def validate(input_file_name,
working_directory,
level=logging.INFO,
logger=None):
"""
Validates any file containing sequence data.
Args:
input_file_name: An input SBML file.
working_directory: A directory where any output files produced by validation can be written.
level: Logging level, defaults to logging.INFO.
Returns:
0 on success, 1 on failure.
All statements passed to standard out via a logger, any errors throw an Exception
and result in a non-zero exit status back to the caller.
Authors:
Srividya Ramikrishnan, Matt Henderson
"""
if logger is None:
logger = script_utils.stdoutlogger(__file__)
command = os.path.join(os.environ.get("KB_TOP"), "bin/validateSBML")
validated = False
fileName = os.path.split(input_file_name)[-1]
if not os.path.isfile(input_file_name):
raise Exception("Not a file {0}".format(fileName))
logger.info("Starting SBML validation of {0}".format(fileName))
arguments = [command, input_file_name]
tool_process = subprocess.Popen(arguments, stderr=subprocess.PIPE)
stdout, stderr = tool_process.communicate()
if len(stderr) > 0:
logger.error("Validation failed on {0}".format(fileName))
else:
logger.info("Validation passed on {0}".format(fileName))
validated = True
if not validated:
raise Exception("Validation failed!")
else:
logger.info("Validation passed.")
def validate(input_file_name, working_directory, level=logging.INFO, logger=None):
"""
Validates any file containing sequence data.
Args:
input_file_name: An input SBML file.
working_directory: A directory where any output files produced by validation can be written.
level: Logging level, defaults to logging.INFO.
Returns:
0 on success, 1 on failure.
All statements passed to standard out via a logger, any errors throw an Exception
and result in a non-zero exit status back to the caller.
Authors:
Srividya Ramikrishnan, Matt Henderson
"""
if logger is None:
logger = script_utils.stdoutlogger(__file__)
command = os.path.join(os.environ.get("KB_TOP"), "bin/validateSBML")
validated = False
fileName = os.path.split(input_file_name)[-1]
if not os.path.isfile(input_file_name):
raise Exception("Not a file {0}".format(fileName))
logger.info("Starting SBML validation of {0}".format(fileName))
arguments = [command, input_file_name]
tool_process = subprocess.Popen(arguments, stderr=subprocess.PIPE)
stdout, stderr = tool_process.communicate()
if len(stderr) > 0:
logger.error("Validation failed on {0}".format(fileName))
else:
logger.info("Validation passed on {0}".format(fileName))
validated = True
if not validated:
raise Exception("Validation failed!")
else:
logger.info("Validation passed.")
def transform(shock_service_url=None, workspace_service_url=None,
workspace_name=None, object_name=None, contigset_object_name=None,
input_directory=None, working_directory=None,
level=logging.INFO, logger=None):
"""
Transforms Genbank file to KBaseGenomes.Genome and KBaseGenomes.ContigSet objects.
Args:
shock_service_url: If you have shock references you need to make.
workspace_service_url: KBase Workspace URL
workspace_name: Name of the workspace to save the data to
object_name: Name of the genome object to save
contigset_object_name: Name of the ContigSet object that is created with this Genome
input_directory: A directory of either a genbank file or a directory of partial genome files to merge
working_directory: A directory where you can do work
Returns:
Workspace objects saved to the user's workspace.
Authors:
Shinjae Yoo, Marcin Joachimiak, Matt Henderson
"""
if logger is None:
logger = script_utils.stdoutlogger(__file__, logging.INFO)
logger.info("Starting transformation of Genbank to KBaseGenomes.Genome")
# TODO get the classpath definition out into the config instead
KB_TOP = os.environ["KB_TOP"]
classpath = ["{}/lib/jars/kbase/transform/kbase_transform_deps.jar".format(KB_TOP),
"{}/lib/jars/kbase/genomes/kbase-genomes-20140411.jar".format(KB_TOP),
"{}/lib/jars/kbase/common/kbase-common-0.0.6.jar".format(KB_TOP),
"{}/lib/jars/jackson/jackson-annotations-2.2.3.jar".format(KB_TOP),
"{}/lib/jars/jackson/jackson-core-2.2.3.jar".format(KB_TOP),
"{}/lib/jars/jackson/jackson-databind-2.2.3.jar".format(KB_TOP),
"{}/lib/jars/kbase/auth/kbase-auth-1398468950-3552bb2.jar".format(KB_TOP),
"{}/lib/jars/kbase/workspace/WorkspaceClient-0.2.0.jar".format(KB_TOP)]
for p in classpath:
try:
assert os.path.exists(p)
except AssertionError, e:
raise IOError("Unable to find classpath library {}".format(p))
def validate(input_directory, working_directory, level=logging.INFO, logger=None):
"""
Validates any file containing sequence data.
