def test_validate_vcf_hc_pairs_raisesIfNoMissingVcfFiles(self):
vcf_hc_pairs = [(MockFileReader("A.vcf"), MockFileReader("A.hc")),
(None, MockFileReader("B.hc"))]
self.assertRaisesRegexp(
utils.UsageError,
r"The VarScan high-confidence file \[B.hc\] has no matching VCF file.",
self.caller._validate_vcf_hc_pairs, vcf_hc_pairs)
def test_claim_vcfAndFilterFileNameGiven(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [self.entab("chrom|position|ref|var"), record1]
content2 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("patientA.indel.Somatic.foo.bar", content1)
reader2 = MockFileReader("patientA.indel.vcf", content2)
reader3 = MockFileReader("patientA.snp.Somatic.foo.bar", content1)
reader4 = MockFileReader("patientA.snp.vcf", content2)
reader5 = MockFileReader("patientA.readme", ["foo"])
file_readers = [reader1, reader2, reader3, reader4, reader5]
caller = varscan.Varscan()
caller.hc_file_pattern = re.compile("foo.bar$")
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(1, len(unrecognized_readers))
self.assertEquals([reader5], unrecognized_readers)
self.assertEquals(2, len(vcf_readers))
self.assertIsInstance(vcf_readers[0], varscan._VarscanVcfReader)
self.assertEquals(reader2.file_name,
vcf_readers[0]._vcf_reader.file_name)
self.assertEquals(reader1.file_name,
vcf_readers[0]._som_hc_file_reader.file_name)
self.assertIsInstance(vcf_readers[1], varscan._VarscanVcfReader)
self.assertEquals(reader4.file_name,
vcf_readers[1]._vcf_reader.file_name)
self.assertEquals(reader3.file_name,
vcf_readers[1]._som_hc_file_reader.file_name)
def append_hc_files(readers,
file1="snp.somatic.hc.fpfilter.pass",
file2="indel.somatic.hc.fpfilter.pass",
content1=None,
content2=None):
if not content1:
content1 = []
if not content2:
content2 = []
readers.append(MockFileReader(file1, content1))
readers.append(MockFileReader(file2, content2))
def test_claim_allSnpOrIndelOkay(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=strelka", "#chrom", record1]
reader1 = MockFileReader("fileA.indels.vcf", content1)
reader2 = MockFileReader("fileB.indels.vcf", content1)
file_readers = [reader1, reader2]
caller = strelka.Strelka()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(0, len(unrecognized_readers))
self.assertEquals(2, len(vcf_readers))
def test_claim_ignoresOtherCallers(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=Foo", "#chrom", record1]
content2 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("fileA.txt", content1)
reader2 = MockFileReader("fileA.vcf", content2)
file_readers = [reader1, reader2]
caller = varscan.Varscan()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(1, len(unrecognized_readers))
self.assertEquals([reader1], unrecognized_readers)
self.assertEquals(1, len(vcf_readers))
def test_claim_VCFHasNoMatchingHc(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [self.entab("chrom|position"), record1]
content2 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("patientA.vcf", content2)
reader2 = MockFileReader("patientB.vcf", content2)
reader3 = MockFileReader("patientB.snp.Somatic.hc.fpfilter.pass",
content1)
file_readers = [reader1, reader2, reader3]
caller = varscan.Varscan()
self.assertRaisesRegexp(
utils.UsageError,
r"The VarScan VCF file \[patientA.vcf\] has no matching high-confidence file.",
caller.claim, file_readers)
def test_claim_mismatchingSnpIndelFiles(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=strelka", "#chrom", record1]
reader1 = MockFileReader("fileA.snvs.vcf", content1)
reader2 = MockFileReader("fileA.indels.vcf", content1)
reader3 = MockFileReader("fileB.indels.vcf", content1)
file_readers = [reader1, reader2, reader3]
caller = strelka.Strelka()
self.assertRaisesRegexp(
utils.JQException,
r"Some Strelka VCFs were missing either a snvs or indels file. Review inputs/command options and try again.",
caller.claim, file_readers)
actual_log_errors = test.utils.mock_logger.messages["ERROR"]
expected_log_errors = ["Strelka VCF [fileB.indels] has no snvs file."]
