def test_sequana_config():
s = snaketools.Module("compressor")
config = snaketools.SequanaConfig(s.config)
assert config.config.get("compressor")["source"] == "fastq.gz"
assert config.config.get("kraken:dummy") == None
# --------------------------------- tests different constructors
config = snaketools.SequanaConfig()
config = snaketools.SequanaConfig({"test": 1})
assert config.config.test == 1
# with a dictionary
config = snaketools.SequanaConfig(config.config)
# with a sequanaConfig instance
config = snaketools.SequanaConfig(config)
# with a non-yaml file
try:
json = sequana_data('test_summary_fastq_stats.json')
config = snaketools.SequanaConfig(json)
assert False
except:
assert True
try:
config = snaketools.SequanaConfig("dummy_dummy")
assert False
except:
assert True
# Test an exception
s = snaketools.Module("compressor")
config = snaketools.SequanaConfig(s.config)
config._recursive_update(config._yaml_code,
{"input_directory_dummy": "test"})
#config.check_config_with_schema(s.schema_config)
# loop over all pipelines, read the config, save it and check the content is
# identical. This requires to remove the templates. We want to make sure the
# empty strings are kept and that "no value" are kept as well
#
# field1: ""
# field2:
#
# is unchanged
from easydev import TempFile
output = TempFile(suffix=".yaml")
for pipeline in snaketools.pipeline_names:
config_filename = Module(pipeline)._get_config()
cfg1 = SequanaConfig(config_filename)
cfg1.cleanup() # remove templates and strip strings
cfg1.save(output.name)
cfg2 = SequanaConfig(output.name)
assert cfg2._yaml_code == cfg1._yaml_code
cfg2._update_config()
assert cfg1.config == cfg2.config
output.delete()
def test_sequana_config():
s = snaketools.Module("quality_control")
config = snaketools.SequanaConfig(s.config)
assert config.config.get("kraken:dummy", "test") == "test"
assert config.config.get("kraken:dummy") == None
# --------------------------------- tests different constructors
config = snaketools.SequanaConfig()
config = snaketools.SequanaConfig({"test":1})
assert config.config.test == 1
# with a dictionary
config = snaketools.SequanaConfig(config.config)
# with a sequanaConfig instance
config = snaketools.SequanaConfig(config)
# with a non-yaml file
try:
json = sequana_data('test_summary_fastq_stats.json')
config = snaketools.SequanaConfig(json)
assert False
except:
assert True
try:
config = snaketools.SequanaConfig("dummy_dummy")
assert False
except:
assert True
# Test an exception
s = snaketools.Module("quality_control")
config = snaketools.SequanaConfig(s.config)
config._recursive_update(config._yaml_code, {"input_directory_dummy": "test"})
# loop over all pipelines, read the config, save it and check the content is
# identical. This requires to remove the templates. We want to make sure the
# empty strings are kept and that "no value" are kept as well
#
# field1: ""
# field2:
#
# is unchanged
from easydev import TempFile
output = TempFile(suffix=".yaml")
for pipeline in snaketools.pipeline_names:
config_filename = Module(pipeline)._get_config()
cfg1 = SequanaConfig(config_filename)
cfg1.cleanup() # remove templates and strip strings
cfg1.save(output.name)
cfg2 = SequanaConfig(output.name)
assert cfg2._yaml_code == cfg1._yaml_code
cfg2._update_config()
assert cfg1.config == cfg2.config
output.delete()
def test_pipeline_manager():
# test missing input_directory
cfg = SequanaConfig({})
try:
pm = snaketools.PipelineManager("custom", cfg)
assert False
except:
assert True
# normal behaviour but no input provided:
config = Module("quality_control")._get_config()
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
try:
pm = snaketools.PipelineManager("custome", cfg)
assert False
except:
assert True
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
file1 = sequana_data("Hm2_GTGAAA_L005_R1_001.fastq.gz")
file2 = sequana_data("Hm2_GTGAAA_L005_R2_001.fastq.gz")
cfg.config.input_samples['file1'] = file1
cfg.config.input_samples['file2'] = file2
pm = snaketools.PipelineManager("custome", cfg)
assert pm.paired == True
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
file1 = sequana_data("Hm2_GTGAAA_L005_R1_001.fastq.gz")
cfg.config.input_samples['file1'] = file1
pm = snaketools.PipelineManager("custome", cfg)
assert pm.paired == False
pm.getlogdir("fastqc")
pm.getwkdir("fastqc")
pm.getrawdata()
pm.getreportdir("test")
pm.getname("fastqc")
def test_pipeline_manager():
# test missing input_directory
cfg = SequanaConfig({})
try:
pm = snaketools.PipelineManager("custom", cfg)
assert False
except:
