def test_run_empty_vcf_input_files(self):
'''test run when input files have no variants'''
outdir = 'tmp.adjudicator.out'
if os.path.exists(outdir):
shutil.rmtree(outdir)
ref_fasta = os.path.join(data_dir, 'run.ref.fa')
reads_file = os.path.join(data_dir, 'run.bwa.bam')
vcf_files = [
os.path.join(data_dir, x)
for x in ['run.calls.empty.1.vcf', 'run.calls.empty.2.vcf']
]
adj = adjudicator.Adjudicator(outdir,
ref_fasta, [reads_file],
vcf_files,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000)
with self.assertRaises(adjudicator.Error):
adj.run()
self.assertTrue(os.path.exists(outdir))
self.assertTrue(os.path.exists(adj.log_file))
self.assertFalse(os.path.exists(adj.final_vcf))
self.assertFalse(os.path.exists(adj.gramtools_build_dir))
self.assertFalse(os.path.exists(adj.gramtools_quasimap_dir))
self.assertFalse(os.path.exists(adj.plots_prefix + '.data.tsv'))
self.assertFalse(os.path.exists(adj.plots_prefix + '.dp_hist.pdf'))
self.assertFalse(
os.path.exists(adj.plots_prefix + '.gt_conf_dp_scatter.pdf'))
self.assertFalse(os.path.exists(adj.plots_prefix +
'.gt_conf_hist.pdf'))
self.assertTrue(os.path.exists(adj.clustered_vcf))
shutil.rmtree(outdir)
def test_0MeanDepth_stillRuns(self):
"""
When mean depth is 0, we can get math errors: math.log(0) in genotype likelihood computation,
and division by 0 in genotype confidence simulation.
Note the former only actually occurs if there is a variant site with non-zero coverage;
in this case, mean depth can get set to 0 due to rounding imprecision. This is tested in genotyper unit tests.
"""
outdir = "tmp.adjudicator.out"
if os.path.exists(outdir):
shutil.rmtree(outdir)
ref_fasta = os.path.join(data_dir, "run.ref.fa")
reads_file = os.path.join(data_dir, "no_map_reads.fastq")
vcf_files = [os.path.join(data_dir, "run.calls.1.vcf")]
adj = adjudicator.Adjudicator(
outdir,
ref_fasta,
[reads_file],
vcf_files,
clean=False,
gramtools_kmer_size=5,
)
adj.run()
# Make sure the coverage is 0
self.assertEqual(adj.mean_depths[0], 0)
def test_run_clean_is_true(self):
"""test run when we do clean files afterwards"""
# We're just testing that it doesn't crash.
# Check the output files exist, but not their contents.
# First run using splitting of VCF file.
# Then run without splitting.
outdir = "tmp.adjudicator.clean.out"
if os.path.exists(outdir):
shutil.rmtree(outdir)
ref_fasta = os.path.join(data_dir, "run.ref.fa")
reads_file = os.path.join(data_dir, "run.bwa.bam")
vcf_files = [
os.path.join(data_dir, x) for x in ["run.calls.1.vcf", "run.calls.2.vcf"]
]
adj = adjudicator.Adjudicator(
outdir,
ref_fasta,
[reads_file],
vcf_files,
clean=True,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000,
)
adj.run()
self.assertTrue(os.path.exists(outdir))
self.assertTrue(os.path.exists(os.path.join(outdir, "final.vcf")))
shutil.rmtree(outdir)
def test_run_empty_vcf_input_files():
"""test run when input files have no variants"""
outdir = "tmp.adjudicator.out"
if os.path.exists(outdir):
shutil.rmtree(outdir)
ref_fasta = os.path.join(data_dir, "run.ref.fa")
reads_file = os.path.join(data_dir, "run.bwa.bam")
vcf_files = [
os.path.join(data_dir, x)
for x in ["run.calls.empty.1.vcf", "run.calls.empty.2.vcf"]
]
adj = adjudicator.Adjudicator(
outdir,
ref_fasta,
[reads_file],
vcf_files,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000,
)
with pytest.raises(Exception):
adj.run()
assert os.path.exists(outdir)
assert os.path.exists(adj.log_file)
assert not os.path.exists(adj.final_vcf)
assert not os.path.exists(adj.gramtools_build_dir)
