def test_bam_extract_01(self):
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
input_file = TEST_DIR + "test_terg_02.bam"
output_file = T_TEST_DIR + "test_terg_02.filtered.bam"
output_file_s = T_TEST_DIR + "test_terg_02.filtered.sam"
test_file = TEST_DIR + "test_terg_02.filtered.sam"
# c = BAMExtract(input_file)
# c.extract("chr21:39000000-40000000", "chr5:1-2", output_file)
command = [
"bin/dr-disco", "bam-extract", "chr21:39000000-40000000",
"chr5:1-2", input_file, output_file
]
self.assertEqual(subprocess.call(command), 0)
# Bam2Sam
fhq = open(output_file_s, "w")
fhq.write(pysam.view(output_file))
fhq.close()
#if not filecmp.cmp(output_file_s, test_file):
# print 'diff \'' + output_file_s + '\' \'' + test_file + '\''
#self.assertTrue(filecmp.cmp(test_file, output_file_s))
self.assertTrue(filecmp.cmp(test_file, output_file_s),
msg=get_diff(test_file, output_file_s))
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_03(self): # +1, +2 -> 0
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = 'frameshift-prediction_03'
# fusions = [(['chr1', 1040604, '+'], ['chr1', 999020, '-']), (['1', 1040604, '+'], ['1', 999020, '-'])]
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
gtf_files = [
TEST_DIR + 'frameshift_example.gtf',
TEST_DIR + 'frameshift_example.no_chr_prefix.gtf'
]
for gtf_file in gtf_files:
command = [
"bin/dr-disco", "integrate", "--gtf", gtf_file, input_file,
output_file
]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
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_01_complementary(self):
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = 'frameshift-prediction_01-complementary'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
gtf_files = [
TEST_DIR + 'frameshift_example.gtf',
TEST_DIR + 'frameshift_example.no_chr_prefix.gtf'
]
for gtf_file in gtf_files:
command = [
"bin/dr-disco", "integrate", "--gtf", gtf_file, input_file,
output_file
]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_01(self): # example of in-frame fusion - strands are RNA strand
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = 'frameshift-prediction_01'
# both do have their DNA strand at minus!! :
#
# <=(-)=| acceptor in negative strand at RNA
# =====(+)=====>| donor in positive strand at RNA
# donor acceptor
# fusions = chr1', 1035203, '+'], ['chr1', 999610, '-'])
# 1', 1035203, '+'], ['1', 999610, '-'])] # strands are at RNA level, and gene order is DONOR, ACCEPTOR
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
gtf_files = [
TEST_DIR + 'frameshift_example.gtf',
TEST_DIR + 'frameshift_example.no_chr_prefix.gtf'
]
for gtf_file in gtf_files:
command = [
"bin/dr-disco", "integrate", "--gtf", gtf_file, input_file,
output_file
]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
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):
input_file = TEST_DIR + "test_01.ref.fa"
output_file_n = T_TEST_DIR + "test_02.logo-n.fa"
output_file_p = T_TEST_DIR + "test_02.logo-p.fa"
test_file_n = TEST_DIR + "test_02.logo-n.fa"
test_file_p = TEST_DIR + "test_02.logo-p.fa"
command = [
'dr-disco', 'logo-sequence', 'chr2:3', input_file, '-n', '4', '-p',
'4', output_file_n, output_file_p
]
print(" ".join(command))
self.assertEqual(
subprocess.call(command),
0) # Ensure error code is 0 - no exceptions have been thrown
self.assertTrue(filecmp.cmp(test_file_n, output_file_n),
msg=get_diff(test_file_n, output_file_n))
self.assertTrue(filecmp.cmp(test_file_p, output_file_p),
msg=get_diff(test_file_p, output_file_p))
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_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_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_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_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_classify_16(self):
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = '16'
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"
command = ["bin/dr-disco", "classify", input_file, output_file]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
