def test_dots_in_filename(self):
k = KmcComplex(os.getcwd(), 1, 1, [SampleData('a.b#c_1.fastq', 'a.b#c_2.fastq'), SampleData('a.b.c.d_1.fastq', 'a.b.c.d_2.fastq')], [SampleData('a.b.c.d.e_1.fastq', 'a.b.c.d.e_2.fastq')], 'intersection', False)
k.create_config_files()
with open(os.path.join(k.temp_working_dir, 'traits_config_file'), 'r') as actual_file:
actual_config_content = actual_file.read()
self.assertEqual(actual_config_content, """\
INPUT:
a_b_c = kmc_a.b#c
a_b_c_d = kmc_a.b.c.d
a_b_c_d_e = kmc_a.b.c.d.e
OUTPUT:
traits = a_b_c*a_b_c_d
OUTPUT_PARAMS:
-ci1
""")
with open(os.path.join(k.temp_working_dir, 'nontraits_config_file'), 'r') as actual_file:
actual_config_content = actual_file.read()
self.assertEqual(actual_config_content, """\
INPUT:
a_b_c = kmc_a.b#c
a_b_c_d = kmc_a.b.c.d
a_b_c_d_e = kmc_a.b.c.d.e
OUTPUT:
nontraits = a_b_c_d_e
OUTPUT_PARAMS:
-ci1
""")
def extract_samples(self):
samples = []
self.logger.warning("Reading input spreadsheet")
with open(self.filename) as csvfile:
spreadsheetreader = csv.reader(csvfile, delimiter=',')
for row in spreadsheetreader:
if len(row) == 2:
forward_file = row[0]
reverse_file = row[1]
for filename in [forward_file, reverse_file]:
if not os.path.exists(filename):
raise Exception(
'Input file in spreadsheet doesnt exit: ' +
filename)
self.logger.warning("Found input files")
samples.append(SampleData(forward_file, reverse_file))
elif len(row) == 1:
filename = row[0]
if not os.path.exists(filename):
raise Exception(
'Input file in spreadsheet doesnt exit: ' +
filename)
self.logger.warning("Found input file")
samples.append(SampleData(filename))
else:
continue
return samples
def test_sample_definitions_str(self):
k = KmcComplex(os.getcwd(), 1, 1, [SampleData('a_1.fastq', 'a_2.fastq'), SampleData('b_1.fastq', 'b_2.fastq')], [SampleData('c_1.fastq', 'c_2.fastq')], 'union', False)
expected_output = """\
a = kmc_a
b = kmc_b
c = kmc_c
"""
self.assertEqual(k.sample_definitions_str(),expected_output)
k.cleanup()
def test_samples_to_set_operation_str(self):
k = KmcComplex(os.getcwd(), 1, 1, [
SampleData('a_1.fastq', 'a_2.fastq'),
SampleData('b_1.fastq', 'b_2.fastq')
], [SampleData('c_1.fastq', 'c_2.fastq')], 'union')
self.assertEqual(k.trait_samples_to_set_operation_str(),
'traits = a+b')
self.assertEqual(k.nontrait_samples_to_set_operation_str(),
'nontraits = c')
k.cleanup()
def test_create_config_file_intersection(self):
k = KmcComplex(os.getcwd(), 1, 1, [
SampleData('a_1.fastq', 'a_2.fastq'),
SampleData('b_1.fastq', 'b_2.fastq')
], [SampleData('c_1.fastq', 'c_2.fastq')], 'intersection')
k.create_config_files()
with open(os.path.join(k.temp_working_dir, 'traits_config_file'),
'r') as actual_file:
actual_config_content = actual_file.read()
self.assertEqual(
actual_config_content, """\
INPUT:
a = kmc_a
b = kmc_b
c = kmc_c
OUTPUT:
traits = a*b
OUTPUT_PARAMS:
-ci1
""")
with open(os.path.join(k.temp_working_dir, 'nontraits_config_file'),
'r') as actual_file:
actual_config_content = actual_file.read()
self.assertEqual(
actual_config_content, """\
INPUT:
a = kmc_a
b = kmc_b
c = kmc_c
OUTPUT:
nontraits = c
OUTPUT_PARAMS:
-ci1
""")
with open(os.path.join(k.temp_working_dir, 'combined_config_file'),
'r') as actual_file:
actual_config_content = actual_file.read()
self.assertEqual(
actual_config_content, """\
INPUT:
set1 = traits
set2 = nontraits
OUTPUT:
result = set1-set2
OUTPUT_PARAMS:
-ci1
""")
k.cleanup()
def test_populate_fofn_name(self):
sample = SampleData('/path/to/sample_1.fastq.gz',
'/path/to/sample_2.fastq.gz')
i = Kmc(os.getcwd(), sample, 1, 51, 30, 200)
self.assertEqual(sample.file_of_fastq_files,
i.temp_working_dir + "/fofn")
i.cleanup()
def test_filtering_low_coverage_contigs(self):
sample = SampleData('/path/to/sample_1.fastq.gz',
'/path/to/sample_2.fastq.gz')
s = SpadesAssembly(sample, 'abc', 1, 1, '', 1, False, 24, True, False,
100)
s.remove_small_large_contigs(
os.path.join(data_dir, 'assembly_with_small_contigs.fa'),
os.path.join(data_dir,
'actual_assembly_without_low_coverage_contigs.fa'))
with open(
os.path.join(
data_dir,
'actual_assembly_without_low_coverage_contigs.fa'),
'r') as actual_file, open(
os.path.join(data_dir, 'expected_assembly_coverage.fa'),
'r') as expected_file:
actual_config_content = actual_file.read()
expected_config_content = expected_file.read()
self.assertEqual(actual_config_content, expected_config_content)
os.remove(
os.path.join(data_dir,
'actual_assembly_without_low_coverage_contigs.fa'))
def test_sample_definition_line(self):
k = KmcComplex(os.getcwd(), 1, 1, [], [], 'union')
s = SampleData('/path/to/sample#ABC_1.fastq',
'/path/to/sample#ABC_2.fastq')
self.assertEqual(k.sample_definition_line(s),
'sample_ABC = kmc_sample#ABC')
k.cleanup()
def test_create_file_of_file_names(self):
sample = SampleData('/path/to/sample_1.fastq.gz',
'/path/to/sample_2.fastq.gz')
i = Kmc(os.getcwd(), sample, 1, 51, 30, 200)
i.create_file_of_file_names(sample.file_of_fastq_files)
with open(sample.file_of_fastq_files, 'r') as actual_file:
actual_fofn_content = actual_file.read()
expected_fofn_content = """\
/path/to/sample_1.fastq.gz
/path/to/sample_2.fastq.gz
"""
self.assertEqual(actual_fofn_content, expected_fofn_content)
i.cleanup()
def test_basename_set_correctly(self):
'''test basename extracted correctly from input file with gz'''
i = SampleData('/path/to/sample_1.fastq.gz',
'/path/to/sample_2.fastq.gz')
self.assertEqual(i.basename, 'sample')