assert not os.path.exists(adj.gramtools_quasimap_dir)
assert os.path.exists(adj.clustered_vcf)
shutil.rmtree(outdir)
def run(options):
adj = adjudicator.Adjudicator(
options.outdir,
options.ref_fasta,
options.reads,
options.vcf_files,
max_read_length=options.max_read_length,
read_error_rate=options.read_error_rate,
overwrite_outdir=options.force,
max_alleles_per_cluster=options.max_alleles_per_cluster,
gramtools_build_dir=options.gramtools_build_dir,
sample_name=options.sample_name,
variants_per_split=options.variants_per_split,
alleles_per_split=options.alleles_per_split,
total_splits=options.total_splits,
clean=not options.debug,
gramtools_kmer_size=options.gramtools_kmer_size,
use_unmapped_reads=options.use_unmapped_reads,
)
adj.run()
def run(options):
adj = adjudicator.Adjudicator(
options.outdir,
options.ref_fasta,
options.reads,
options.vcf_files,
read_error_rate=options.read_error_rate,
overwrite_outdir=options.force,
max_alleles_per_cluster=options.max_alleles_per_cluster,
gramtools_build_dir=options.gramtools_build_dir,
sample_name=options.sample_name,
variants_per_split=options.variants_per_split,
alleles_per_split=options.alleles_per_split,
total_splits=options.total_splits,
clean=not options.debug,
gramtools_kmer_size=options.gramtools_kmer_size,
use_unmapped_reads=options.use_unmapped_reads,
filter_min_dp=options.filter_min_dp,
filter_min_gcp=options.filter_min_gcp,
filter_min_frs=options.filter_min_frs,
call_hets=options.include_het_calls,
debug=options.debug,
)
adj.run()
def test_run(self):
'''test run'''
# We're just testing that it doesn't crash.
# Check the output files exist, but not their contents.
# First run using splitting of VCF file.
# Then run without splitting.
outdir = 'tmp.adjudicator.out'
if os.path.exists(outdir):
shutil.rmtree(outdir)
ref_fasta = os.path.join(data_dir, 'run.ref.fa')
reads_file = os.path.join(data_dir, 'run.bwa.bam')
vcf_files = [
os.path.join(data_dir, x)
for x in ['run.calls.1.vcf', 'run.calls.2.vcf']
]
adj = adjudicator.Adjudicator(outdir,
ref_fasta, [reads_file],
vcf_files,
variants_per_split=3,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000)
adj.run()
self.assertTrue(os.path.exists(outdir))
self.assertTrue(os.path.exists(adj.log_file))
self.assertTrue(os.path.exists(adj.final_vcf))
self.assertTrue(os.path.exists(adj.clustered_vcf))
# Clean up and then run without splitting
shutil.rmtree(outdir)
adj = adjudicator.Adjudicator(outdir,
ref_fasta, [reads_file],
vcf_files,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000)
adj.run()
self.assertTrue(os.path.exists(outdir))
self.assertTrue(os.path.exists(adj.log_file))
self.assertTrue(os.path.exists(adj.final_vcf))
self.assertTrue(os.path.exists(adj.gramtools_build_dir))
self.assertTrue(os.path.exists(adj.gramtools_quasimap_dir))
self.assertTrue(os.path.exists(adj.clustered_vcf))
self.assertTrue(os.path.exists(adj.plots_prefix + '.data.tsv'))
self.assertTrue(os.path.exists(adj.plots_prefix + '.dp_hist.pdf'))
self.assertTrue(
os.path.exists(adj.plots_prefix + '.gt_conf_dp_scatter.pdf'))
self.assertTrue(os.path.exists(adj.plots_prefix + '.gt_conf_hist.pdf'))
# Now we've run the adjudicator, we have a gramtools
# build directory. Rerun, but this time use the build
# directory, so we test the gramtools_build_dir option
outdir2 = 'tmp.adjudicator.out.2'
gramtools_build_dir = adj.gramtools_build_dir
if os.path.exists(outdir2):
shutil.rmtree(outdir2)
ref_fasta = os.path.join(data_dir, 'run.ref.fa')
reads_file = os.path.join(data_dir, 'run.bwa.bam')
