def test_difference_when_many_diffs_h_vs_ss(self):
header_metadata = SAMFileHeaderMetadata('/seq/123.bam',
samples={'name': set(['S9']), 'accession_number': set(['Acc9'])},
libraries={'internal_id': set(['123'])}, studies={})
seqscape_metadata = SeqscapeMetadata(samples={'name': set(['S1'])}, libraries={'internal_id': set(['444'])}, studies={})
result = header_metadata.difference(seqscape_metadata)
self.assertEqual(result,{'samples': {'name': set(['S9'])}, 'libraries': {'internal_id': set(['123'])}})
def test_difference_when_diffs_h_vs_i(self):
irods_metadata = IrodsSeqFileMetadata('/seq/123.bam',
samples={
'name': set(['S1']),
'accession_number':
set(['EGA1']),
'internal_id': set(['1'])
},
libraries={
'name': set(['123']),
'internal_id': set(['123'])
},
studies={
'name':
set(["Crohns disease"]),
'accession_number':
set(['EGAS4']),
'internal_id':
set(['4'])
})
header_metadata = SAMFileHeaderMetadata(
'/seq/123.bam',
samples={
'name': set(['S100']),
'accession_number': set(),
'internal_id': set()
},
libraries={'internal_id': set(['123'])},
studies={})
result = header_metadata.difference(irods_metadata)
self.assertDictEqual(result, {'samples': {'name': set(['S100'])}})
def test_difference_when_there_are_diffs_h_vs_ss2(self):
header_metadata = SAMFileHeaderMetadata('/seq/123.bam',
samples={'name': set(['S9'])},
libraries={},
studies={})
seqscape_metadata = SeqscapeMetadata(samples={'name': set(['S1'])},
libraries={},
studies={})
result = header_metadata.difference(seqscape_metadata)
self.assertEqual(result, {'samples': {'name': set(['S9'])}})
def test_difference_when_diffs_h_vs_i(self):
irods_metadata = IrodsSeqFileMetadata('/seq/123.bam',
samples={'name': set(['S1']), 'accession_number': set(['EGA1']), 'internal_id': set(['1'])},
libraries={'name': set(['123']), 'internal_id': set(['123'])},
studies={'name': set(["Crohns disease"]), 'accession_number': set(['EGAS4']), 'internal_id': set(['4'])})
header_metadata = SAMFileHeaderMetadata('/seq/123.bam',
samples={'name': set(['S100']), 'accession_number' : set(), 'internal_id': set()},
libraries={'internal_id': set(['123'])}, studies={})
result = header_metadata.difference(irods_metadata)
self.assertDictEqual(result, {'samples': {'name': set(['S100'])}})
def test_fix_metadata_2(self):
metadata = SAMFileHeaderMetadata(fpath='some_path')
metadata.samples = {'name': {'sample1', 'sample2', 'sample3'}}
metadata.libraries = {'internal_id': {-1, '', 123}, 'name': set()}
expected = SAMFileHeaderMetadata(fpath='some_path')
expected.samples = {'name': {'sample1', 'sample2', 'sample3'}}
expected.libraries = {'internal_id': {123}, 'name': set()}
metadata.fix_metadata()
self.assertDictEqual(metadata.samples, expected.samples)
self.assertDictEqual(metadata.libraries, expected.libraries)
def test_difference_when_there_are_diff_type_ids_h_vs_ss(self):
header_metadata = SAMFileHeaderMetadata('/seq/123.bam',
samples={
'name':
set(['S1']),
'accession_number':
set(['Acc1'])
},
libraries={},
studies={})
seqscape_metadata = SeqscapeMetadata(samples={'name': set(['S1'])},
libraries={},
studies={})
result = header_metadata.difference(seqscape_metadata)
self.assertEqual(result, {})
def test_fetch(self):
fpath = 'test_data/A.bam'
result = SAMFileHeaderMetadataProvider.fetch_metadata(fpath)
