def test_validation_correction_valid(self):
""" Test that the validation allele fields are determined automatically when not specified by the user for a valid mutation.
"""
m = MutationDataFactory.default_create()
m.chr = "3"
m.start = "178948145"
m.end = "178948145"
m.alt_allele = "A"
m.ref_allele = "G"
m['validation_status'] = "Valid"
m['Match_Norm_Validation_Allele1'] = ""
m['Match_Norm_Validation_Allele2'] = ""
m['Tumor_Validation_Allele1'] = ""
m['Tumor_Validation_Allele2'] = ""
m['Mutation_Status'] = "Somatic"
output_filename = os.path.join("out", "test_validation_correction2.maf.tsv")
outputRenderer = TcgaMafOutputRenderer(output_filename,
configFile=os.path.join("configs", "tcgaMAF2.4_output.config"))
outputRenderer.renderMutations([m].__iter__())
tsv_reader = GenericTsvReader(output_filename)
for line_dict in tsv_reader:
self.assertTrue(line_dict['Match_Norm_Validation_Allele1'] == line_dict['Match_Norm_Validation_Allele2'], "Matched norm alleles did not match.")
self.assertTrue(line_dict['Tumor_Validation_Allele1'] == line_dict['Reference_Allele'], "Tumor validation allele 1 did not match reference for a valid validation result.")
self.assertTrue(line_dict['Tumor_Validation_Allele2'] == line_dict['Tumor_Seq_Allele2'], "Tumor validation allele 2 did not match Tumor_Seq_Allele2 for a valid validation result.")
self.assertTrue(line_dict['Match_Norm_Validation_Allele1'] == line_dict['Tumor_Validation_Allele1'], "Tumor allele 1 did not match normal alleles for a valid validation result.")
self.assertTrue(line_dict['Match_Norm_Validation_Allele1'] == line_dict['Reference_Allele'], "Norm validation alleles did not match reference (norm, reference): (%s, %s)" %(line_dict['Match_Norm_Validation_Allele1'] ,line_dict['Reference_Allele']) )
self.assertTrue("G" == line_dict['Reference_Allele'], "Reference allele should have been G, but was " + line_dict['Reference_Allele'])
self.assertTrue("A" == line_dict['Tumor_Seq_Allele2'], "Alt allele should have been A, but was " + line_dict['Tumor_Seq_Allele2'])
def testTCGAMAFAsInput(self):
""" Test that we can take in a TCGA MAF (using MAFLITE), do no annotations, and still render it properly """
tmp = MafliteInputMutationCreator("testdata/maf/Patient0.maf.annotated", 'configs/maflite_input.config')
muts = tmp.createMutations()
outputFilename = "out/testTCGAMAFAsInput.tsv"
outputRenderer = TcgaMafOutputRenderer(outputFilename, 'configs/tcgaMAF2.4_output.config')
outputRenderer.renderMutations(muts, tmp.getComments())
def test_validation_correction(self):
""" Test that the validation allele fields are determined automatically when not specified by the user for invalid mutation.
"""
m = MutationDataFactory.default_create()
m.chr = "3"
m.start = "178948145"
m.end = "178948145"
m.alt_allele = "A"
m.ref_allele = "G"
m['validation_status'] = "Invalid"
m['Match_Norm_Validation_Allele1'] = ""
m['Match_Norm_Validation_Allele2'] = ""
m['Tumor_Validation_Allele1'] = ""
m['Tumor_Validation_Allele2'] = ""
m['Mutation_Status'] = "Somatic"
output_filename = os.path.join("out",
"test_validation_correction1.maf.tsv")
outputRenderer = TcgaMafOutputRenderer(output_filename,
configFile=os.path.join(
"configs",
"tcgaMAF2.4_output.config"))
outputRenderer.renderMutations([m].__iter__())
tsv_reader = GenericTsvReader(output_filename)
for line_dict in tsv_reader:
self.assertTrue(
line_dict['Match_Norm_Validation_Allele1'] ==
line_dict['Match_Norm_Validation_Allele2'],
"Matched norm alleles did not match.")
self.assertTrue(
line_dict['Tumor_Validation_Allele1'] ==
line_dict['Tumor_Validation_Allele2'],
"Tumor alleles did not match for an invalid validation result."
)
self.assertTrue(
line_dict['Match_Norm_Validation_Allele1'] ==
line_dict['Tumor_Validation_Allele2'],
"Tumor alleles did not match normal alleles for an invalid validation result."
