def _get_chosen_tx_and_transcript_ds(self, chrom, loc):
config = TestUtils.createUnitTestConfig()
transcript_ds = TestUtils.createTranscriptProviderDatasource(config)
transcript_ds.set_tx_mode(TranscriptProvider.TX_MODE_CANONICAL)
start_txs = transcript_ds.get_transcripts_by_pos(chr=chrom, start=str(loc), end=str(loc))
chosen_tx = transcript_ds._choose_transcript(start_txs, transcript_ds.get_tx_mode(),
VariantClassification.VT_SNP, "", "", str(loc), str(loc))
return chosen_tx, transcript_ds
def _get_chosen_tx_and_transcript_ds(self, chrom, loc):
config = TestUtils.createUnitTestConfig()
transcript_ds = TestUtils.createTranscriptProviderDatasource(config)
transcript_ds.set_tx_mode(TranscriptProvider.TX_MODE_CANONICAL)
start_txs = transcript_ds.get_transcripts_by_pos(chr=chrom, start=str(loc), end=str(loc))
chosen_tx = transcript_ds._choose_transcript(start_txs, transcript_ds.get_tx_mode(),
VariantClassification.VT_SNP, "", "", str(loc), str(loc))
return chosen_tx, transcript_ds
def test_retrieve_transcripts_from_region(self):
"""Test that we can retrieve a large number of transcripts. Requires a full gencode datasource."""
config = TestUtils.createUnitTestConfig()
transcript_ds = TestUtils.createTranscriptProviderDatasource(config)
filtered_txs = transcript_ds.get_transcripts_by_pos(chr="1", start="1", end="100000000")
self.assertTrue(len(filtered_txs) > 4000)
gene_set = set([tx.get_gene() for tx in filtered_txs])
self.assertTrue(len(gene_set) > 1500)
def test_retrieve_transcripts_from_region(self):
"""Test that we can retrieve a large number of transcripts. Requires a full gencode datasource."""
config = TestUtils.createUnitTestConfig()
transcript_ds = TestUtils.createTranscriptProviderDatasource(config)
filtered_txs = transcript_ds.get_transcripts_by_pos(chr="1", start="1", end="100000000")
self.assertTrue(len(filtered_txs) > 4000)
gene_set = set([tx.get_gene() for tx in filtered_txs])
self.assertTrue(len(gene_set) > 1500)
def test_single_sample_onp_combiner(self):
"""test that we can create an onp combined TCGA maf without crashing"""
input_filename = 'testdata/maflite/onp.singlesample.maf.txt'
output_filename = 'out/testSingleSampleOnpCombiner.maf'
config = TestUtils.createUnitTestConfig()
defaultdb = config.get('DEFAULT',"dbDir")
spec = RunSpecificationFactory.create_run_spec("MAFLITE","TCGAMAF", input_filename, output_filename,datasourceDir=defaultdb,
other_opts={OptionConstants.INFER_ONPS: True})
annotator = Annotator()
annotator.initialize(spec)
annotator.annotate()
def test_single_sample_onp_combiner(self):
"""test that we can create an onp combined TCGA maf without crashing"""
input_filename = 'testdata/maflite/onp.singlesample.maf.txt'
output_filename = 'out/testSingleSampleOnpCombiner.maf'
config = TestUtils.createUnitTestConfig()
defaultdb = config.get('DEFAULT',"dbDir")
spec = RunSpecificationFactory.create_run_spec("MAFLITE","TCGAMAF", input_filename, output_filename,
datasource_dir=defaultdb,
other_opts={OptionConstants.INFER_ONPS: True})
annotator = Annotator()
annotator.initialize(spec)
annotator.annotate()
def test_protein_position_off_by_one(self, chrom, start, end, ref, alt, gt_prot_change):
config = TestUtils.createUnitTestConfig()
transcript_ds = TestUtils.createTranscriptProviderDatasource(config)
cc_txs_fp = file("testdata/tx_exact_uniprot_matches.txt", 'r')
cc_txs = [tx.rsplit(".", 1)[0] for tx in cc_txs_fp]
cc_txs.append("ENST00000338368") # Add a transcript that is not exactly the same, but close
cc_txs_fp.close()
transcript_ds.set_custom_canonical_txs(cc_txs)
m = MutationData()
m.chr = chrom
m.start = start
m.end = end
m.ref_allele = ref
m.alt_allele = alt
m2 = transcript_ds.annotate_mutation(m)
self.assertEqual(m2['protein_change'], gt_prot_change)
def test_protein_position_off_by_one(self, chrom, start, end, ref, alt, gt_prot_change):
config = TestUtils.createUnitTestConfig()
transcript_ds = TestUtils.createTranscriptProviderDatasource(config)
cc_txs_fp = file("testdata/tx_exact_uniprot_matches.txt", 'r')
cc_txs = [tx.rsplit(".", 1)[0] for tx in cc_txs_fp]
cc_txs.append("ENST00000338368") # Add a transcript that is not exactly the same, but close
cc_txs_fp.close()
transcript_ds.set_custom_canonical_txs(cc_txs)
m = MutationDataFactory.default_create()
m.chr = chrom
m.start = start
m.end = end
m.ref_allele = ref
m.alt_allele = alt
m2 = transcript_ds.annotate_mutation(m)
self.assertEqual(m2['protein_change'], gt_prot_change)
def test_onp_combiner_snp_then_multiallelic(self):
"""test that we can handle reading a SNP then multiallelic from a VCF without crashing"""
input_filename = 'testdata/vcf/infer_onp_fail_snp_then_multiallelic.vcf'
output_filename = 'out/testSNPThenMultiallelic.maf.annotated'
config = TestUtils.createUnitTestConfig()
default_db = config.get('DEFAULT', "dbDir")
spec = RunSpecificationFactory.create_run_spec(
"VCF",
"TCGAMAF",
input_filename,
output_filename,
datasource_dir=default_db,
is_skip_no_alts=True,
other_opts={
OptionConstants.INFER_ONPS: True,
OptionConstants.COLLAPSE_NUMBER_ANNOTATIONS: True
})
annotator = Annotator()
annotator.initialize(spec)
annotator.annotate()
def setUp(self):
self.config = TestUtils.createUnitTestConfig()
pass
def setUp(self):
self.config = TestUtils.createUnitTestConfig()