def test_vcf_records_raisesTypeErrorWhenClosed(self):
file_contents = [
"##metaheader1\n",
self.entab("#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
record_iter = reader.vcf_records()
self.assertRaises(TypeError, next, record_iter)
def test_vcf_records_raisesTypeErrorWhenClosed(self):
file_contents = [
"##metaheader1\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
record_iter = reader.vcf_records()
self.assertRaises(TypeError, next, record_iter)
def test_get_format_tag_list(self):
file_contents = [
'##FORMAT=<ID=GT,Number=1>\n',
'##FORMAT=<ID=GQ,Number=1,Description="bar">\n', '#columnHeader\n',
'record1\n', 'record2'
]
mock_file_reader = MockFileReader("my_dir/my_file.txt", file_contents)
vcf_reader = VcfReader(mock_file_reader)
actual_format_set = vcf_reader.format_metaheaders
self.assertEquals(["GQ", "GT"], sorted(actual_format_set.keys()))
def test_init_sampleNamesInitialized(self):
file_contents = [
"##metaheader1\n", "##metaheader2\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleA|SampleB\n"
), "record1\n", "record2"
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
actual_vcf_reader = VcfReader(mock_reader)
self.assertEquals(["SampleA", "SampleB"],
actual_vcf_reader.sample_names)
def test_sort_delegatesToFileReader(self):
_FILE_CONTENTS = [
"##FORMAT=<ID=DP,Number=1,Type=Integer,Description='Read Depth'>\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
]
class ReversedSortMockFileReader(MockFileReader):
def __init__(self, filename):
MockFileReader.__init__(self, filename, _FILE_CONTENTS)
self.filename = filename
def __lt__(self, other):
return self.filename > other.filename
reader1 = VcfReader(ReversedSortMockFileReader("1.txt"))
reader2 = VcfReader(ReversedSortMockFileReader("2.txt"))
reader3 = VcfReader(ReversedSortMockFileReader("3.txt"))
actual_readers = sorted([reader1, reader2, reader3])
self.assertEquals([reader3, reader2, reader1], actual_readers)
def test_format_tag_ids_emptyWhenNoFormatTags(self):
file_contents = [
"##metaheader1\n",
"##INFO=<ID=AF,Number=A,Type=Float,Description='Allele Frequency'>\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
self.assertEquals(0, len(reader.format_metaheaders))
def test_get_info_field_list(self):
file_contents = [
'##INFO=<ID=AF,Number=1>\n',
'##FORMAT=<ID=GQ,Number=1,Description="bar">\n',
'##INFO=<ID=AA,Number=1>\n', '#columnHeader\n', 'record1\n',
'record2'
]
mock_file_reader = MockFileReader("my_dir/my_file.txt", file_contents)
vcf_reader = VcfReader(mock_file_reader)
actual_format_set = vcf_reader.info_metaheaders
expected_format_set = ["AA", "AF"]
self.assertEquals(expected_format_set,
sorted(actual_format_set.keys()))
def test_format_tag_ids_immutable(self):
file_contents = [
"##metaheader1\n",
"##FORMAT=<ID=DP,Number=1,Type=Integer,Description='Read Depth'>\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
self.assertEquals(["DP"], sorted(reader.format_metaheaders.keys()))
del reader.format_metaheaders["DP"]
self.assertEquals(["DP"], sorted(reader.format_metaheaders.keys()))
def test_format_tag_ids_ignoresRelatedFieldNames(self):
file_contents = [
"##metaheader1\n", "##FORMAT=<UUID=DPX1,ID=DP1>\n",
"##FORMAT=<ID=DP2,UUID=DPX2>\n", "##FORMAT=<ID=DP3,ID=DPX3>\n",
"##FORMAT=<NOID=DPX>\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
self.assertEquals(set(["DP1", "DP2", "DP3"]),
set(reader.format_metaheaders.keys()))
def test_init(self):
file_contents = [
