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_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_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_07(self):
test_id = '07'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
def test_detect_02(self):
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = '02'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
command = ["bin/dr-disco", "detect", "-m", "0", fixed_bam, output_file]
self.assertEqual(subprocess.call(command), 0)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_03(self):
test_id = '03'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_01(self):
test_id = 'artifact_reads_01'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
fh = open(output_file, "w")
ic.export(fh)
fh.close()
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_32_pruning_hotfix(self):
test_id = '32'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
# Test data not checked, should just not throw an exception
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_01(self):
test_id = '01'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
fh = open(output_file, "w")
ic.export(fh)
fh.close()
#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_21_tests_extracting_subnetworks_in_ideal_optimization_usecase(
self):
test_id = '21'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
# Test data not checked, should just not throw an exception
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_08(self):
# True TMPRSS2-ERG inv/del; requires to pass, which is possible if stranding is taken into account correctly
test_id = 'artifact_reads_08'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
fh = open(output_file, "w")
ic.export(fh)
fh.close()
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_29_mismatch_ratio_and_stddev_b11383(self):
# sum matches needs to be: 5675 (2919 + 2756 = 5675)
# sum mismatches needs to be: 149 (73 + 76 = 149)
test_id = '29'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
# Test data not checked, should just not throw an exception
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
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 ))
def test_18_s27(self):
test_id = '18'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
# Test data not checked, should just not throw an exception
#self.assertTrue(filecmp.cmp(test_file, output_file), msg=get_diff(test_file, output_file))
# order is different, but content matching is still valid:
self.assertTrue(
sorted(open(test_file, 'r').read().split("\n")) == sorted(
open(output_file, 'r').read().split("\n")))
def test_31_many_muts_on_test_12_for_testing_manymuts_merging_in_subgraphs(
self):
"""
mm needs to be 4*1=4
m needs to be 4*50=200
"""
test_id = '31'
input_file_a = TEST_DIR + "test_" + test_id + ".sam"
fixed_bam = T_TEST_DIR + "test_" + test_id + ".fixed.bam"
test_file = TEST_DIR + "test_" + test_id + ".out.dbed"
output_file = T_TEST_DIR + "test_" + test_id + ".out.dbed"
sam_to_fixed_bam(input_file_a, fixed_bam, T_TEST_DIR)
ic = IntronDecomposition(fixed_bam)
ic.decompose(0)
with open(output_file, "w") as fh:
ic.export(fh)
# Test data not checked, should just not throw an exception
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
