def test_run_with_filter_cluster_include_ref_alleles(self):
'''test run with filtering and clustering'''
vcf_file_in = os.path.join(data_dir, 'run.calls.include_ref_calls.vcf')
vcf_reference_file = os.path.join(data_dir, 'run.ref.fa')
verify_reference_file = os.path.join(data_dir, 'run.ref.mutated.fa')
tmp_out = 'tmp.mapping_based_verifier.out.include_ref_calls'
verifier = mapping_based_verifier.MappingBasedVerifier(
vcf_file_in,
vcf_reference_file,
verify_reference_file,
tmp_out,
flank_length=31,
filter_and_cluster_vcf=True,
discard_ref_calls=False)
verifier.run()
expected_out = os.path.join(data_dir, 'run.out.include_ref_calls')
for suffix in [
'.false_negatives.vcf', '.stats.tsv', '.vcf',
'.gt_conf_hist.TP.tsv', '.gt_conf_hist.FP.tsv'
]:
expected_file = expected_out + suffix
got_file = tmp_out + suffix
self.assertTrue(filecmp.cmp(expected_file, got_file,
shallow=False))
os.unlink(got_file)
samfile = tmp_out + '.sam'
self.assertTrue(os.path.exists(samfile))
os.unlink(samfile)
for suffix in [
'.dnadiff.merged.vcf', '.dnadiff.qdiff', '.dnadiff.raw.vcf',
'.dnadiff.snps', '.filter.vcf', '.filter.cluster.vcf'
]:
os.unlink(tmp_out + suffix)
def run(options):
verifier = mapping_based_verifier.MappingBasedVerifier(
options.vcf_file,
options.vcf_ref,
options.truth_ref,
options.outprefix,
flank_length=options.flank_length,
expected_variants_vcf=options.expected_variants_vcf,
run_dnadiff=options.run_dnadiff,
filter_and_cluster_vcf=not options.no_filter_cluster,
discard_ref_calls=not options.include_ref_calls,
allow_flank_mismatches=not options.no_flank_mismatches,
merge_length=options.variant_merge_length,
exclude_regions_bed_file=options.exclude_bed,
max_soft_clipped=options.max_soft_clipped,
)
verifier.run()
def test_run_no_filter_cluster_with_exclude_region(self):
"""test run without filtering and clustering but with region excluded"""
vcf_file_in = os.path.join(data_dir, "run.calls.vcf")
vcf_reference_file = os.path.join(data_dir, "run.ref.fa")
verify_reference_file = os.path.join(data_dir, "run.ref.mutated.fa")
exclude_regions_bed_file = os.path.join(data_dir, "run.exclude.bed")
tmp_out = "tmp.mapping_based_verifier.out.no_filter_cluster.with_exclude"
verifier = mapping_based_verifier.MappingBasedVerifier(
vcf_file_in,
vcf_reference_file,
verify_reference_file,
tmp_out,
flank_length=31,
filter_and_cluster_vcf=False,
exclude_regions_bed_file=exclude_regions_bed_file,
)
verifier.run()
expected_out = os.path.join(data_dir,
"run.out.no_filter_cluster_with_exclude")
for suffix in [
".false_negatives.vcf",
".stats.tsv",
".vcf",
".gt_conf_hist.TP.tsv",
".gt_conf_hist.FP.tsv",
]:
expected_file = expected_out + suffix
got_file = tmp_out + suffix
self.assertTrue(filecmp.cmp(expected_file, got_file,
shallow=False))
os.unlink(got_file)
samfile = tmp_out + ".sam"
self.assertTrue(os.path.exists(samfile))
os.unlink(samfile)
for suffix in [
".dnadiff.merged.vcf",
".dnadiff.qdiff",
".dnadiff.raw.vcf",
".dnadiff.snps",
]:
os.unlink(tmp_out + suffix)