def test_merge_header_definitions_no_conflicting_headers(self):
lines_1 = [
'##FORMAT=<ID=NS,Number=1,Type=Float,Description="Number samples">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample1 Sample2\n'
]
lines_2 = [
'##FORMAT=<ID=DP,Number=2,Type=Float,Description="Total Depth">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample3\n'
]
vcf_reader_1 = vcf.Reader(fsock=iter(lines_1))
vcf_reader_2 = vcf.Reader(fsock=iter(lines_2))
headers_1 = self._get_vcf_header_from_reader(vcf_reader_1, 'file1')
headers_2 = self._get_vcf_header_from_reader(vcf_reader_2, 'file2')
pipeline = TestPipeline()
merged_definitions = (
pipeline
| Create([headers_1, headers_2])
|
'MergeDefinitions' >> merge_header_definitions.MergeDefinitions())
expected = VcfHeaderDefinitions()
expected._formats = {
'NS': {
Definition(1, 'Float'): ['file1']
},
'DP': {
Definition(2, 'Float'): ['file2']
}
}
assert_that(merged_definitions, equal_to([expected]))
pipeline.run()
def test_merge_header_definitions_save_five_copies(self):
lines_1 = [
'##INFO=<ID=NS,Number=1,Type=Float,Description="Number samples">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample1 Sample2\n'
]
lines_2 = [
'##INFO=<ID=NS,Number=1,Type=Integer,Description="Number samples">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample3\n'
]
vcf_reader_1 = vcf.Reader(fsock=iter(lines_1))
vcf_reader_2 = vcf.Reader(fsock=iter(lines_2))
file_names = ['file1', 'file2', 'file3', 'file4', 'file5', 'file6']
headers = []
for file_name in file_names:
headers.append(self._get_vcf_header_from_reader(vcf_reader_1, file_name))
headers.append(self._get_vcf_header_from_reader(vcf_reader_2, 'file7'))
pipeline = TestPipeline()
merged_definitions = (
pipeline
| Create(headers)
| 'MergeDefinitions' >> merge_header_definitions.MergeDefinitions())
expected = VcfHeaderDefinitions()
expected._infos = {
'NS': {Definition(1, 'Float'):
['file1', 'file2', 'file3', 'file4', 'file5'],
Definition(1, 'Integer'): ['file7']}}
assert_that(merged_definitions, equal_to([expected]))
pipeline.run()
def test_report_conflicted_and_inferred_headers(self):
header_definitions = VcfHeaderDefinitions()
header_definitions._infos = {
'NS': {
Definition(1, 'Float'): ['file1'],
Definition(1, 'Integer'): ['file2']
}
}
infos = OrderedDict([('NS',
createInfo('NS', 1, 'Float', 'Number samples',
None, None))])
formats = OrderedDict([('DP',
createFormat('DP', 2, 'Float',
'Total Depth'))])
resolved_headers = VcfHeader(infos=infos, formats=formats)
inferred_headers = VcfHeader(formats=formats)
expected = [
preprocess_reporter._InconsistencyType.HEADER_CONFLICTS + '\n',
preprocess_reporter._HeaderLine.CONFLICTS_HEADER + '\n',
(preprocess_reporter._DELIMITER).join([
'NS', 'INFO', 'num=1 type=Float', 'file1', 'num=1 type=Float\n'
]), (preprocess_reporter._DELIMITER).join(
[' ', ' ', 'num=1 type=Integer', 'file2', ' \n']), '\n',
preprocess_reporter._InconsistencyType.INFERRED_HEADERS + '\n',
preprocess_reporter._HeaderLine.INFERRED_FIELD_HEADER + '\n',
(preprocess_reporter._DELIMITER).join(
['DP', 'FORMAT', 'num=2 type=Float\n']), '\n'
]
self._generate_report_and_assert_contents_equal(
expected, header_definitions, resolved_headers, inferred_headers)
def test_report_multiple_files(self):
header_definitions = VcfHeaderDefinitions()
