def test_indexing_to_custom_location_works(self):
'''test indexing a file with a non-default location.'''
index_path = get_temp_filename(suffix='custom.tbi')
pysam.tabix_index(self.tmpfilename, preset="gff",
index=index_path, force=True)
self.assertTrue(checkBinaryEqual(index_path, self.filename_idx))
os.unlink(index_path)
def testSaveStdout(self):
outfile = get_temp_filename(suffix=".tsv")
r = pysam.samtools.flagstat(os.path.join(BAM_DATADIR, "ex1.bam"),
save_stdout=outfile)
self.assertEqual(r, None)
with open(outfile) as inf:
r = inf.read()
self.assertTrue(len(r) > 0)
def testSaveStdout(self):
outfile = get_temp_filename(suffix=".tsv")
r = pysam.samtools.flagstat(
os.path.join(DATADIR, "ex1.bam"),
save_stdout=outfile)
self.assertEqual(r, None)
with open(outfile) as inf:
r = inf.read()
self.assertTrue(len(r) > 0)
def test_indexing_to_custom_location_works(self):
'''test indexing a file with a non-default location.'''
index_path = get_temp_filename(suffix='custom.tbi')
pysam.tabix_index(self.tmpfilename,
preset="gff",
index=index_path,
force=True)
self.assertTrue(checkGZBinaryEqual(index_path, self.filename_idx))
os.unlink(index_path)
def test_read_can_be_written_to_file(self):
tmpfilename = get_temp_filename(".bam")
with pysam.AlignmentFile(tmpfilename, "wb",
reference_names=["chr1", "chr2", "chr3"],
reference_lengths=[1000, 2000, 3000]) as outf:
read = self.build_read()
read.reference_id = 2
outf.write(read)
stdout = pysam.samtools.view(tmpfilename)
chromosome = stdout.split("\t")[2]
self.assertEqual(chromosome, "chr3")
os.unlink(tmpfilename)
def testConstructionFromCopy(self):
fn_in = os.path.join(DATADIR, self.filename)
fn_out = get_temp_filename(suffix=".vcf")
vcf_in = pysam.VariantFile(fn_in)
vcf_out = pysam.VariantFile(fn_out, "w", header=vcf_in.header)
for record in vcf_in:
vcf_out.write(record)
vcf_out.close()
self.complete_check(fn_in, fn_out)
def testConstructionFromCopy(self):
fn_in = os.path.join(DATADIR, self.filename)
fn_out = get_temp_filename(suffix=".vcf")
vcf_in = pysam.VariantFile(fn_in)
vcf_out = pysam.VariantFile(fn_out, "w", header=vcf_in.header)
for record in vcf_in:
vcf_out.write(record)
vcf_out.close()
self.complete_check(fn_in, fn_out)
def test_read_can_be_written_to_file(self):
tmpfilename = get_temp_filename(".bam")
with pysam.AlignmentFile(tmpfilename, "wb",
reference_names=["chr1", "chr2", "chr3"],
reference_lengths=[1000, 2000, 3000]) as outf:
read = self.build_read()
read.reference_id = 2
outf.write(read)
stdout = pysam.samtools.view(tmpfilename)
chromosome = stdout.split("\t")[2]
self.assertEqual(chromosome, "chr3")
os.unlink(tmpfilename)
def test_open_file_with_explicit_abritrarily_named_index_succeeds(self):
tmpfilename = get_temp_filename(self.data_suffix)
shutil.copyfile(self.filename, tmpfilename)
filepath_index = self.filename + ".fai"
filepath_index_compressed = self.filename + ".gzi"
if not os.path.exists(filepath_index_compressed):
filepath_index_compressed = None
with pysam.FastaFile(tmpfilename,
filepath_index=filepath_index,
filepath_index_compressed=filepath_index_compressed) as inf:
self.assertEqual(len(inf), 2)
# index should not be auto-generated
self.assertFalse(os.path.exists(tmpfilename + ".fai"))
os.unlink(tmpfilename)
def test_open_file_with_explicit_abritrarily_named_index_succeeds(self):
tmpfilename = get_temp_filename(self.data_suffix)
shutil.copyfile(self.filename, tmpfilename)
filepath_index = self.filename + ".fai"
filepath_index_compressed = self.filename + ".gzi"
if not os.path.exists(filepath_index_compressed):
filepath_index_compressed = None
with pysam.FastaFile(tmpfilename,
filepath_index=filepath_index,
filepath_index_compressed=filepath_index_compressed) as inf:
self.assertEqual(len(inf), 2)
# index should not be auto-generated
self.assertFalse(os.path.exists(tmpfilename + ".fai"))
os.unlink(tmpfilename)
def check_read_write(self, flag_write, header):
fn = get_temp_filename()
with pysam.AlignmentFile(
fn,
flag_write,
header=header,
reference_filename=os.path.join(BAM_DATADIR, "ex1.fa")) as outf:
a = pysam.AlignedSegment()
a.query_name = "abc"
outf.write(a)
with pysam.AlignmentFile(fn) as inf:
read_header = inf.header
os.unlink(fn)
self.compare_headers(header, read_header)
def check_read_write(self, flag_write, header):
fn = get_temp_filename()
with pysam.AlignmentFile(