self.assertEquals(expected_log_errors, actual_log_errors)
def test_claim_varscanVcfOnly(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("fileA.snp.vcf", content1)
reader2 = MockFileReader("fileB.snp.vcf", content1)
file_readers = [reader1, reader2]
caller = varscan.Varscan()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(0, len(unrecognized_readers))
self.assertEquals([], unrecognized_readers)
self.assertEquals(2, len(vcf_readers))
self.assertIsInstance(vcf_readers[0], varscan._VarscanVcfReader)
self.assertEquals(reader1.file_name,
vcf_readers[0]._vcf_reader.file_name)
def test_claim(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=strelka", "#chrom", record1]
content2 = ["##foo", "##MuTect", "#chrom", record1]
reader1 = MockFileReader("fileA.vcf", content1)
reader2 = MockFileReader("fileB.vcf", content2)
file_readers = [reader1, reader2]
caller = strelka.Strelka()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(1, len(unrecognized_readers))
self.assertEquals([reader2], unrecognized_readers)
self.assertEquals(1, len(vcf_readers))
self.assertIsInstance(vcf_readers[0], strelka._StrelkaVcfReader)
self.assertEquals(reader1.file_name,
vcf_readers[0]._vcf_reader.file_name)
def test_claim_vcfAndInvalidFilterFile(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [self.entab("chrom|pos|ref|alt"), record1]
content2 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("patientA.indel.Somatic.hc.fpfilter.pass",
content1)
reader2 = MockFileReader("patientA.indel.vcf", content2)
reader3 = MockFileReader("patientA.snp.Somatic.hc.fpfilter.pass",
content1)
reader4 = MockFileReader("patientA.snp.vcf", content2)
reader5 = MockFileReader("patientA.readme", ["foo"])
file_readers = [reader1, reader2, reader3, reader4, reader5]
caller = varscan.Varscan()
self.assertRaisesRegexp(
utils.JQException,
r"The \[2\] input files \[.*\] match high-confidence file names, but the file header is invalid or missing. Review inputs and try again.",
caller.claim, file_readers)
def test_add_tag_values_highConfidenceDoesNotChangeFilter(self):
record = vcf.VcfRecord("chr1", "42", "ref", "alt", vcf_filter="pass")
input_reader = MockFileReader("foo.txt", [
"chrom\tposition\tref\tvar", "chr1\t42\tref\tvar",
"chr2\t50\tref\tvar"
])
expected = "pass"
actual = _HCTag(input_reader).add_tag_values(record).filter
self.assertEquals(expected, actual)
def test_add_tag_values_lowConfidencePassingReplacesFilter(self):
record = vcf.VcfRecord("chr1", "30", "ref", "alt", vcf_filter="pass")
input_reader = MockFileReader("foo.txt", [
"chrom\tposition\tref\tvar", "chr1\t42\tref\tvar",
"chr2\t50\tref\tvar"
])
tag = _HCTag(input_reader)
expected = tag._TAG_ID
actual = tag.add_tag_values(record).filter
self.assertEquals(expected, actual)
def test_claim(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=strelka", "#chrom", record1]
content2 = [
"##foo", "##MuTect=123", "##FORMAT=<ID=AF,...>", "#chrom", record1
]
reader1 = MockFileReader("fileA.vcf", content1)
reader2 = MockFileReader("fileB.vcf", content2)
file_readers = [reader1, reader2]
caller = mutect.Mutect()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(1, len(unrecognized_readers))
self.assertEquals([reader1], unrecognized_readers)
self.assertEquals(1, len(vcf_readers))
self.assertIsInstance(vcf_readers[0], mutect._MutectVcfReader)