expected = SAMFileHeaderMetadata(
fpath='test_data/A.bam',
fname='A.bam',
samples={
'name': set(['A_J']),
'internal_id': set(),
'accession_number': set()
},
platforms={'SLX'},
libraries={
'name':
set([
'A_J_SLX_500_HC_2', 'A_J_SLX_200_NOPCR_3',
'A_J_SLX_200_NOPCR_2', 'A_J_SLX_200_NOPCR_2',
'A_J_SLX_200_NOPCR_3', 'A_J_SLX_300_DSS_1',
'A_J_SLX_300_DSS_2', 'A_J_SLX_300_DSS_3',
'A_J_SLX_500_DSS_1', 'A_J_SLX_500_DSS_2',
'A_J_SLX_500_DSS_3'
]),
'internal_id':
set(),
'accession_number':
set()
})
self.assertEqual(result, expected)
def test_fix_metadata_1(self):
metadata = SAMFileHeaderMetadata(fpath='some_path')
metadata.samples = {'name': {'sample1', 'sample2', 'sample3', None}}
metadata.libraries = {'internal_id': {-1, '', 123}, 'name': {'', None}}
expected = SAMFileHeaderMetadata(fpath='some_path')
expected.samples = {'name': {'sample1', 'sample2', 'sample3'}}
expected.libraries = {'internal_id': {123}, 'name': set()}
metadata.fix_metadata()
self.assertDictEqual(metadata.samples, expected.samples)
self.assertDictEqual(metadata.libraries, expected.libraries)
def test_check_for_invalid_ids_4(self):
multi_ids = {
'name': {'Ana', ''},
'id': set(),
'accession_number': {'EGA1', None}
}
result = SAMFileHeaderMetadata._check_for_invalid_ids(
multi_ids, 'sample')
self.assertEqual(result.result, RESULT.FAILURE)
def test_difference_when_many_diffs_h_vs_ss(self):
header_metadata = SAMFileHeaderMetadata(
'/seq/123.bam',
samples={
'name': set(['S9']),
'accession_number': set(['Acc9'])
},
libraries={'internal_id': set(['123'])},
studies={})
seqscape_metadata = SeqscapeMetadata(
samples={'name': set(['S1'])},
libraries={'internal_id': set(['444'])},
studies={})
result = header_metadata.difference(seqscape_metadata)
self.assertEqual(
result, {
'samples': {
'name': set(['S9'])
},
'libraries': {
'internal_id': set(['123'])
}
})
def fetch_metadata(cls, fpath, irods=False):
if irods:
header_as_text = IrodsSamFileHeaderExtractor.extract(fpath)
else:
header_as_text = LustreSamFileHeaderExtractor.extract(fpath)
raw_header = SAMFileHeaderParser.parse(header_as_text)
rg_tags_parsed = SAMFileRGTagParser.parse(raw_header.rg_tags)
samples = EntityIdentifier.separate_identifiers_by_type(
rg_tags_parsed.samples)
libraries = EntityIdentifier.separate_identifiers_by_type(
rg_tags_parsed.libraries)
return SAMFileHeaderMetadata(fpath=fpath,
samples=samples,
libraries=libraries,
platforms=rg_tags_parsed.platforms)
def test_mdata_from_diff_srcs_when_different_id_types(self):
irods_metadata = IrodsSeqFileMetadata('/seq/123.bam',
samples={'name': set(['S1']), 'accession_number': set(['EGA1']),
'internal_id': set()},
libraries={}, studies={})
header_metadata = SAMFileHeaderMetadata('/seq/123.bam', samples={'name': set(['S1'])}, libraries={},
studies={})
seqscape_metadata = SeqscapeMetadata(samples={'name': set(['S1'])}, libraries={}, studies={})
issues_dict = defaultdict(list)
FileMetadataComparison.check_metadata_across_different_sources({'/seq/213.bam': irods_metadata},
{'/seq/213.bam': header_metadata},
{'/seq/213.bam': seqscape_metadata},
issues_dict)
check_results = issues_dict['/seq/213.bam']
self.assertEqual(4, len(check_results))
results = {c.result for c in check_results}
self.assertSetEqual(results, {RESULT.SUCCESS})