assert True
# normal behaviour but no input provided:
config = Module("quality_control")._get_config()
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
try:
pm = snaketools.PipelineManager("custome", cfg)
assert False
except:
assert True
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
file1 = sequana_data("Hm2_GTGAAA_L005_R1_001.fastq.gz")
file2 = sequana_data("Hm2_GTGAAA_L005_R2_001.fastq.gz")
cfg.config.input_samples['file1'] = file1
cfg.config.input_samples['file2'] = file2
pm = snaketools.PipelineManager("custome", cfg)
assert pm.paired == True
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
file1 = sequana_data("Hm2_GTGAAA_L005_R1_001.fastq.gz")
cfg.config.input_samples['file1'] = file1
pm = snaketools.PipelineManager("custome", cfg)
assert pm.paired == False
pm.getlogdir("fastqc")
pm.getwkdir("fastqc")
pm.getrawdata()
pm.getreportdir("test")
pm.getname("fastqc")
def test_pipeline_manager():
# test missing input_directory
cfg = SequanaConfig({})
try:
pm = snaketools.PipelineManager("custom", cfg)
assert False
except:
assert True
# normal behaviour but no input provided:
config = Module("compressor")._get_config()
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
try:
pm = snaketools.PipelineManager("custome", cfg)
assert False
except:
assert True
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
file1 = sequana_data("Hm2_GTGAAA_L005_R1_001.fastq.gz")
cfg.config.input_directory, cfg.config.input_pattern = os.path.split(file1)
#file2 = sequana_data("Hm2_GTGAAA_L005_R2_001.fastq.gz")
pm = snaketools.PipelineManager("custom", cfg)
assert pm.paired == False
cfg = SequanaConfig(config)
cfg.cleanup() # remove templates
cfg.config.input_directory, cfg.config.input_pattern = os.path.split(file1)
cfg.config.input_pattern = "Hm*gz"
#file1 = sequana_data("Hm2_GTGAAA_L005_R1_001.fastq.gz")
pm = snaketools.PipelineManager("custom", cfg)
pm.plot_stats()
assert pm.paired == True
pm.getlogdir("fastqc")
pm.getwkdir("fastqc")
pm.getrawdata()
pm.getreportdir("test")
pm.getname("fastqc")
# Test different configuration of input_directory, input_readtag,
# input_pattern
# Test the _R[12]_ paired
with tempfile.TemporaryDirectory() as tmpdir:
cfg = SequanaConfig()
cfgname = tmpdir + "/config.yaml"
cfg.config.input_pattern = "*fastq.gz"
cfg.config.input_directory = tmpdir
cfg.config.input_readtag = "_R[12]_"
cfg._update_yaml()
cfg.save(cfgname)
cmd = "touch {}/test_R1_.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
cmd = "touch {}/test_R2_.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
pm = snaketools.PipelineManager("test", cfgname)
assert pm.paired == True
# Test the _[12]_ paired
with tempfile.TemporaryDirectory() as tmpdir:
cfg = SequanaConfig()
cfgname = tmpdir + "/config.yaml"
cfg.config.input_pattern = "*fastq.gz"
cfg.config.input_directory = tmpdir
cfg.config.input_readtag = "_[12]."
cfg._update_yaml()
cfg.save(cfgname)
cmd = "touch {}/test_1.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
cmd = "touch {}/test_2.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
pm = snaketools.PipelineManager("test", cfgname)
assert pm.paired is True
# Test the _R[12]_ single end
with tempfile.TemporaryDirectory() as tmpdir:
cfg = SequanaConfig()
cfgname = tmpdir + "/config.yaml"
cfg.config.input_pattern = "*fastq.gz"
cfg.config.input_directory = tmpdir
cfg.config.input_readtag = "_R[12]_"
cfg._update_yaml()
cfg.save(cfgname)
cmd = "touch {}/test_R1_.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
pm = snaketools.PipelineManager("test", cfgname)
assert pm.paired is False
# Test the _R[12]_ single end
with tempfile.TemporaryDirectory() as tmpdir:
cfg = SequanaConfig()
cfgname = tmpdir + "/config.yaml"
cfg.config.input_pattern = "*fq.gz" # wrong on purpose
cfg.config.input_directory = tmpdir
cfg.config.input_readtag = "_R[12]_"
cfg._update_yaml()
cfg.save(cfgname)
cmd = "touch {}/test_R1_.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
try:
pm = snaketools.PipelineManager("test", cfgname)
assert False
except:
assert True
# Test the _R[12]_ single end
with tempfile.TemporaryDirectory() as tmpdir:
cfg = SequanaConfig()
cfgname = tmpdir + "/config.yaml"
cfg.config.input_pattern = "*fastq.gz"
cfg.config.input_directory = tmpdir
cfg.config.input_readtag = "R[12]_"
cfg._update_yaml()
cfg.save(cfgname)
cmd = "touch {}/testR1_.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
cmd = "touch {}/testR2_.fastq.gz".format(tmpdir)
subprocess.call(cmd.split())
try:
pm = snaketools.PipelineManager("test", cfgname)
assert False
except:
assert True