self.assertEquals(reader2.file_name,
vcf_readers[0]._vcf_reader.file_name)
def test_claim_multiplePatients(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=VarScan2", "#chrom", record1]
content2 = ["chrom\tposition", "1\t23"]
reader1 = MockFileReader("p2.fileA.vcf", content1)
reader2 = MockFileReader("p2.fileA.Somatic.hc.fpfilter.pass", content2)
reader3 = MockFileReader("p3.fileA.Somatic.hc.fpfilter.pass", content2)
reader4 = MockFileReader("p3.fileA.vcf", content1)
file_readers = [reader1, reader2, reader3, reader4]
caller = varscan.Varscan()
dummy, vcf_readers = caller.claim(file_readers)
self.assertEquals(2, len(vcf_readers))
self.assertIsInstance(vcf_readers[0], varscan._VarscanVcfReader)
self.assertEquals("p2.fileA.vcf", vcf_readers[0]._vcf_reader.file_name)
self.assertIn("_HCTag", self._get_tag_class_names(vcf_readers[0]))
self.assertIn("_HCTag", self._get_tag_class_names(vcf_readers[1]))
self.assertEquals(reader1.file_name,
vcf_readers[0]._vcf_reader.file_name)
def test_claim_filterRegexDoesNotMatch(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("patientA.vcf", content1)
file_readers = [reader1]
caller = varscan.Varscan()
caller.hc_file_pattern = re.compile("foo.bar$")
self.assertRaisesRegexp(
utils.UsageError,
r"The VarScan high-confidence filename regex \[foo\.bar\$\] didn't match any files in the input directory. The beginning of the high-confidence filename must exactly match a VCF filename up to the .vcf extension. Review inputs/command options and try again.",
caller.claim, file_readers)
def test_claim_ignoresNonVcfExtensions(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##MuTect=123", "#chrom", record1]
reader1 = MockFileReader("fileA.txt", content1)
file_readers = [reader1]
caller = mutect.Mutect()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(1, len(unrecognized_readers))
self.assertEquals([reader1], unrecognized_readers)
self.assertEquals(0, len(vcf_readers))
def test_claim_ignores_non_vcf_files(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = ["##foo", "##source=strelka", "#chrom", record1]
reader1 = MockFileReader("fileA.txt", content1)
file_readers = [reader1]
caller = strelka.Strelka()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(1, len(unrecognized_readers))
self.assertEquals([reader1], unrecognized_readers)
self.assertEquals(0, len(vcf_readers))
def test_claim_metaheaderRecognizesMutectV2x(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [
"##foo", "##MuTect=2.1", "##FORMAT=<ID=FA,...>", "#chrom", record1
]
reader1 = MockFileReader("fileA.vcf", content1)
file_readers = [reader1]
caller = mutect.Mutect()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(0, len(unrecognized_readers))
self.assertEquals(1, len(vcf_readers))
def test_claim_vcfExtensionCaseInsensitive(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [
"##foo", "##MuTect=123", "##FORMAT=<ID=AF,...>", "#chrom", record1
]
reader1 = MockFileReader("fileA.VcF", content1)
file_readers = [reader1]
caller = mutect.Mutect()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(0, len(unrecognized_readers))
self.assertEquals(1, len(vcf_readers))
def test_claim_metaheaderRecognizesMutectV4x(self):
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [
"##foo",