)
self.assertTrue(
line_dict['Match_Norm_Validation_Allele1'] ==
line_dict['Reference_Allele'],
"Norm validation alleles did not match reference (norm, reference): (%s, %s)"
% (line_dict['Match_Norm_Validation_Allele1'],
line_dict['Reference_Allele']))
self.assertTrue(
"G" == line_dict['Reference_Allele'],
"Reference allele should have been G, but was " +
line_dict['Reference_Allele'])
self.assertTrue(
"None" == line_dict['Mutation_Status'],
"Mutation Status must be None when Validation Status is Invalid: "
+ line_dict['Mutation_Status'])
def testTCGAMAFAsInput(self):
""" Test that we can take in a TCGA MAF (using MAFLITE), do no annotations, and still render it properly """
tmp = MafliteInputMutationCreator(
"testdata/maf/Patient0.maf.annotated", None,
'configs/maflite_input.config')
muts = tmp.createMutations()
outputFilename = "out/testTCGAMAFAsInput.tsv"
outputRenderer = TcgaMafOutputRenderer(
outputFilename, 'configs/tcgaMAF2.4_output.config')
outputRenderer.renderMutations(muts, tmp.getComments())
def testInternalFieldsSkipPrepend(self):
""" Test that no prepending of "i_" is honored."""
outputFilename = "out/testInternalFields_v2.4.maf.tsv"
m = MutationDataFactory.default_create()
m.createAnnotation("TEST", "THIS IS A TEST", "TESTING")
# The next annotation is real and should not be considered internal.
m.createAnnotation("gene", "EGFR")
outputRenderer = TcgaMafOutputRenderer(
outputFilename,
configFile='configs/tcgaMAF2.4_output.config',
other_options={OptionConstants.NO_PREPEND: True})
outputRenderer.renderMutations(iter([m]), ['No comments'])
configFile = ConfigUtils.createConfigParser(
'configs/tcgaMAF2.4_output.config')
requiredColumns = configFile.get("general", "requiredColumns")
self.assertTrue(
"Hugo_Symbol" in requiredColumns,
" This test assumes that Hugo_Symbol is a required column in the TCGA MAF. If not, the test must be modified."
)
statinfo = os.stat(outputFilename)
self.assertTrue(
statinfo.st_size > 0,
"Generated MAF file (" + outputFilename + ") is empty.")
tsvReader = GenericTsvReader(outputFilename)
headers = tsvReader.getFieldNames()
self.assertTrue("Hugo_Symbol" in headers,
"Hugo_Symbol not found in output headers")
self.assertTrue(
"i_TEST" not in headers,
"i_TEST was found in output headers when prepend was disabled.")
self.assertTrue("TEST" in headers,
"TEST was not found in output headers.")
def testInternalFieldsSkipPrepend(self):
""" Test that no prepending of "i_" is honored."""
outputFilename = "out/testInternalFields_v2.4.maf.tsv"
m = MutationDataFactory.default_create()
m.createAnnotation("TEST", "THIS IS A TEST", "TESTING")
# The next annotation is real and should not be considered internal.
m.createAnnotation("gene", "EGFR")
outputRenderer = TcgaMafOutputRenderer(outputFilename, configFile='configs/tcgaMAF2.4_output.config', other_options={OptionConstants.NO_PREPEND:True})
outputRenderer.renderMutations(iter([m]), ['No comments'])
configFile = ConfigUtils.createConfigParser('configs/tcgaMAF2.4_output.config')
requiredColumns = configFile.get("general", "requiredColumns")
self.assertTrue("Hugo_Symbol" in requiredColumns, " This test assumes that Hugo_Symbol is a required column in the TCGA MAF. If not, the test must be modified.")
statinfo = os.stat(outputFilename)
self.assertTrue(statinfo.st_size > 0, "Generated MAF file (" + outputFilename + ") is empty.")
tsvReader = GenericTsvReader(outputFilename)
headers = tsvReader.getFieldNames()
self.assertTrue("Hugo_Symbol" in headers, "Hugo_Symbol not found in output headers")
self.assertTrue("i_TEST" not in headers, "i_TEST was found in output headers when prepend was disabled.")
self.assertTrue("TEST" in headers, "TEST was not found in output headers.")
def testInternalFields(self):
""" Test that an annotation that is not listed explicitly in the required or optional columns is rendered with i_ prepended """
outputFilename = "out/testInternalFields_v2.4.maf.tsv"
m = MutationData()
m.createAnnotation("TEST", "THIS IS A TEST", "TESTING")
# The next annotation is real and should not be considered internal.
m.createAnnotation("gene", "EGFR")
outputRenderer = TcgaMafOutputRenderer(outputFilename, configFile='configs/tcgaMAF2.4_output.config')
outputRenderer.renderMutations(iter([m]), ['No comments'])
configFile = ConfigUtils.createConfigParser('configs/tcgaMAF2.4_output.config')
requiredColumns = configFile.get("general", "requiredColumns")
self.assertTrue("Hugo_Symbol" in requiredColumns, " This test assumes that Hugo_Symbol is a required column in the TCGA MAF. If not, the test must be modified.")
statinfo = os.stat(outputFilename)
self.assertTrue(statinfo.st_size > 0, "Generated MAF file (" + outputFilename + ") is empty.")
