def test_no_overwriting_muts(self):
"""Ensure that (given configuration that disallows) we cannot annotate from a datasource when a value was specified in the input."""
# We will have an input with a "Who" annotation that this datasource will try to write.
gene_ds = DatasourceFactory.createDatasource(
"testdata/thaga_janakari_gene_ds/hg19/tj_data.config",
"testdata/thaga_janakari_gene_ds/hg19/")
input_filename = "testdata/maflite/who_alt1_vs_alt2.maflite"
output_filename = "out/who_alt1_vs_alt2.maf.annotated"
input_format = "MAFLITE"
output_format = "TCGAMAF"
other_opts = {
OptionConstants.ALLOW_ANNOTATION_OVERWRITING: False,
OptionConstants.NO_PREPEND: True
}
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(
input_format,
output_format,
input_filename,
output_filename,
datasource_list=[gene_ds],
other_opts=other_opts)
annotator = Annotator()
annotator.initialize(run_spec)
self.assertRaises(DuplicateAnnotationException, annotator.annotate)
def test_overwriting_muts(self):
"""Ensure that (given correct configuration) we can annotate from a datasource, even if the datasource will overwrite an existing mutation."""
# We will have an input with a "Who" annotation that this datasource will try to write.
gene_ds = DatasourceFactory.createDatasource(
"testdata/thaga_janakari_gene_ds/hg19/tj_data.config",
"testdata/thaga_janakari_gene_ds/hg19/")
input_filename = "testdata/maflite/who_alt1_vs_alt2.maflite"
output_filename = "out/who_alt1_vs_alt2.maf.annotated"
input_format = "MAFLITE"
output_format = "TCGAMAF"
other_opts = {
OptionConstants.ALLOW_ANNOTATION_OVERWRITING: True,
OptionConstants.NO_PREPEND: True
}
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(
input_format,
output_format,
input_filename,
output_filename,
datasource_list=[gene_ds],
other_opts=other_opts)
annotator = Annotator()
annotator.initialize(run_spec)
annotator.annotate()
tsv_reader = GenericTsvReader(output_filename)
for i, line_dict in enumerate(tsv_reader):
self.assertTrue(line_dict.get('TJ_Data_Who', "") != "Tromokratis")
def test_no_overwriting_muts(self):
"""Ensure that (given configuration that disallows) we cannot annotate from a datasource when a value was specified in the input."""
# We will have an input with a "Who" annotation that this datasource will try to write.
gene_ds = DatasourceFactory.createDatasource(
"testdata/thaga_janakari_gene_ds/hg19/tj_data.config", "testdata/thaga_janakari_gene_ds/hg19/"
)
input_filename = "testdata/maflite/who_alt1_vs_alt2.maflite"
output_filename = "out/who_alt1_vs_alt2.maf.annotated"
input_format = "MAFLITE"
output_format = "TCGAMAF"
other_opts = {OptionConstants.ALLOW_ANNOTATION_OVERWRITING: False, OptionConstants.NO_PREPEND: True}
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(
input_format,
output_format,
input_filename,
output_filename,
datasource_list=[gene_ds],
other_opts=other_opts,
)
annotator = Annotator()
annotator.initialize(run_spec)
self.assertRaises(DuplicateAnnotationException, annotator.annotate)
def test_overwriting_muts(self):
"""Ensure that (given correct configuration) we can annotate from a datasource, even if the datasource will overwrite an existing mutation."""
