def test_intersect(self):
isamp = self.gen_fake_isamp()
series = Series('GSE0')
sample_list = [Sample('GSM10', series), Sample('GSM20', series)]
for __ in sample_list :
series.add_passed_sample(__)
self.assertEqual(ppr.intersect(series, isamp), sample_list)
def test_gen_orig_params_per_with_a_single_sra(self):
# mock a series and sample
series = Series('GSE123456', 'GSE123456_family.soft.subset')
sample = Sample('GSM1', series)
sample.outdir = 'some_outdir/GSE123456/some_species/GSM1'
series.add_passed_sample(sample)
with mock.patch('rsempipeline.utils.download.open',
mock.mock_open(read_data=SRA_INFO_YAML_SINGLE_SRA)):
vals = download.gen_orig_params_per(sample)
self.assertEqual(vals, [
[None, ['some_outdir/GSE123456/some_species/GSM1/SRX685892/SRR1557065/SRR1557065.sra',
'some_outdir/GSE123456/some_species/GSM1/SRR1557065.sra.download.COMPLETE'], sample]])
class SampleTestCase(unittest.TestCase):
def setUp(self):
self.series = Series("GSE123456", "GSE123456_family.soft.subset")
self.sample = Sample("GSM1", self.series)
def test_is_info_complete(self):
self.assertFalse(self.sample.is_info_complete())
self.sample.organism = "Mus musculus"
self.assertFalse(self.sample.is_info_complete())
self.sample.url = "ftp://ftp-trace.ncbi.nlm.nih.gov/sra/sra-instant/reads/ByExp/sra/SRX/SRX000/SRX000000"
self.assertTrue(self.sample.is_info_complete())
@mock.patch("rsempipeline.utils.objs.Sample.is_info_complete")
def test_gen_outdir(self, mock_is_info_complete):
self.sample.organism = "Mus musculus"
self.sample.url = "ftp://ftp-trace.ncbi.nlm.nih.gov/sra/sra-instant/reads/ByExp/sra/SRX/SRX000/SRX000000"
self.assertEqual(self.sample.gen_outdir("some_outdir"), "some_outdir/GSE123456/mus_musculus/GSM1")
self.assertEqual(self.sample.outdir, "some_outdir/GSE123456/mus_musculus/GSM1")
mock_is_info_complete.return_value = False
self.assertRaisesRegexp(ValueError, "not information complete", self.sample.gen_outdir, "some_outdir")
def test___str__(self):
self.sample.organism = "Mus musculus"
self.sample.url = "ftp://ftp-trace.ncbi.nlm.nih.gov/sra/sra-instant/reads/ByExp/sra/SRX/SRX000/SRX000000"
self.sample.gen_outdir("some_outdir")
self.assertEqual(str(self.sample), "<GSM1 (0/0/0) of GSE123456 at some_outdir/GSE123456/mus_musculus/GSM1>")
def test___repr__(self):
self.sample.organism = "Mus musculus"
self.sample.url = "ftp://ftp-trace.ncbi.nlm.nih.gov/sra/sra-instant/reads/ByExp/sra/SRX/SRX000/SRX000000"
self.sample.gen_outdir("some_outdir")
self.assertEqual(str(self.sample), "<GSM1 (0/0/0) of GSE123456 at some_outdir/GSE123456/mus_musculus/GSM1>")
def test_gen_all_samples_from_soft_and_isamp(
self, mock_get_isamp, mock_analyze_one, mock_sanity_check):
mock_sanity_check.return_value = True
mock_get_isamp.return_value = self.gen_fake_isamp()
fake_series = Series('GSE0')
sample_list = [Sample('GSM10', fake_series), Sample('GSM20', fake_series)]
for __ in sample_list :
__.organism = 'H**o Sapiens'
fake_series.add_passed_sample(__)
mock_analyze_one.return_value = sample_list
self.assertEqual(ppr.gen_all_samples_from_soft_and_isamp(
['soft1'], 'isamp_file_or_str', {'INTERESTED_ORGANISMS': ['H**o Sapiens']}),
sample_list)
def test_add(self, mock_is_info_complete):
series = Series('GSE123456', 'GSE123456_family.soft.subset')
current_sample = Sample('GSM1', series)
mock_is_info_complete.return_value = True
self.assertEqual(soft_parser.add(current_sample, series, 1), 2)
mock_is_info_complete.return_value = False
self.assertEqual(soft_parser.add(current_sample, series, 1), 1)
def test_gen_orig_params_per_with_a_single_sra(self):
# mock a series and sample
series = Series('GSE123456', 'GSE123456_family.soft.subset')
sample = Sample('GSM1', series)
sample.outdir = 'some_outdir/GSE123456/some_species/GSM1'
series.add_passed_sample(sample)
with mock.patch('rsempipeline.utils.download.open',
mock.mock_open(read_data=SRA_INFO_YAML_SINGLE_SRA)):
vals = download.gen_orig_params_per(sample)
self.assertEqual(vals, [[
None,
[
'some_outdir/GSE123456/some_species/GSM1/SRX685892/SRR1557065/SRR1557065.sra',
'some_outdir/GSE123456/some_species/GSM1/SRR1557065.sra.download.COMPLETE'
], sample
]])
def test_gen_orig_params_per_with_multiple_sras(self):
# mock a series and sample
series = Series('GSE123456', 'GSE123456_family.soft.subset')
sample = Sample('GSM1', series)
sample.outdir = 'some_outdir/GSE123456/some_species/GSM1'
series.add_passed_sample(sample)
with mock.patch('rsempipeline.utils.download.open',