Args:
input_directory: A directory containing one or more SequenceRead files.
working_directory: A directory where any output files produced by validation can be written.
level: Logging level, defaults to logging.INFO.
Returns:
Currently writes to stderr with a Java Exception trace on error, otherwise no output.
Authors:
Srividya Ramikrishnan, Jason Baumohl, Matt Henderson
"""
if logger is None:
logger = script_utils.stdoutlogger(__file__, level)
# TODO get classpaths and binary paths into the config
KB_TOP = os.environ["KB_TOP"]
fasta_executable = "{}/lib/jars/FastaValidator/FastaValidator-1.0.jar".format(KB_TOP)
fastq_executable = "fastQValidator"
fasta_validator_present = False
fastq_validator_present = False
fastq_validator_runnable = False
if os.path.isfile(fasta_executable):
fasta_validator_present = True
for path in os.environ["PATH"].split(os.pathsep):
path = path.strip('"')
exe_file = os.path.join(path, fastq_executable)
if os.path.isfile(exe_file) and os.access(exe_file, os.X_OK):
fastq_validator_present = True
fastq_validator_runnable = True
break
elif os.path.isfile(exe_file):
fastq_validator_present = True
break
if not fasta_validator_present:
logger.warning("FASTA validator executable FastaValidator-1.0.jar could not be found.")
if not fastq_validator_present:
logger.warning("FASTQ validator executable fastQValidator could not be found.")
elif not fastq_validator_runnable:
logger.warning("FASTQ validator executable fastQValidator does not have execute permissions.")
fasta_extensions = [".fa",".fas",".fasta",".fna"]
fastq_extensions = [".fq",".fastq",".fnq"]
extensions = fasta_extensions + fastq_extensions
checked = False
validated = True
for input_file_name in os.listdir(input_directory):
logger.info("Checking for SequenceReads file : {0}".format(input_file_name))
filePath = os.path.abspath(os.path.join(input_directory, input_file_name))
if not os.path.isfile(filePath):
logger.warning("Skipping directory {0}".format(input_file_name))
continue
elif os.path.splitext(input_file_name)[-1] not in extensions:
logger.warning("Unrecognized file type {}, skipping.".format(os.path.splitext(input_file_name)[-1]))
continue
logger.info("Starting SequenceReads validation of {0}".format(input_file_name))
if os.path.splitext(input_file_name)[-1] in fasta_extensions:
# TODO This needs to be changed, this is really just a demo program for this library and not a serious tool
arguments = ["java", "-classpath", fasta_executable, "FVTester", "'{}'".format(filePath)]
elif os.path.splitext(input_file_name)[-1] in fastq_extensions:
logger.info("Checking FASTQ line count for errors.")
line_number = 0
with open(filePath, 'rb') as seqfile:
for line in seqfile:
line_number += 1
logger.info("FASTQ line count check completed.")
if line_number % 4 > 0:
logger.warning("Found extra lines, removing blank lines.")
out = open(filePath + ".tmp", 'w')
with open(filePath, 'r') as seqfile:
for line in seqfile:
if len(line.strip()) == 0:
pass
out.write(line)
out.close()
os.remove(filePath)
os.rename(filePath + ".tmp", filePath)
logger.warning("Blank lines removed from FASTQ.")
arguments = [fastq_executable, "--file", "'{}'".format(filePath), "--maxErrors", "10"]
if (check_interleavedPE(filePath) == 1):
arguments.append("--disableSeqIDCheck")
logger.info("Running {}".format(" ".join(arguments).replace(filePath, input_file_name)))
tool_process = subprocess.Popen(" ".join(arguments), shell=True)
tool_process.wait()
if tool_process.returncode != 0:
logger.error("Validation failed on {0}".format(input_file_name))
validated = False
break
else:
logger.info("Validation passed on {0}".format(input_file_name))
checked = True
if not validated:
raise Exception("Validation failed!")
elif not checked:
raise Exception("No files were found that had a valid fasta or fastq extension.")
else:
logger.info("Validation passed.")
raise Exception("Validation failed!")
else:
logger.info("Validation passed.")
if __name__ == "__main__":
script_details = script_utils.parse_docs(validate.__doc__)
import argparse
parser = argparse.ArgumentParser(prog=__file__,
description=script_details["Description"],
epilog=script_details["Authors"])
parser.add_argument("--input_file_name", help=script_details["Args"]["input_file_name"], type=str, nargs="?", required=True)
parser.add_argument("--working_directory", help=script_details["Args"]["working_directory"], type=str, nargs="?", required=True)
args = parser.parse_args()
logger = script_utils.stdoutlogger(__file__)
try:
validate(input_file_name = args.input_file_name,
working_directory = args.working_directory,
level = logging.DEBUG,
logger = logger)
except Exception, e:
logger.exception(e)
sys.exit(1)
sys.exit(0)
import argparse
parser = argparse.ArgumentParser(prog=__file__,
description=script_details["Description"],
epilog=script_details["Authors"])
parser.add_argument("--input_file_name",
help=script_details["Args"]["input_file_name"],
type=str,
nargs="?",
required=True)
parser.add_argument("--working_directory",
help=script_details["Args"]["working_directory"],
type=str,
nargs="?",
required=True)
args = parser.parse_args()
logger = script_utils.stdoutlogger(__file__)
try:
validate(input_file_name=args.input_file_name,
working_directory=args.working_directory,
level=logging.DEBUG,
logger=logger)
except Exception, e:
logger.exception(e)
sys.exit(1)
sys.exit(0)
def validate(input_directory,
working_directory,
level=logging.INFO,
logger=None):
"""
Validates any file containing sequence data.
