def test_01_gzip(self):
test_id = '01'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
input_file_gz = T_TEST_DIR + "test_" + test_id + ".in.dbed.gz"
# create gzip archive
with open(input_file, 'r') as fh_in:
with gzip.open(input_file_gz, 'wt') as fh_out:
#fh_out.write(str_to_bytearray(fh_in.read()))
d = fh_in.read()
print(type(d))
#d = str_to_bytearray(fh_in.read())
#d = d.encode()
#print(type(d))
#print(d)
d = str(d)
fh_out.write(d)
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.gz.dbed"
cl = DetectOutput(input_file_gz)
cl.classify(output_file, False, Blacklist(), 1, True)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_sj_02(self):
test_id = 'splice_site_motif_02'
input_sam = TEST_DIR + "test_" + test_id + ".in.sam"
input_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
input_file = T_TEST_DIR + "test_" + test_id + ".dbed"
gtf_file = None
fasta_file = TEST_DIR + "test_" + test_id + ".in.fa"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
# sam -> fixed bam
sam_to_fixed_bam(input_sam, input_bam, T_TEST_DIR)
# fixed bam -> dr-disco detect
ic = IntronDecomposition(input_bam)
ic.decompose(0)
fh = open(input_file, "w")
ic.export(fh)
fh.close()
# dr-disco-detect (skip classify) -> dr-disco integrate
cl = DetectOutput(input_file)
cl.integrate(output_file, gtf_file, fasta_file)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_sj_04__CA_GT_d4(self):
test_id = 'splice_site_motif_04'
input_sam = TEST_DIR + "test_" + test_id + ".in.sam"
input_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
input_file = T_TEST_DIR + "test_" + test_id + ".dbed"
gtf_file = None
fasta_file = TEST_DIR + "test_" + test_id + ".in.fa"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
# sam -> fixed bam
sam_to_fixed_bam(input_sam, input_bam, T_TEST_DIR)
# fixed bam -> dr-disco detect
ic = IntronDecomposition(input_bam)
ic.decompose(0)
fh = open(input_file, "w")
ic.export(fh)
fh.close()
# dr-disco-detect (skip classify) -> dr-disco integrate
cl = DetectOutput(input_file)
# originally, this triggered an exception, now we just log an error
# cl.integrate(output_file, gtf_file, fasta_file)
# self.assertRaises(Exception, cl.integrate, output_file, gtf_file, fasta_file)
cl.integrate(output_file, gtf_file, fasta_file)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_02(self):
test_id = 'vcap_err_02'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
detect_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.classified.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
fh = open(detect_file, "w")
ic.export(fh)
fh.close()
cl = DetectOutput(detect_file)
cl.classify(output_file, False, Blacklist(), 1, True)
n_valid = 0
with open(output_file) as fh:
for line in fh:
if line.find('valid') > -1:
n_valid += 1
self.assertEqual(n_valid, 0)
def test_02(self):
test_id = '02'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
cl = DetectOutput(input_file)
cl.classify(output_file, False, Blacklist(), 1, True)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_01__only_valid(self):
test_id = '01'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.only-valid.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.only-valid.dbed"
cl = DetectOutput(input_file)
cl.classify(output_file, True, Blacklist(), 1, True)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_s041_nocrash(self):
test_id = 'terg_s041_b'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
gtf_file = TEST_DIR + "example_refseq.gff"
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
cl = DetectOutput(input_file)
cl.integrate(output_file, gtf_file, None)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_s041_no_gtf(self):
test_id = 'terg_s041_b'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
gtf_file = None
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
cl = DetectOutput(input_file)
cl.integrate(output_file, gtf_file, None)
#self.assertTrue(filecmp.cmp(test_file, output_file), msg="diff '" + test_file + "' '" + output_file + "':\n" + subprocess.Popen(['diff', test_file, output_file], stdout=subprocess.PIPE).stdout.read())
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_s041(self):
test_id = 'terg_s041'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
gtf_files = [
TEST_DIR + "test_" + test_id + ".in.gtf",
TEST_DIR + "test_" + test_id + ".in.no_chr_prefix.gtf"
]
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
for gtf_file in gtf_files:
cl = DetectOutput(input_file)
cl.integrate(output_file, gtf_file, None)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_in_frame_non_hybrid_protein(self):
test_id = 'in_frame_non_hybrid_protein'
# Transcript ID's necessary:
# - TMPRSS2: ENST00000424093
# - ERG: ENST00000398910
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
gtf_files = [TEST_DIR + "test_" + test_id + ".gtf"]
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
for gtf_file in gtf_files:
cl = DetectOutput(input_file)
cl.integrate(output_file, gtf_file, None)
#self.assertTrue(filecmp.cmp(test_file, output_file), msg="diff '" + test_file + "' '" + output_file + "':\n" + subprocess.Popen(['diff', test_file, output_file], stdout=subprocess.PIPE).stdout.read())
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_blacklists(
self): # only test if they don't crash - do not test actual output
test_id = '01'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
blacklists = [Blacklist(), Blacklist(), Blacklist(), Blacklist()]
blacklists[0].add_junctions_from_file(
'share/blacklist-junctions.hg19.txt')
blacklists[1].add_junctions_from_file(
'share/blacklist-junctions.hg38.txt')
blacklists[2].add_regions_from_bed('share/blacklist-regions.hg19.bed')
blacklists[3].add_regions_from_bed('share/blacklist-regions.hg38.bed')
for blacklist in blacklists:
cl = DetectOutput(input_file)
cl.classify(output_file, False, blacklist, 1, True)
def test_01(self):
test_id = '01'
unfixed_sam = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
drdisco_detect = T_TEST_DIR + "test_" + test_id + "_detect.out.txt"
drdisco_detect_test = TEST_DIR + "test_" + test_id + "_detect.out.txt"
drdisco_classify = T_TEST_DIR + "test_" + test_id + "_classify.out.txt"
drdisco_classify_test = TEST_DIR + "test_" + test_id + "_classify.out.txt"
drdisco_integrate = T_TEST_DIR + "test_" + test_id + "_integrate.out.txt"
drdisco_integrate_test = TEST_DIR + "test_" + test_id + "_integrate.out.txt"
# Step 01: dr-disco fix (don't check please)
sam_to_fixed_bam(unfixed_sam, fixed_bam, T_TEST_DIR)
# Step 02: dr-disco detect (check appropriate values and columns)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(drdisco_detect, "w") as fh:
ic.export(fh)
self.assertTrue(filecmp.cmp(drdisco_detect_test, drdisco_detect), msg=get_diff( drdisco_detect_test , drdisco_detect ))
# Step 03: dr-disco classify
cl = DetectOutput(drdisco_detect)
cl.classify(drdisco_classify, False, Blacklist(), 25, True)
self.assertTrue(filecmp.cmp(drdisco_classify_test, drdisco_classify), msg=get_diff( drdisco_classify_test , drdisco_classify ))
# Step 04: dr-disco integrate
cl = DetectOutput(drdisco_classify)
cl.integrate(drdisco_integrate, None, None)
self.assertTrue(filecmp.cmp(drdisco_integrate_test, drdisco_integrate), msg=get_diff( drdisco_integrate_test , drdisco_integrate ))