"##metaheader1\n", "##metaheader2\n", "#columnHeader\n",
"record1\n", "record2"
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
actual_vcf_reader = VcfReader(mock_reader)
self.assertEquals("my_dir/my_file.txt",
actual_vcf_reader.input_filepath)
self.assertEquals("my_file.txt", actual_vcf_reader.file_name)
self.assertEquals("#columnHeader", actual_vcf_reader.column_header)
self.assertEquals(["##metaheader1", "##metaheader2"],
actual_vcf_reader.metaheaders)
self.assertEquals([], actual_vcf_reader.sample_names)
def test_format_tag_ids_idsAreUnique(self):
file_contents = [
"##metaheader1\n",
"##FORMAT=<ID=AF,Description='Allele Frequency 1'>\n",
"##FORMAT=<ID=AF,Description='Allele Frequency 2'>\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
self.assertEquals(["AF"], sorted(reader.format_metaheaders.keys()))
self.assertEquals("##FORMAT=<ID=AF,Description='Allele Frequency 2'>",
reader.format_metaheaders["AF"])
def test_format_metaheaders(self):
file_contents = [
"##metaheader1\n",
"##FORMAT=<ID=AF,Number=A,Type=Float,Description=\"Allele Frequency\">\n",
"##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"Read Depth\">\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
expected_metaheaders = {
"AF":
'##FORMAT=<ID=AF,Number=A,Type=Float,Description="Allele Frequency">',
"DP":
'##FORMAT=<ID=DP,Number=1,Type=Integer,Description="Read Depth">'
}
self.assertEquals(expected_metaheaders, reader.format_metaheaders)
def test_filter_metaheaders(self):
file_contents = [
"##metaheader1\n",
"##FORMAT=<ID=AF,Number=A,Type=Float,Description=\"Allele Frequency\">\n",
"##INFO=<ID=SNP,Number=1,Type=Integer,Description=\"snp\">\n",
"##FILTER=<ID=.,Number=1,Type=Integer,Description=\"foo\">\n",
"##FILTER=<ID=PASS,Number=1,Type=Integer,Description=\"bar\">\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
expected_metaheaders = {
".": "##FILTER=<ID=.,Number=1,Type=Integer,Description=\"foo\">",
"PASS":
"******"bar\">",
}
self.assertEquals(expected_metaheaders, reader.filter_metaheaders)
def test_vcf_records(self):
file_contents = [
"##metaheader1\n", "##metaheader2\n",
self.entab(
"#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"
),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR")
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
actual_vcf_records = []
reader.open()
for vcf_record in reader.vcf_records():
actual_vcf_records.append(vcf_record)
reader.close()
self.assertEquals(2, len(actual_vcf_records))
self.assertEquals('chr1', actual_vcf_records[0].chrom)
self.assertEquals('chr2', actual_vcf_records[1].chrom)
self.assertTrue(mock_reader.open_was_called)
self.assertTrue(mock_reader.close_was_called)
def test_vcf_records(self):
file_contents = [
"##metaheader1\n",
"##metaheader2\n",
self.entab("#CHROM|POS|ID|REF|ALT|QUAL|FILTER|INFO|FORMAT|SampleNormal|SampleTumor\n"),
self.entab("chr1|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR\n"),
self.entab("chr2|1|.|A|C|.|.|INFO|FORMAT|NORMAL|TUMOR"),
]
mock_reader = MockFileReader("my_dir/my_file.txt", file_contents)
reader = VcfReader(mock_reader)
actual_vcf_records = []
reader.open()
for vcf_record in reader.vcf_records():
actual_vcf_records.append(vcf_record)
reader.close()
self.assertEquals(2, len(actual_vcf_records))
self.assertEquals("chr1", actual_vcf_records[0].chrom)
self.assertEquals("chr2", actual_vcf_records[1].chrom)
self.assertTrue(mock_reader.open_was_called)
self.assertTrue(mock_reader.close_was_called)