tsvReader = GenericTsvReader(outputFilename)
headers = tsvReader.getFieldNames()
self.assertTrue("Hugo_Symbol" in headers, "Hugo_Symbol not found in output headers")
self.assertTrue("TEST" not in headers, "TEST was found in output headers when it should have been renamed to i_TEST")
self.assertTrue("i_TEST" in headers, "i_TEST not found in output headers")
def testInternalFields(self):
""" Test that an annotation that is not listed explicitly in the required or optional columns is rendered with i_ prepended """
outputFilename = "out/testInternalFields_v2.4.maf.tsv"
m = MutationData()
m.createAnnotation("TEST", "THIS IS A TEST", "TESTING")
# The next annotation is real and should not be considered internal.
m.createAnnotation("gene", "EGFR")
outputRenderer = TcgaMafOutputRenderer(
outputFilename, configFile='configs/tcgaMAF2.4_output.config')
outputRenderer.renderMutations(iter([m]), ['No comments'])
configFile = ConfigUtils.createConfigParser(
'configs/tcgaMAF2.4_output.config')
requiredColumns = configFile.get("general", "requiredColumns")
self.assertTrue(
"Hugo_Symbol" in requiredColumns,
" This test assumes that Hugo_Symbol is a required column in the TCGA MAF. If not, the test must be modified."
)
statinfo = os.stat(outputFilename)
self.assertTrue(
statinfo.st_size > 0,
"Generated MAF file (" + outputFilename + ") is empty.")
tsvReader = GenericTsvReader(outputFilename)
headers = tsvReader.getFieldNames()
self.assertTrue("Hugo_Symbol" in headers,
"Hugo_Symbol not found in output headers")
self.assertTrue(
"TEST" not in headers,
"TEST was found in output headers when it should have been renamed to i_TEST"
)
self.assertTrue("i_TEST" in headers,
"i_TEST not found in output headers")
def testTCGAMAFAsInputAndQuickAnnotate(self):
""" Test that we can take in a TCGA MAF (using MAFLITE), do annotating, and still render it properly """
inputFilename = "testdata/maf/Patient0.maf.annotated"
tmp = MafliteInputMutationCreator(inputFilename,
'configs/maflite_input.config')
outputFilename = "out/testTCGAMAFAsInputAndQuickAnnotate.tsv"
outputRenderer = TcgaMafOutputRenderer(
outputFilename, 'configs/tcgaMAF2.4_output.config')
annotator = Annotator()
annotator.setInputCreator(tmp)
annotator.setOutputRenderer(outputRenderer)
ds = DatasourceFactory.createDatasource(
"testdata/thaga_janakari_gene_ds/hg19/tj_data.config",
"testdata/thaga_janakari_gene_ds/hg19/")
annotator.addDatasource(ds)
annotator.annotate()
statinfo = os.stat(outputFilename)
self.assertTrue(
statinfo.st_size > 0,
"Generated MAF file (" + outputFilename + ") is empty.")
tsvReaderIn = GenericTsvReader(inputFilename)
tsvReader = GenericTsvReader(outputFilename)
self.assertTrue(tsvReader.getComments().find('#version') != -1,
"First line did not specify a version number")
self.assertTrue("i_TJ_Data_Why" in tsvReader.getFieldNames(),
"New field missing (i_TJ_Data_Why) from header")
self.assertTrue("i_TJ_Data_Who" in tsvReader.getFieldNames(),
"New field missing (i_TJ_Data_Who) from header")
ctrOut = 0
for lineDict in tsvReader:
ctrOut += 1
ctrIn = 0
for lineDict in tsvReaderIn:
ctrIn += 1
ctrIn += len(tsvReaderIn.getCommentsAsList())
ctrOut += len(tsvReader.getCommentsAsList())
self.assertTrue(
ctrOut == (ctrIn + 2),
"Output file should have same number of lines plus two (for maf version and Oncotator version comments) as input file. (In,Out): "
+ str(ctrIn) + ", " + str(ctrOut))
def testTCGAMAFRendering(self):
"""
Tests the ability to render a germline VCF file as a TCGA MAF file.
"""
inputFilename = os.path.join(*["testdata", "vcf", "example.vcf"])
outputFilename = os.path.join("out", "example.vcf.maf.annotated")
creator = VcfInputMutationCreator(inputFilename)
creator.createMutations()
renderer = TcgaMafOutputRenderer(outputFilename)
annotator = Annotator()
annotator.setInputCreator(creator)
annotator.setOutputRenderer(renderer)
annotator.setManualAnnotations(self._createTCGAMAFOverridesForVCF())
datasources = self._createDatasourceCorpus()
for ds in datasources:
annotator.addDatasource(ds)
filename = annotator.annotate()
self._validateTcgaMafContents(filename)