# We will have an input with a "Who" annotation that this datasource will try to write.
gene_ds = DatasourceFactory.createDatasource(
"testdata/thaga_janakari_gene_ds/hg19/tj_data.config", "testdata/thaga_janakari_gene_ds/hg19/"
)
input_filename = "testdata/maflite/who_alt1_vs_alt2.maflite"
output_filename = "out/who_alt1_vs_alt2.maf.annotated"
input_format = "MAFLITE"
output_format = "TCGAMAF"
other_opts = {OptionConstants.ALLOW_ANNOTATION_OVERWRITING: True, OptionConstants.NO_PREPEND: True}
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(
input_format,
output_format,
input_filename,
output_filename,
datasource_list=[gene_ds],
other_opts=other_opts,
)
annotator = Annotator()
annotator.initialize(run_spec)
annotator.annotate()
tsv_reader = GenericTsvReader(output_filename)
for i, line_dict in enumerate(tsv_reader):
self.assertTrue(line_dict.get("TJ_Data_Who", "") != "Tromokratis")
def test_run_spec_creation_no_datasources(self):
"""Test that we can create a run spec with no datasources"""
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(input_format="VCF",
input_filename="testdata/m2_support/phasingExample.vcf",
output_format="TCGAMAF",
output_filename="out/foo.maf.annotated",
datasource_list=[])
self.assertTrue(isinstance(run_spec.inputCreator, InputMutationCreator))
self.assertTrue(isinstance(run_spec.outputRenderer, OutputRenderer))
self.assertTrue(run_spec.is_allow_annotation_overwriting==False)
def test_run_spec_creation_no_datasources(self):
"""Test that we can create a run spec with no datasources"""
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(
input_format="VCF",
input_filename="testdata/m2_support/phasingExample.vcf",
output_format="TCGAMAF",
output_filename="out/foo.maf.annotated",
datasource_list=[])
self.assertTrue(isinstance(run_spec.inputCreator,
InputMutationCreator))
self.assertTrue(isinstance(run_spec.outputRenderer, OutputRenderer))
self.assertTrue(run_spec.is_allow_annotation_overwriting == False)
def test_tcgamaf_invalid_input_file(self):
"""Test a case where TCGAMAF specified as input and we get an error (as we should) for a missing file"""
is_exception_seen = False
try:
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(input_format="TCGAMAF",
input_filename="testdata/Idonotexist",
output_format="TCGAMAF",
output_filename="out/foo.maf.annotated",
datasource_list=[])
except IOError as ie:
is_exception_seen = True
self.assertTrue(is_exception_seen)
def test_tcgamaf_invalid_input_file(self):
"""Test a case where TCGAMAF specified as input and we get an error (as we should) for a missing file"""
is_exception_seen = False
try:
run_spec = RunSpecificationFactory.create_run_spec_given_datasources(
input_format="TCGAMAF",
input_filename="testdata/Idonotexist",
output_format="TCGAMAF",
output_filename="out/foo.maf.annotated",
datasource_list=[])
except IOError as ie:
is_exception_seen = True
self.assertTrue(is_exception_seen)
def test_reannotating_actual_file(self):
"""Test that we can take in a file, annotate, similar to M2 process (VCF to TCGA MAF no ONPs, then TCGA MAF to TCGA MAF with ONPs) and collapse values."""
# This test assumes that the numeric values are not being collapsed.
input_filename = "testdata/m2_support/phasingExample.vcf"
midpoint_output_filename = "out/m2_support/reannotating_tcga_maf_midpoint.maf.annotated"
output_filename = "out/m2_support/reannotating_tcga_maf.maf.annotated"
options_step1 = {OptionConstants.COLLAPSE_FILTER_COLS: True, OptionConstants.NO_PREPEND: False,
OptionConstants.SPLIT_ALLELIC_DEPTH: True, OptionConstants.INFER_ONPS: False}
# Note that this will also test collapsing numeric values.
options_step2 = {OptionConstants.REANNOTATE_TCGA_MAF_COLS: True, OptionConstants.INFER_ONPS: True,
OptionConstants.ALLOW_ANNOTATION_OVERWRITING: True, OptionConstants.NO_PREPEND: False,
OptionConstants.COLLAPSE_NUMBER_ANNOTATIONS: True}
run_spec_step1 = RunSpecificationFactory.create_run_spec("VCF", "TCGAMAF", input_filename, midpoint_output_filename,
is_skip_no_alts=True, other_opts=options_step1,
datasource_dir=self._determine_db_dir())
annotator = Annotator()
annotator.initialize(run_spec_step1)
annotator.annotate()
# To speed up this test, use the same datasources from step 1
ds_list = run_spec_step1.get_datasources()
tsv_reader = GenericTsvReader(midpoint_output_filename)
i = -1
for i, line in enumerate(tsv_reader):
self.assertTrue(line["i_QSS"].find("|") == -1, "i_QSS annotation should not have a '|' in it in mutation: " + str(i+1))
self.assertTrue(i == 2, 'Mutation count flawed... should have been three mutations: ' + str(i+1))
run_spec_step2 = RunSpecificationFactory.create_run_spec_given_datasources("TCGAMAF", "TCGAMAF", midpoint_output_filename, output_filename,
other_opts=options_step2, datasource_list=ds_list)
annotator.initialize(run_spec_step2)
annotator.annotate()
gt_alt_count = [80, 7]
gt_alt_count_full = ["82|80", "7"]
gt_ref_count = [68, 151]
# Please note that this is not "68|68" since these were collapsed by ONP combiner.
gt_ref_count_full = ["68", "151"]
gt_tumor_f = [.5375, .046]
gt_tumor_f_full = ["0.538|0.537", "0.046"]
tsv_reader = GenericTsvReader(output_filename)
i = -1
for i, line in enumerate(tsv_reader):
is_good_prefix = [not ks.startswith('i_i_') for ks in line.keys()]
self.assertTrue(all(is_good_prefix), "i_i_ prefix found.")