'##GATKCommandLine=<ID=Mutect2,CommandLine="Mutect2 ...">',
"##FORMAT=<ID=AF,...>", "#chrom", record1
]
reader1 = MockFileReader("fileB.vcf", content1)
file_readers = [reader1]
caller = mutect.Mutect()
unrecognized_readers, vcf_readers = caller.claim(file_readers)
self.assertEquals(0, len(unrecognized_readers))
self.assertEquals(1, len(vcf_readers))
def test_vcf_records_SomHcFileSNP(self):
record1 = vcf.VcfRecord(chrom="chr1",
pos="21",
ref="A",
alt="G",
vcf_filter="PASS")
record2 = vcf.VcfRecord(chrom="chr1",
pos="22",
ref="A",
alt="T",
vcf_filter="PASS")
vcf_reader = MockVcfReader(records=[record1, record2])
content1 = ["chrom\tposition", "chr1\t21", "chr1\t22"]
somatic_hc_reader = MockFileReader("fileA.Somatic.hc.fpfilter.pass",
content1)
varscan_vcf_reader = varscan._VarscanVcfReader(vcf_reader,
somatic_hc_reader)
vcf_records = [record for record in varscan_vcf_reader.vcf_records()]
self.assertEquals(2, len(vcf_records))
self.assertIn("PASS", vcf_records[0].filter)
self.assertIn("PASS", vcf_records[1].filter)
def test_validate_vcf_hc_pairs(self):
self.caller._validate_vcf_hc_pairs([
(MockFileReader("A.vcf"), MockFileReader("A.hc")),
(MockFileReader("B.vcf"), MockFileReader("B.hc"))
])
self.ok()
def test_validate_vcf_hc_pairs_okIfNoHcFiles(self):
self.caller._validate_vcf_hc_pairs([(MockFileReader("A.vcf"), None),
(MockFileReader("B.vcf"), None)])
self.ok()
def test_validate_filter_file_validFile(self):
file_reader = MockFileReader("p1.hc.fpfilter.pass",
["chrom\tposition"])
caller = varscan.Varscan()
valid_reader = caller._validate_filter_file(file_reader)
self.assertEquals("p1.hc.fpfilter.pass", valid_reader.file_name)
def test_validate_filter_file_invalidFile(self):
file_reader = MockFileReader("p1.hc.fpfilter.pass",
["chrom\tpos\tref"])
caller = varscan.Varscan()
valid_reader = caller._validate_filter_file(file_reader)
self.assertEquals(None, valid_reader)
def test_claim_vcfAnd6InvalidFilterFiles(self):
#pylint: disable=too-many-locals
record1 = "chr1\t.\t.\t.\t.\t.\t.\t.\t."
content1 = [self.entab("chrom|pos|ref|alt"), record1]
content2 = ["##foo", "##source=VarScan2", "#chrom", record1]
reader1 = MockFileReader("patientA.snp.Somatic.hc.fpfilter.pass",
content1)
reader2 = MockFileReader("patientA.snp.vcf", content2)
reader3 = MockFileReader("patientB.snp.Somatic.hc.fpfilter.pass",
content1)
reader4 = MockFileReader("patientB.snp.vcf", content2)
reader5 = MockFileReader("patientC.snp.Somatic.hc.fpfilter.pass",
content1)
reader6 = MockFileReader("patientC.snp.vcf", content2)
reader7 = MockFileReader("patientD.snp.Somatic.hc.fpfilter.pass",
content1)
reader8 = MockFileReader("patientD.snp.vcf", content2)
reader9 = MockFileReader("patientE.snp.Somatic.hc.fpfilter.pass",
content1)
reader10 = MockFileReader("patientE.snp.vcf", content2)
reader11 = MockFileReader("patientF.snp.Somatic.hc.fpfilter.pass",
content1)
reader12 = MockFileReader("patientF.snp.vcf", content2)
file_readers = [
reader1, reader2, reader3, reader4, reader5, reader6, reader7,
reader8, reader9, reader10, reader11, reader12
]
caller = varscan.Varscan()
self.assertRaisesRegexp(
utils.JQException,
r"The \[6\] input files \[.*\(1 file\(s\) omitted\)\] match high-confidence file names, but the file header is invalid or missing. Review inputs and try again.",
caller.claim, file_readers)
