def setUp(self):
self.tmp_root = tempfile.mkdtemp()
args = get_default_args()
provider = gcf.GcfProvider(self.tmp_root)
test_utils.set_pattern(provider, 'basic')
self.proj = project.Project(args, provider)
test_utils.create_fastq_files("gcf", "basic", self.tmp_root)
def setUp(self):
self.tmp_root = tempfile.mkdtemp()
args = get_default_args()
provider = sll.SllProvider(self.tmp_root)
test_utils.set_pattern(provider, 'project_first')
self.proj = project.Project(args, provider)
test_utils.create_fastq_files("sll", "project_first", self.tmp_root)
def test_verify_flowcell_dir():
good1 = ["150606_ABCDE12345"]
good2 = ["000000_AAAAAAAAAA"]
bad = ["1"]
provider = sll.SllProvider(".")
test_utils.set_pattern(provider, 'project_first')
assert_equal(provider.verify_flowcell_dir(good1), True)
assert_equal(provider.verify_flowcell_dir(good2), True)
assert_raises(ValueError, provider.verify_flowcell_dir, bad)
assert_raises(ValueError, provider.verify_flowcell_dir, good1 + good2)
def test_verify_fastq_files_lane_first():
fastq1 = '5_150702_BC6U56ANXX_P1954_101_1.fastq.gz'
fastq2 = '5_150702_BC6U56ANXX_P1954_101_2.fastq.gz'
good = [fastq1, fastq2]
filtered = ["abc123", "foo.txt"]
bad = ["abc123.fastq.gz"]
provider = sll.SllProvider(".")
test_utils.set_pattern(provider, 'lane_first')
assert_equal(provider.verify_fastq_files(good), True)
assert_equal(provider.verify_fastq_files(filtered), True)
assert_raises(ValueError, provider.verify_fastq_files, bad)
def test_verify_fastq_files_project_first():
fastq1 = 'P1777_101_GTTTCG_L001_R1_001.fastq.gz'
fastq2 = 'P1777_101_GTTTCG_L001_R2_001.fastq.gz'
good = [fastq1, fastq2]
filtered = ["abc123", "foo.txt"]
bad = ["abc123.fastq.gz"]
provider = sll.SllProvider(".")
test_utils.set_pattern(provider, 'project_first')
assert_equal(provider.verify_fastq_files(good), True)
assert_equal(provider.verify_fastq_files(filtered), True)
assert_raises(ValueError, provider.verify_fastq_files, bad)
def setUp(self):
self.tmp_root = tempfile.mkdtemp()
self.provider = sll.SllProvider(self.tmp_root)
test_utils.set_pattern(self.provider, 'project_first')
sample_prefix = "P1777"
sample_format = "{0}_{1}"
flowcell_dir = "150305_AHCYF2ADXX"
for i in range(1, 4):
root = os.path.join(self.tmp_root,
sample_format.format(sample_prefix, i))
fastq_root = os.path.join(root, flowcell_dir)
os.makedirs(fastq_root)
test_utils.create_fastq_files("sll", "project_first",
fastq_root, i, bad_format=True)
def test_get_samples_basic(self):
fastq_format = "{3}/ABC012345_S{0}_{1}_R{2}_001.fastq.gz"
lane1_1 = project.Lane("L001", reads=[
project.Read(fastq_format.format(1, "L001", 1, self.tmp_root)),
project.Read(fastq_format.format(1, "L001", 2, self.tmp_root))
])
lane2_1 = project.Lane("L002", reads=[
project.Read(fastq_format.format(1, "L002", 1, self.tmp_root)),
project.Read(fastq_format.format(1, "L002", 2, self.tmp_root))
])
lane1_2 = project.Lane("L001", reads=[
project.Read(fastq_format.format(2, "L001", 1, self.tmp_root)),
project.Read(fastq_format.format(2, "L001", 2, self.tmp_root))
])
lane2_2 = project.Lane("L002", reads=[
project.Read(fastq_format.format(2, "L002", 1, self.tmp_root)),
project.Read(fastq_format.format(2, "L002", 2, self.tmp_root))
])
lane1_3 = project.Lane("L001", reads=[
project.Read(fastq_format.format(3, "L001", 1, self.tmp_root)),
project.Read(fastq_format.format(3, "L001", 2, self.tmp_root))
])
lane2_3 = project.Lane("L002", reads=[