mock.mock_open(read_data=SRA_INFO_YAML_MULTIPLE_SRAS)):
vals = download.gen_orig_params_per(sample)
self.assertEqual(vals, [
# in the format of input, outputs, other params
[None, ['some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453140/SRR453140.sra', 'some_outdir/GSE123456/some_species/GSM1/SRR453140.sra.download.COMPLETE'], sample],
[None, ['some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453141/SRR453141.sra', 'some_outdir/GSE123456/some_species/GSM1/SRR453141.sra.download.COMPLETE'], sample],
[None, ['some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453142/SRR453142.sra', 'some_outdir/GSE123456/some_species/GSM1/SRR453142.sra.download.COMPLETE'], sample],
[None, ['some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453143/SRR453143.sra', 'some_outdir/GSE123456/some_species/GSM1/SRR453143.sra.download.COMPLETE'], sample]
])
def test_intersect_with_discrenpacy(self, L):
isamp = self.gen_fake_isamp()
series = Series('GSE0')
sample_list = [Sample('GSM10', series)]
for __ in sample_list :
series.add_passed_sample(__)
self.assertEqual(ppr.intersect(series, isamp), sample_list)
L.check(('rsempipeline.utils.pre_pipeline_run', 'ERROR',
'Discrepancy for GSE0: 1 GSMs in soft, 2 GSMs in isamp, and only 1 left after intersection.'),)
def test_gen_orig_params_per_with_multiple_sras(self):
# mock a series and sample
series = Series('GSE123456', 'GSE123456_family.soft.subset')
sample = Sample('GSM1', series)
sample.outdir = 'some_outdir/GSE123456/some_species/GSM1'
series.add_passed_sample(sample)
with mock.patch('rsempipeline.utils.download.open',
mock.mock_open(read_data=SRA_INFO_YAML_MULTIPLE_SRAS)):
vals = download.gen_orig_params_per(sample)
self.assertEqual(
vals,
[
# in the format of input, outputs, other params
[
None,
[
'some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453140/SRR453140.sra',
'some_outdir/GSE123456/some_species/GSM1/SRR453140.sra.download.COMPLETE'
], sample
],
[
None,
[
'some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453141/SRR453141.sra',
'some_outdir/GSE123456/some_species/GSM1/SRR453141.sra.download.COMPLETE'
], sample
],
[
None,
[
'some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453142/SRR453142.sra',
'some_outdir/GSE123456/some_species/GSM1/SRR453142.sra.download.COMPLETE'
], sample
],
[
None,
[
'some_outdir/GSE123456/some_species/GSM1/SRX135160/SRR453143/SRR453143.sra',
'some_outdir/GSE123456/some_species/GSM1/SRR453143.sra.download.COMPLETE'
], sample
]
])
def test_analyze_one(self, mock_parse, L):
fake_isamp = self.gen_fake_isamp()
fake_series = Series('GSE0')
sample_list = [Sample('GSM10', fake_series)]
for __ in sample_list :
__.organism = 'H**o Sapiens'
fake_series.add_passed_sample(__)
mock_parse.return_value = fake_series
self.assertEqual(ppr.analyze_one('GSE0_family.soft.subset', fake_isamp, ['H**o sapiens']),
sample_list)
L.check(('rsempipeline.utils.pre_pipeline_run', 'ERROR',
'Discrepancy for GSE0: 1 GSMs in soft, 2 GSMs in isamp, and only 1 left after intersection.'),)
def parse(soft_file, interested_organisms):
"""Parse the soft file
:param interested_organisms: a list of interested organisms: ['H**o
sapiens', 'Mus musculus']
"""
logger.info("Parsing file: {0} ...".format(soft_file))
series_name_from_file = get_series_name_from(soft_file)
print series_name_from_file
# Assume one GSE per soft file
# index: the index of all passed samples, unpassed samples are not indexed
index, series, current_sample = 1, None, None
with open(soft_file, 'rb') as inf:
for line in inf:
label, value = [__.strip() for __ in line.split('=')]
if label == '^SERIES':
series = Series(value, os.path.abspath(soft_file))
if series.name != series_name_from_file:
msg = ('series contained in the soft file doesn\'t match '
'that in the filename: {0} != {1}'.format(
series, series_name_from_file))
raise ValueError(msg)
elif label == '^SAMPLE':
index = add(current_sample, series, index)
current_sample = Sample(name=value, series=series)
if current_sample:
current_sample = update(current_sample, label, value,
interested_organisms)
if series is not None:
# add the last sample
add(current_sample, series, index)
logger.info("{0}: {1}/{2} samples passed".format(
series.name, series.num_passed_samples(),
series.num_samples()))
logger.info('=' * 30)
return series
def setUp(self):
self.series = Series("GSE123456", "GSE123456_family.soft.subset")
self.sample = Sample("GSM1", self.series)
