expected_files, os.listdir(missingdata_dir))
with open(os.path.join(missingdata_dir, expected_files[0])) as handle:
# The file contains all the expected rows.
for expected_line, line in itertools.zip_longest(expected_lines, handle):
self.assertEqual(expected_line, line)
# The file does not contain any unexpected rows.
# self.assertEqual([], handle.readlines())
# No other artifacts were created in the tmpdir.
self.assertListEqual(expected_files, os.listdir(missingdata_dir))
def test_withallrefpos(self):
expected = (
'LocusID',
'Reference',
'sample1::aligner,snpcaller',
'sample2::aligner1,snpcaller',
'sample2::aligner2',
'sample3::aligner,snpcaller',
'#SNPcall',
'#Indelcall',
'#Refcall',
'#CallWasMade',
'#PassedDepthFilter',
'#PassedProportionFilter',
'#A',
'#C',
'#G',
'#T',
'#Indel',
'#NXdegen',
'Contig',
'Position',
'InDupRegion',
'SampleConsensus',
'Pattern',
'Pattern#'
)
self.assertEqual(expected, write_matrix.get_header('withallrefpos', identifiers))
def test_vcf(self):
expected = (
'#CHROM',
'POS',
'ID',
'REF',
'ALT',
'QUAL',
'FILTER',
'INFO',
'FORMAT',
'sample1::aligner,snpcaller',
'sample2::aligner1,snpcaller',
'sample2::aligner2',
'sample3::aligner,snpcaller'
)
self.assertEqual(expected, write_matrix.get_header('vcf', identifiers))
def test_write_general_stats(self):
expected_files = ['general_stats.tsv']
expected_lines = (
self.metadata,
'\t'.join(write_matrix.get_header('vcf', identifiers)) + '\n',
'',
)
contig_stats = Counter(
{'reference_clean': 8, 'all_passed_proportion': 8, 'quality_breadth': 8,
'Contig': 'ContigWithFilePositionOffset', 'reference_length': 8, 'best_snps': 0,
'all_passed_coverage': 8, 'reference_duplicated': 0, 'all_passed_consensus': 8, 'all_called': 8,
'any_snps': 0}
)
with TemporaryDirectory() as tmpdir:
writer = write_matrix.write_general_stats(os.path.join(tmpdir, expected_files[0]), contig_stats)
writer.send(None)
for position in self.positions:
writer.send(position)
writer.close()
# The file was created.
self.assertListEqual(expected_files, os.listdir(tmpdir))
with open(os.path.join(tmpdir, expected_files[0])) as handle:
# The file contains all the expected rows.
for expected_line, line in zip(expected_lines, handle):
self.assertEqual(expected_line, line)
# The file does not contain any unexpected rows.
self.assertEqual([], handle.readlines())
# No other artifacts were created in the tmpdir.
self.assertListEqual(expected_files, os.listdir(tmpdir))
def test_undefined_header(self):
# It should warn the developer if they request an undefined header.
with self.assertRaises(ValueError):
write_matrix.get_header('undefined', identifiers)