project.Read(fastq_format.format(3, "L002", 1, self.tmp_root)),
project.Read(fastq_format.format(3, "L002", 2, self.tmp_root))
])
lane1_4 = project.Lane("L001", reads=[
project.Read(fastq_format.format(4, "L001", 1, self.tmp_root)),
project.Read(fastq_format.format(4, "L001", 2, self.tmp_root))
])
lane2_4 = project.Lane("L002", reads=[
project.Read(fastq_format.format(4, "L002", 1, self.tmp_root)),
project.Read(fastq_format.format(4, "L002", 2, self.tmp_root))
])
sample1 = project.Sample("S1", [lane1_1, lane2_1])
sample2 = project.Sample("S2", [lane1_2, lane2_2])
sample3 = project.Sample("S3", [lane1_3, lane2_3])
sample4 = project.Sample("S4", [lane1_4, lane2_4])
expected = [sample1, sample2, sample3, sample4]
provider = gcf.GcfProvider(self.tmp_root)
test_utils.set_pattern(provider, "basic")
result = provider.get_samples()
for res, exp in zip(result, expected):
assert_equal(res.sample_id, exp.sample_id)
for res_lane, exp_lane in zip(res.lanes, exp.lanes):
assert_equal(res_lane, exp_lane)
def test_get_samples_lane_first(self):
fastq_format = "{3}/{1}_151224_A1B2C3XYZ0_P1234_{0}_{2}.fastq.gz"
expected = self.create_expected(self.lane_first_dir,
fastq_format, "")
provider = sll.SllProvider(self.lane_first_dir)
test_utils.set_pattern(provider, "lane_first")
result = provider.get_samples()
for res, exp in zip(result, expected):
assert_equal(res.sample_id, exp.sample_id)
for res_lane, exp_lane in zip(res.lanes, exp.lanes):
assert_equal(res_lane, exp_lane)
def test_get_samples_project_first(self):
fastq_format = "{3}/P1234_{0}_TAGCTT_{1}_R{2}_001.fastq.gz"
expected = self.create_expected(self.project_first_dir,
fastq_format, "L00")
provider = sll.SllProvider(self.project_first_dir)
test_utils.set_pattern(provider, "project_first")
result = provider.get_samples()
for res, exp in zip(result, expected):
assert_equal(res.sample_id, exp.sample_id)
for res_lane, exp_lane in zip(res.lanes, exp.lanes):
assert_equal(res_lane, exp_lane)
def test_basic_file_pattern():
name = "ABC012345_S1_L001_R1_001.fastq.gz"
provider = gcf.GcfProvider(".")
test_utils.set_pattern(provider, 'basic')
assert_equal(provider.get_filename_part(name),
"ABC012345_S1_L001_R1_001.fastq.gz")
assert_equal(provider.get_filename_part(name, 'project_id'),
"ABC012345")
assert_equal(provider.get_filename_part(name, 'sample_id'),
"S1")
assert_equal(provider.get_filename_part(name, 'lane'),
"L001")
assert_equal(provider.get_filename_part(name, 'read'),
"1")
assert_equal(provider.get_filename_part(name, 'file_ending'),
".fastq.gz")
assert_equal(provider.get_filename_part(name, 'gzip'),
".gz")
def test_get_filename_part_lane_first():
lane_first = '5_150702_BC6U56ANXX_P1954_101_1.fastq.gz'
provider = sll.SllProvider(".")
test_utils.set_pattern(provider, 'lane_first')
assert_equal(provider.get_filename_part(lane_first, 'all'),
lane_first)
assert_equal(provider.get_filename_part(lane_first, 'sample_id'),
"P1954_101")
assert_equal(provider.get_filename_part(lane_first, 'project_id'),
"P1954")
assert_equal(provider.get_filename_part(lane_first, 'scilab_id'),
"101")
assert_equal(provider.get_filename_part(lane_first, 'lane'),
"5")
assert_equal(provider.get_filename_part(lane_first, 'read'),
"1")
assert_equal(provider.get_filename_part(lane_first, 'file_ending'),
".fastq.gz")
def test_get_filename_part_project_first():
project_first = 'P1777_101_GTTTCG_L001_R2_001.fastq'
provider = sll.SllProvider(".")