header_definitions._infos = {
'NS': {
Definition(1, 'Float'): ['file1', 'file2'],
Definition(1, 'Integer'): ['file3']
}
}
infos = OrderedDict([('NS',
createInfo('NS', 1, 'Float', 'Number samples',
None, None))])
resolved_headers = VcfHeader(infos=infos)
expected = [
preprocess_reporter._InconsistencyType.HEADER_CONFLICTS + '\n',
preprocess_reporter._HeaderLine.CONFLICTS_HEADER + '\n',
(preprocess_reporter._DELIMITER).join([
'NS', 'INFO', 'num=1 type=Float', 'file1', 'num=1 type=Float\n'
]), (preprocess_reporter._DELIMITER).join(
[' ', ' ', ' ', 'file2', ' \n']),
(preprocess_reporter._DELIMITER).join(
[' ', ' ', 'num=1 type=Integer', 'file3', ' \n']), '\n'
]
self._generate_report_and_assert_contents_equal(
expected, header_definitions, resolved_headers)
def test_create_empty_header_defitions(self):
expected_info = {}
expected_format = {}
header_definitions = VcfHeaderDefinitions()
self.assertDictEqual(header_definitions.infos, expected_info)
self.assertDictEqual(header_definitions.formats, expected_format)
def test_type_check(self):
merger = DefinitionsMerger()
empty_header_definitions = VcfHeaderDefinitions()
with self.assertRaises(NotImplementedError):
merger.merge(empty_header_definitions, None)
with self.assertRaises(NotImplementedError):
merger.merge(None, empty_header_definitions)
def test_create_definitions_multi(self):
lines = [
'##INFO=<ID=NS,Number=1,Type=Integer,Description="Number samples">\n',
'##INFO=<ID=DP,Number=2,Type=Float,Description="Number samples">\n',
'##FORMAT=<ID=NS,Number=3,Type=Integer,Description="Number samples">\n',
'##FORMAT=<ID=DP,Number=4,Type=Float,Description="Number samples">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample1 Sample2\n'
]
vcf_reader = vcf.Reader(fsock=iter(lines))
header = self._get_vcf_header_from_reader(vcf_reader, 'file1')
expected_info = {
'NS': {
Definition(1, 'Integer'): ['file1']
},
'DP': {
Definition(2, 'Float'): ['file1']
}
}
expected_format = {
'NS': {
Definition(3, 'Integer'): ['file1']
},
'DP': {
Definition(4, 'Float'): ['file1']
}
}
header_definitions = VcfHeaderDefinitions(header)
self.assertDictEqual(header_definitions.infos, expected_info)
self.assertDictEqual(header_definitions.formats, expected_format)
def test_merge_header_definitions_one_header(self):
lines = [
'##INFO=<ID=NS,Number=1,Type=Integer,Description="Number samples">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample1 Sample2\n'
]
vcf_reader = vcf.Reader(fsock=iter(lines))
headers = self._get_vcf_header_from_reader(vcf_reader, 'file1')
pipeline = TestPipeline()
merged_definitions = (
pipeline
| Create([headers])
| 'MergeDefinitions' >> merge_header_definitions.MergeDefinitions())
expected = VcfHeaderDefinitions()
expected._infos = {'NS': {Definition(1, 'Integer'): ['file1']}}
assert_that(merged_definitions, equal_to([expected]))
pipeline.run()
def test_report_no_conflicts(self):
header_definitions = VcfHeaderDefinitions()
header_definitions._infos = {'NS': {Definition(1, 'Float'): ['file1']}}
header_definitions._formats = {
'NS': {
Definition(1, 'Float'): ['file2']
}
}
infos = OrderedDict([('NS',
Info('NS', 1, 'Integer', 'Number samples', None,
None))])
formats = OrderedDict([('NS', Format('NS', 1, 'Float',
'Number samples'))])
resolved_headers = VcfHeader(infos=infos, formats=formats)