if i == 0:
self.assertTrue(line["i_QSS"].find("|") != -1, "i_QSS tag should have a '|' in it for the first mutation")
self.assertEqual(int(line['t_alt_count']), gt_alt_count[i])
self.assertEqual(int(line['t_ref_count']), gt_ref_count[i])
self.assertEqual(float(line['i_tumor_f']), gt_tumor_f[i])
self.assertEqual(line['i_t_alt_count_full'], gt_alt_count_full[i])
self.assertEqual(line['i_t_ref_count_full'], gt_ref_count_full[i])
self.assertEqual(line['i_tumor_f_full'], gt_tumor_f_full[i])
self.assertTrue(i == 1, 'Mutation count flawed... should have been two mutations: ' + str(i+1))
def test_reannotating_actual_file(self):
"""Test that we can take in a file, annotate, similar to M2 process (VCF to TCGA MAF no ONPs, then TCGA MAF to TCGA MAF with ONPs) and collapse values."""
# This test assumes that the numeric values are not being collapsed.
input_filename = "testdata/m2_support/phasingExample.vcf"
midpoint_output_filename = "out/m2_support/reannotating_tcga_maf_midpoint.maf.annotated"
output_filename = "out/m2_support/reannotating_tcga_maf.maf.annotated"
options_step1 = {
OptionConstants.COLLAPSE_FILTER_COLS: True,
OptionConstants.NO_PREPEND: False,
OptionConstants.SPLIT_ALLELIC_DEPTH: True,
OptionConstants.INFER_ONPS: False
}
# Note that this will also test collapsing numeric values.
options_step2 = {
OptionConstants.REANNOTATE_TCGA_MAF_COLS: True,
OptionConstants.INFER_ONPS: True,
OptionConstants.ALLOW_ANNOTATION_OVERWRITING: True,
OptionConstants.NO_PREPEND: False,
OptionConstants.COLLAPSE_NUMBER_ANNOTATIONS: True
}
run_spec_step1 = RunSpecificationFactory.create_run_spec(
"VCF",
"TCGAMAF",
input_filename,
midpoint_output_filename,
is_skip_no_alts=True,
other_opts=options_step1,
datasource_dir=self._determine_db_dir())
annotator = Annotator()
annotator.initialize(run_spec_step1)
annotator.annotate()
# To speed up this test, use the same datasources from step 1
ds_list = run_spec_step1.get_datasources()
tsv_reader = GenericTsvReader(midpoint_output_filename)
i = -1
for i, line in enumerate(tsv_reader):
self.assertTrue(
line["i_QSS"].find("|") == -1,
"i_QSS annotation should not have a '|' in it in mutation: " +
str(i + 1))
self.assertTrue(
i == 2,
'Mutation count flawed... should have been three mutations: ' +
str(i + 1))
run_spec_step2 = RunSpecificationFactory.create_run_spec_given_datasources(
"TCGAMAF",
"TCGAMAF",
midpoint_output_filename,
output_filename,
other_opts=options_step2,
datasource_list=ds_list)
annotator.initialize(run_spec_step2)
annotator.annotate()
gt_alt_count = [80, 7]
gt_alt_count_full = ["82|80", "7"]
gt_ref_count = [68, 151]
# Please note that this is not "68|68" since these were collapsed by ONP combiner.
gt_ref_count_full = ["68", "151"]
gt_tumor_f = [.5375, .046]
gt_tumor_f_full = ["0.538|0.537", "0.046"]
tsv_reader = GenericTsvReader(output_filename)
i = -1
for i, line in enumerate(tsv_reader):
is_good_prefix = [not ks.startswith('i_i_') for ks in line.keys()]
self.assertTrue(all(is_good_prefix), "i_i_ prefix found.")
if i == 0:
self.assertTrue(
line["i_QSS"].find("|") != -1,
"i_QSS tag should have a '|' in it for the first mutation")
self.assertEqual(int(line['t_alt_count']), gt_alt_count[i])
self.assertEqual(int(line['t_ref_count']), gt_ref_count[i])
self.assertEqual(float(line['i_tumor_f']), gt_tumor_f[i])
self.assertEqual(line['i_t_alt_count_full'], gt_alt_count_full[i])
self.assertEqual(line['i_t_ref_count_full'], gt_ref_count_full[i])
self.assertEqual(line['i_tumor_f_full'], gt_tumor_f_full[i])
self.assertTrue(
i == 1,
'Mutation count flawed... should have been two mutations: ' +
str(i + 1))