test_utils.set_pattern(provider, 'project_first')
assert_equal(provider.get_filename_part(project_first, 'all'),
project_first)
assert_equal(provider.get_filename_part(project_first, 'sample_id'),
"P1777_101")
assert_equal(provider.get_filename_part(project_first, 'project_id'),
"P1777")
assert_equal(provider.get_filename_part(project_first, 'scilab_id'),
"101")
assert_equal(provider.get_filename_part(project_first, 'lane'),
"L001")
assert_equal(provider.get_filename_part(project_first, 'read'),
"R2")
assert_equal(provider.get_filename_part(project_first, 'file_ending'),
".fastq")
def test_aligning_cmd():
args = get_default_args()
expected = {}
expected["L001"] = ("STAR --runThreadN 8 --genomeDir ./genomes "
"--readFilesIn "
"{0}/{1}/output_paired_1.fastq.gz "
"{0}/{1}/output_paired_2.fastq.gz "
"--readFilesCommand zcat").format(args['run_root'],
"L001")
expected["L002"] = ("STAR --runThreadN 8 --genomeDir ./genomes "
"--readFilesIn "
"{0}/{1}/output_paired_1.fastq "
"{0}/{1}/output_paired_2.fastq").format(
args['run_root'],
"L002")
trimmis = trimmostar.Trimmostar(args)
provider = trimmis.provider
test_utils.set_pattern(provider, 'project_first')
assert_equal(trimmis.aligning_cmd(), expected)
def test_trimming_cmd():
args = get_default_args()
args.update({
'trimmomatic_jar': "trimmomatic-0.33.jar",
'threads': '8',
'quality_type': "phred33",
'trimmomatic_log_file': "trimmomatic.log",
'mode': "paired",
'trimming_steps': ["LEADING:20", "TRAILING:20"]
})
expected_pattern = ("java -jar trimmomatic-0.33.jar PE -threads 8 "
"-phred33 -trimlog trimmomatic.log "
"{0} "
"{1} "
"LEADING:20 TRAILING:20")
expected = {}
expected["L001"] = expected_pattern.format(
("{0}/P1777_101_GTTTCG_L001_R1_001.fastq.gz "
"{0}/P1777_101_GTTTCG_L001_R2_001.fastq.gz").format(args['run_root']),
("{0}/{1}/output_paired_1.fastq.gz "
"{0}/{1}/output_unpaired_1.fastq.gz "
"{0}/{1}/output_paired_2.fastq.gz "
"{0}/{1}/output_unpaired_2.fastq.gz").format(args['run_root'], "L001")
)
expected["L002"] = expected_pattern.format(
("{0}/P1777_101_GTTTCG_L002_R1_001.fastq "
"{0}/P1777_101_GTTTCG_L002_R2_001.fastq").format(args['run_root']),
("{0}/{1}/output_paired_1.fastq "
"{0}/{1}/output_unpaired_1.fastq "
"{0}/{1}/output_paired_2.fastq "
"{0}/{1}/output_unpaired_2.fastq").format(args['run_root'], "L002")
)
trimmis = trimmostar.Trimmostar(args)
provider = trimmis.provider
test_utils.set_pattern(provider, 'project_first')
assert_equal(trimmis.trimming_cmd(), expected)
def test_verify_directory_structure_bad_directory(self):
provider = gcf.GcfProvider(self.bad_root)
test_utils.set_pattern(provider, 'basic')
assert(provider.verify_directory_structure())
def test_generate_commands(self):
trimmis = trimmostar.Trimmostar(self.args)
provider = trimmis.provider
test_utils.set_pattern(provider, 'project_first')
assert_equal(trimmis.generate_commands(), self.expected_cmds)
def test_get_filename_part_bad_name():
bad_name = '123.fastq.gz'
provider = sll.SllProvider(".")
test_utils.set_pattern(provider, 'project_first')
assert_raises(ValueError, provider.get_filename_part, bad_name, 'all')