expected = ['No Header Conflicts Found.\n', '\n']
self._generate_report_and_assert_contents_equal(
expected, header_definitions, resolved_headers)
def test_report_no_resolved_headers(self):
header_definitions = VcfHeaderDefinitions()
header_definitions._infos = {
'NS': {
Definition(1, 'Float'): ['file1'],
Definition(1, 'Integer'): ['file2']
}
}
expected = [
preprocess_reporter._InconsistencyType.HEADER_CONFLICTS + '\n',
preprocess_reporter._HeaderLine.CONFLICTS_HEADER + '\n',
(preprocess_reporter._DELIMITER).join(
['NS', 'INFO', 'num=1 type=Float', 'file1',
'Not resolved.\n']), (preprocess_reporter._DELIMITER).join(
[' ', ' ', 'num=1 type=Integer', 'file2', ' \n']), '\n'
]
self._generate_report_and_assert_contents_equal(
expected, header_definitions)
def test_report_no_inconsistencies(self):
header_definitions = VcfHeaderDefinitions()
inferred_headers = VcfHeader()
expected = [
'No Header Conflicts Found.\n', '\n',
'No Inferred Headers Found.\n', '\n',
'No Malformed Records Found.\n', '\n'
]
self._generate_report_and_assert_contents_equal(
expected,
header_definitions,
inferred_headers=inferred_headers,
malformed_records=[])
def test_create_definitions_with_info(self):
lines = [
'##INFO=<ID=NS,Number=1,Type=Float,Description="Number samples">\n',
'#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT Sample1 Sample2\n'
]
header = self._get_header_from_lines(lines, 'file1')
expected_info = {'NS': {Definition(1, 'Float'): ['file1']}}
expected_format = {}
header_definitions = VcfHeaderDefinitions(header)
self.assertDictEqual(header_definitions.infos, expected_info)
self.assertDictEqual(header_definitions.formats, expected_format)
def test_report_inferred_headers_only(self):
header_definitions = VcfHeaderDefinitions()
formats = OrderedDict([('DP', Format('DP', 2, 'Float',
'Total Depth'))])
inferred_headers = VcfHeader(formats=formats)
expected = [
'No Header Conflicts Found.\n', '\n',
preprocess_reporter._InconsistencyType.INFERRED_HEADERS + '\n',
preprocess_reporter._HeaderLine.INFERRED_FIELD_HEADER + '\n',
(preprocess_reporter._DELIMITER).join(
['DP', 'FORMAT', 'num=2 type=Float\n']), '\n'
]
self._generate_report_and_assert_contents_equal(
expected, header_definitions, inferred_headers=inferred_headers)
def test_report_malformed_records(self):
header_definitions = VcfHeaderDefinitions()
inferred_headers = VcfHeader()
records = [
vcfio.MalformedVcfRecord('file1', 'rs6054257 G A 29 PASS',
'Invalid literal')
]
expected = [
'No Header Conflicts Found.\n', '\n',
'No Inferred Headers Found.\n', '\n',
preprocess_reporter._InconsistencyType.MALFORMED_RECORDS + '\n',
preprocess_reporter._HeaderLine.MALFORMED_RECORDS_HEADER + '\n',
(preprocess_reporter._DELIMITER).join(
['file1', 'rs6054257 G A 29 PASS',
'Invalid literal\n']), '\n'
]
self._generate_report_and_assert_contents_equal(
expected,
header_definitions,
inferred_headers=inferred_headers,
malformed_records=records)
def _create_definitions_from_lines(self, lines, file_name): header = self._get_header_from_lines(lines, file_name) return VcfHeaderDefinitions(header)
def _create_definitions_from_lines(self, lines, file_name):
vcf_reader = vcf.Reader(fsock=iter(lines))
header = self._get_vcf_header_from_reader(vcf_reader, file_name)
return VcfHeaderDefinitions(header)