# When gramtools build dir supplied, the Adjudicator assumes
# one clsutered VCF file that matches the gramtools build run.
# This is the clustered VCF made by the Adjudicator, so we
# use that instead of the list of original VCF files
vcf_files = [adj.clustered_vcf]
adj = adjudicator.Adjudicator(outdir2,
ref_fasta, [reads_file],
vcf_files,
gramtools_build_dir=gramtools_build_dir,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000)
adj.run()
self.assertTrue(os.path.exists(outdir2))
self.assertTrue(os.path.exists(adj.log_file))
self.assertTrue(os.path.exists(adj.final_vcf))
self.assertTrue(os.path.exists(adj.gramtools_build_dir))
self.assertTrue(os.path.exists(adj.gramtools_quasimap_dir))
self.assertTrue(os.path.exists(adj.clustered_vcf))
self.assertTrue(os.path.exists(adj.plots_prefix + '.data.tsv'))
self.assertTrue(os.path.exists(adj.plots_prefix + '.dp_hist.pdf'))
self.assertTrue(
os.path.exists(adj.plots_prefix + '.gt_conf_dp_scatter.pdf'))
self.assertTrue(os.path.exists(adj.plots_prefix + '.gt_conf_hist.pdf'))
self.assertFalse(
os.path.exists(os.path.join(outdir2, 'gramtools.build')))
shutil.rmtree(outdir)
shutil.rmtree(outdir2)
def test_run_clean_is_false():
"""test run when not cleaning up files afterwards"""
# We're just testing that it doesn't crash.
# Check the output files exist, but not their contents.
# First run using splitting of VCF file.
# Then run without splitting.
outdir = "tmp.adjudicator.noclean.out"
if os.path.exists(outdir):
shutil.rmtree(outdir)
ref_fasta = os.path.join(data_dir, "run.ref.fa")
reads_file = os.path.join(data_dir, "run.bwa.bam")
vcf_files = [
os.path.join(data_dir, x)
for x in ["run.calls.1.vcf", "run.calls.2.vcf"]
]
adj = adjudicator.Adjudicator(
outdir,
ref_fasta,
[reads_file],
vcf_files,
variants_per_split=3,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000,
)
adj.run()
assert os.path.exists(outdir)
assert os.path.exists(adj.log_file)
assert os.path.exists(adj.final_vcf)
assert os.path.exists(adj.clustered_vcf)
# Clean up and then run without splitting
shutil.rmtree(outdir)
adj = adjudicator.Adjudicator(
outdir,
ref_fasta,
[reads_file],
vcf_files,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000,
)
adj.run()
assert os.path.exists(outdir)
assert os.path.exists(adj.log_file)
assert os.path.exists(adj.final_vcf)
assert os.path.exists(adj.gramtools_build_dir)
assert os.path.exists(adj.gramtools_quasimap_dir)
assert os.path.exists(adj.clustered_vcf)
# Now we've run the adjudicator, we have a gramtools
# build directory. Rerun, but this time use the build
# directory, so we test the gramtools_build_dir option
outdir2 = "tmp.adjudicator.out.2"
gramtools_build_dir = adj.gramtools_build_dir
if os.path.exists(outdir2):
shutil.rmtree(outdir2)
ref_fasta = os.path.join(data_dir, "run.ref.fa")
reads_file = os.path.join(data_dir, "run.bwa.bam")
# When gramtools build dir supplied, the Adjudicator assumes
# one clsutered VCF file that matches the gramtools build run.
# This is the clustered VCF made by the Adjudicator, so we
# use that instead of the list of original VCF files
vcf_files = [adj.clustered_vcf]
adj = adjudicator.Adjudicator(
outdir2,
ref_fasta,
[reads_file],
vcf_files,
gramtools_build_dir=gramtools_build_dir,
clean=False,
gramtools_kmer_size=5,
genotype_simulation_iterations=1000,
)
adj.run()
assert os.path.exists(outdir2)
assert os.path.exists(adj.log_file)
assert os.path.exists(adj.final_vcf)
assert os.path.exists(adj.gramtools_build_dir)
assert os.path.exists(adj.gramtools_quasimap_dir)
assert os.path.exists(adj.clustered_vcf)
assert not os.path.exists(os.path.join(outdir2, "gramtools.build"))
shutil.rmtree(outdir)
shutil.rmtree(outdir2)