def test_02_c(self):
input_file = TEST_DIR + "test_terg_02.bam"
output_file = T_TEST_DIR + "test_terg_02.filtered.bam"
output_file_s = T_TEST_DIR + "test_terg_02.filtered.sam"
test_file = TEST_DIR + "test_terg_02.filtered.sam"
c = BAMExtract(input_file, False)
c.extract("chr7:151000000-153000000", "chr5:1-2", output_file)
# Bam2Sam
fhq = open(output_file_s, "w")
fhq.write(pysam.view(output_file))
fhq.close()
self.assertTrue(filecmp.cmp(test_file, output_file_s), msg=get_diff(test_file, output_file_s))
def test_01_a(self):
# Tests a file that has not (yet) been fixed with `dr-disco fix`
input_file = TEST_DIR + "test_terg_01.bam"
output_file = T_TEST_DIR + "test_terg_01.filtered.bam"
output_file_s = T_TEST_DIR + "test_terg_01.filtered.sam"
test_file = TEST_DIR + "test_terg_02.filtered.sam"
c = BAMExtract(input_file, False)
c.extract("chr21:39000000-40000000", "chr5:1-2", output_file)
# Bam2Sam
fhq = open(output_file_s, "w")
fhq.write(pysam.view(output_file))
fhq.close()
self.assertTrue(filecmp.cmp(test_file, output_file_s), msg=get_diff(test_file, output_file_s))
def test_02_s041_no_gtf(self):
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = 'terg_s041_b'
input_file = TEST_DIR + "test_" + test_id + ".in.dbed"
test_file = TEST_DIR + "test_" + test_id + ".out.txt"
output_file = T_TEST_DIR + "test_" + test_id + ".out.txt"
command = ["bin/dr-disco", "integrate", input_file, output_file]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
self.assertTrue(filecmp.cmp(test_file, output_file),
msg=get_diff(test_file, output_file))
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_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_01(self):
input_file = TEST_DIR + "test_terg_01.filtered.bam"
output_file = T_TEST_DIR + "test_terg_01.filtered.fixed.bam"
output_file_s = T_TEST_DIR + "test_terg_01.filtered.fixed.sam"
test_file = TEST_DIR + "test_terg_01.filtered.fixed.sam"
alignment_handle = ChimericAlignment(input_file)
alignment_handle.convert(output_file, "tmp")
# Bam2Sam
fhq = open(output_file_s, "w")
fhq.write(pysam.view(output_file))
fhq.close()
#self.assertTrue(filecmp.cmp(test_file, output_file_s), msg="diff '" + test_file + "' '" + output_file_s + "':\n" + subprocess.Popen(['diff', test_file, output_file_s], stdout=subprocess.PIPE).stdout.read())
self.assertTrue(filecmp.cmp(test_file, output_file_s),
msg=get_diff(test_file, output_file_s))
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_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_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 = '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_02(self):
if not os.path.exists("tmp"):
os.mkdir("tmp")
input_file = TEST_DIR + "test_terg_02.filtered.bam"
output_file = T_TEST_DIR + "test_terg_02.filtered.fixed.bam"
output_file_s = T_TEST_DIR + "test_terg_02.filtered.fixed.sam"
test_file = TEST_DIR + "test_terg_02.filtered.fixed.sam"
alignment_handle = ChimericAlignment(input_file)
alignment_handle.convert(output_file, "tmp")
# Bam2Sam
fhq = open(output_file_s, "w")
fhq.write(pysam.view(output_file))
fhq.close()
self.assertTrue(filecmp.cmp(test_file, output_file_s),
msg=get_diff(test_file, output_file_s))
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_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_sj_01(self):
TEST_DIR, T_TEST_DIR = self.__get_temp_dirs()
test_id = 'splice_site_motif_01'
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
command = ["bin/dr-disco", "fix", input_sam, input_bam]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
# fixed bam -> dr-disco detect
command = ["bin/dr-disco", "detect", "-m", "0", input_bam, input_file]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
# dr-disco-detect (skip classify) -> dr-disco integrate
command = [
"bin/dr-disco", "integrate", "--fasta", fasta_file, input_file,
output_file
]
self.assertEqual(subprocess.call(command),
0,
msg=" ".join([str(x) for x in command]))
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))