fn,
flag_write,
header=header,
reference_filename=os.path.join(BAM_DATADIR, "ex1.fa")) as outf:
a = pysam.AlignedSegment()
a.query_name = "abc"
outf.write(a)
with pysam.AlignmentFile(fn) as inf:
read_header = inf.header
os.unlink(fn)
self.compare_headers(header, read_header)
def testSingleThreadEqualsMultithreadResult(self):
with pysam.VariantFile(self.filename) as inf:
header = inf.header
single = [r for r in inf]
with pysam.VariantFile(self.filename, threads=2) as inf:
multi = [r for r in inf]
for r1, r2 in zip(single, multi):
assert str(r1) == str(r2)
bcf_out = get_temp_filename(suffix=".bcf")
with pysam.VariantFile(bcf_out, mode='wb', header=header,
threads=2) as out:
for r in single:
out.write(r)
with pysam.VariantFile(bcf_out) as inf:
multi_out = [r for r in inf]
for r1, r2 in zip(single, multi_out):
assert str(r1) == str(r2)
def testConstructionWithRecords(self):
fn_in = os.path.join(DATADIR, self.filename)
fn_out = get_temp_filename(suffix=".vcf")
vcf_in = pysam.VariantFile(fn_in)
header = pysam.VariantHeader()
for record in vcf_in.header.records:
header.add_record(record)
fn = str("tmp_VariantFileTest_testConstructionWithRecords") + ".vcf"
vcf_out = pysam.VariantFile(fn, "w", header=header)
for record in vcf_in:
# currently segfaults here:
# vcf_out.write(record)
pass
return
vcf_out.close()
self.complete_check(fn_in, fn_out)
def testConstructionWithRecords(self):
fn_in = os.path.join(DATADIR, self.filename)
fn_out = get_temp_filename(suffix=".vcf")
vcf_in = pysam.VariantFile(fn_in)
header = pysam.VariantHeader()
for record in vcf_in.header.records:
header.add_record(record)
fn = str("tmp_VariantFileTest_testConstructionWithRecords") + ".vcf"
vcf_out = pysam.VariantFile(fn, "w", header=header)
for record in vcf_in:
# currently segfaults here:
# vcf_out.write(record)
pass
return
vcf_out.close()
self.complete_check(fn_in, fn_out)
def testConstructionWithLines(self):
fn_in = os.path.join(CBCF_DATADIR, self.filename)
fn_out = get_temp_filename(suffix=".vcf")
vcf_in = pysam.VariantFile(fn_in)
header = pysam.VariantHeader()
for sample in vcf_in.header.samples:
header.add_sample(sample)
for hr in vcf_in.header.records:
header.add_line(str(hr))
vcf_out = pysam.VariantFile(fn_out, "w", header=header)
for record in vcf_in:
vcf_out.write(record)
vcf_out.close()
vcf_in.close()
self.complete_check(fn_in, fn_out)
def check_read_write(self, flag_write, header):
fn = get_temp_filename()
print(fn)
with pysam.AlignmentFile(fn,
flag_write,
header=header,
reference_filename=os.path.join(
BAM_DATADIR, "ex1.fa")) as outf:
a = pysam.AlignedSegment()
a.query_name = "abc"
outf.write(a)
with pysam.AlignmentFile(fn) as inf:
read_header = inf.header
# os.unlink(fn)
self.compare_headers(header, read_header)
expected_lengths = dict([(x["SN"], x["LN"]) for x in header["SQ"]])
self.assertEqual(
expected_lengths,
dict(zip(read_header.references, read_header.lengths)))
def testConstructionWithLines(self):
fn_in = os.path.join(DATADIR, self.filename)
fn_out = get_temp_filename(suffix=".vcf")
vcf_in = pysam.VariantFile(fn_in)
header = pysam.VariantHeader()
for sample in vcf_in.header.samples:
header.add_sample(sample)
for hr in vcf_in.header.records:
header.add_line(str(hr))
vcf_out = pysam.VariantFile(fn_out, "w", header=header)
for record in vcf_in:
vcf_out.write(record)
vcf_out.close()
vcf_in.close()
self.complete_check(fn_in, fn_out)
def setUp(self):
self.tmpfilename = get_temp_filename(suffix="vcf")
shutil.copyfile(self.filename, self.tmpfilename)
pysam.tabix_index(self.tmpfilename, preset="vcf")
def setUp(self):
self.tmpfilename = get_temp_filename(suffix="vcf")
shutil.copyfile(self.filename, self.tmpfilename)
pysam.tabix_index(self.tmpfilename, preset="vcf")
def setUp(self):
self.tmpfilename = get_temp_filename(suffix="gtf.gz")
shutil.copyfile(self.filename, self.tmpfilename)
def setUp(self):
self.tmpfilename = get_temp_filename(suffix="gtf")
with gzip.open(self.filename, "rb") as infile, \
open(self.tmpfilename, "wb") as outfile:
outfile.write(infile.read())
def setUp(self):
self.tmpfilename = get_temp_filename(suffix="gtf.gz")
shutil.copyfile(self.filename, self.tmpfilename)
def setUp(self):
self.tmpfilename = get_temp_filename(suffix="gtf")
with gzip.open(self.filename, "rb") as infile, \
open(self.tmpfilename, "wb") as outfile:
outfile.write(infile.read())