Args:
input_directory: A directory containing one or more SequenceRead files.
working_directory: A directory where any output files produced by validation can be written.
level: Logging level, defaults to logging.INFO.
Returns:
Currently writes to stderr with a Java Exception trace on error, otherwise no output.
Authors:
Srividya Ramikrishnan, Jason Baumohl, Matt Henderson
"""
if logger is None:
logger = script_utils.stdoutlogger(__file__, level)
# TODO get classpaths and binary paths into the config
KB_TOP = os.environ["KB_TOP"]
fasta_executable = "{}/lib/jars/FastaValidator/FastaValidator-1.0.jar".format(
KB_TOP)
fastq_executable = "fastQValidator"
fasta_validator_present = False
fastq_validator_present = False
fastq_validator_runnable = False
if os.path.isfile(fasta_executable):
fasta_validator_present = True
for path in os.environ["PATH"].split(os.pathsep):
path = path.strip('"')
exe_file = os.path.join(path, fastq_executable)
if os.path.isfile(exe_file) and os.access(exe_file, os.X_OK):
fastq_validator_present = True
fastq_validator_runnable = True
break
elif os.path.isfile(exe_file):
fastq_validator_present = True
break
if not fasta_validator_present:
logger.warning(
"FASTA validator executable FastaValidator-1.0.jar could not be found."
)
if not fastq_validator_present:
logger.warning(
"FASTQ validator executable fastQValidator could not be found.")
elif not fastq_validator_runnable:
logger.warning(
"FASTQ validator executable fastQValidator does not have execute permissions."
)
fasta_extensions = [".fa", ".fas", ".fasta", ".fna"]
fastq_extensions = [".fq", ".fastq", ".fnq"]
extensions = fasta_extensions + fastq_extensions
checked = False
validated = True
for input_file_name in os.listdir(input_directory):
logger.info(
"Checking for SequenceReads file : {0}".format(input_file_name))
filePath = os.path.abspath(
os.path.join(input_directory, input_file_name))
if not os.path.isfile(filePath):
logger.warning("Skipping directory {0}".format(input_file_name))
continue
elif os.path.splitext(input_file_name)[-1] not in extensions:
logger.warning("Unrecognized file type {}, skipping.".format(
os.path.splitext(input_file_name)[-1]))
continue
logger.info(
"Starting SequenceReads validation of {0}".format(input_file_name))
if os.path.splitext(input_file_name)[-1] in fasta_extensions:
# TODO This needs to be changed, this is really just a demo program for this library and not a serious tool
arguments = [
"java", "-classpath", fasta_executable, "FVTester",
"'{}'".format(filePath)
]
elif os.path.splitext(input_file_name)[-1] in fastq_extensions:
logger.info("Checking FASTQ line count for errors.")
line_number = 0
with open(filePath, 'rb') as seqfile:
for line in seqfile:
line_number += 1
logger.info("FASTQ line count check completed.")
if line_number % 4 > 0:
logger.warning("Found extra lines, removing blank lines.")
out = open(filePath + ".tmp", 'w')
with open(filePath, 'r') as seqfile:
for line in seqfile:
if len(line.strip()) == 0:
pass
out.write(line)
out.close()
os.remove(filePath)
os.rename(filePath + ".tmp", filePath)
logger.warning("Blank lines removed from FASTQ.")
arguments = [
fastq_executable, "--file", "'{}'".format(filePath),
"--maxErrors", "10"
]
if (check_interleavedPE(filePath) == 1):
arguments.append("--disableSeqIDCheck")
logger.info("Running {}".format(" ".join(arguments).replace(
filePath, input_file_name)))
tool_process = subprocess.Popen(" ".join(arguments), shell=True)
tool_process.wait()
if tool_process.returncode != 0:
logger.error("Validation failed on {0}".format(input_file_name))
validated = False
break
else:
logger.info("Validation passed on {0}".format(input_file_name))
checked = True
if not validated:
raise Exception("Validation failed!")
elif not checked:
raise Exception(
"No files were found that had a valid fasta or fastq extension.")
else:
logger.info("Validation passed.")