def test_difference_when_no_diffs_i_vs_h(self):
irods_metadata = IrodsSeqFileMetadata('/seq/123.bam',
samples={
'name': set(['S1']),
'accession_number': set(),
'internal_id': set()
},
libraries={},
studies={})
header_metadata = SAMFileHeaderMetadata('/seq/123.bam',
samples={
'name': set(['S1']),
'accession_number': set(),
'internal_id': set()
},
libraries={},
studies={})
result = irods_metadata.difference(header_metadata)
self.assertDictEqual(result, {})
def test_check_for_invalid_ids_3(self):
multi_ids = {'name': {'Ana'}}
result = SAMFileHeaderMetadata._check_for_invalid_ids(
multi_ids, 'sample')
self.assertEqual(result.result, RESULT.SUCCESS)
def test_check_for_invalid_ids_2(self):
multi_ids = {'name': {'Ana', ''}, 'ids': {-1, 0}}
result = SAMFileHeaderMetadata._check_for_invalid_ids(
multi_ids, 'sample')
self.assertEqual(result.result, RESULT.FAILURE)
def test_filter_out_invalid_ids_4(self):
ids = {None, 123}
expected_result = {123}
actual_result = SAMFileHeaderMetadata._filter_out_invalid_ids(ids)
self.assertSetEqual(expected_result, actual_result)
def test_is_id_valid_6(self):
id = "123"
self.assertTrue(SAMFileHeaderMetadata._is_id_valid(id))
def test_difference_when_there_are_diffs_h_vs_ss2(self):
header_metadata = SAMFileHeaderMetadata('/seq/123.bam', samples={'name': set(['S9'])}, libraries={}, studies={})
seqscape_metadata = SeqscapeMetadata(samples={'name': set(['S1'])}, libraries={}, studies={})
result = header_metadata.difference(seqscape_metadata)
self.assertEqual(result,{'samples': {'name': set(['S9'])}})
def test_is_id_valid_5(self):
id = None
self.assertFalse(SAMFileHeaderMetadata._is_id_valid(id))
def test_check_for_invalid_ids_2(self):
multi_ids = {'name': {'Ana', ''}, 'ids': {-1, 0}}
result = SAMFileHeaderMetadata._check_for_invalid_ids(multi_ids, 'sample')
self.assertEqual(result.result, RESULT.FAILURE)
def test_check_for_invalid_ids_3(self):
multi_ids = {'name': {'Ana'}}
result = SAMFileHeaderMetadata._check_for_invalid_ids(multi_ids, 'sample')
self.assertEqual(result.result, RESULT.SUCCESS)
def test_check_for_invalid_ids_4(self):
multi_ids = {'name': {'Ana', ''}, 'id': set(), 'accession_number': {'EGA1', None}}
result = SAMFileHeaderMetadata._check_for_invalid_ids(multi_ids, 'sample')
self.assertEqual(result.result, RESULT.FAILURE)
def test_filter_out_invalid_ids_4(self):
ids = {None, 123}
expected_result = {123}
actual_result = SAMFileHeaderMetadata._filter_out_invalid_ids(ids)
self.assertSetEqual(expected_result, actual_result)
def test_is_id_valid_2(self):
id = 'undefined'
self.assertFalse(SAMFileHeaderMetadata._is_id_valid(id))
def test_difference_when_there_are_diff_type_ids_h_vs_ss(self):
header_metadata = SAMFileHeaderMetadata('/seq/123.bam', samples={'name': set(['S1']), 'accession_number': set(['Acc1'])}, libraries={}, studies={})
seqscape_metadata = SeqscapeMetadata(samples={'name': set(['S1'])}, libraries={}, studies={})
result = header_metadata.difference(seqscape_metadata)
self.assertEqual(result, {})
def test_is_id_valid_5(self):
id = None
self.assertFalse(SAMFileHeaderMetadata._is_id_valid(id))
def test_is_id_valid_2(self):
id = 'undefined'
self.assertFalse(SAMFileHeaderMetadata._is_id_valid(id))
def test_is_id_valid_6(self):
id = "123"
self.assertTrue(SAMFileHeaderMetadata._is_id_valid(id))