def run(self):
while True:
try:
self.waiting()
self.launch()
self.update_runstats()
except exceptions.RerunnableError as e:
logger.warn("[%s] A job needs to be rerun: %s" % (self.name, e))
self.update_runstats(1)
except exceptions.FatalError as e:
logger.error("[%s] A fatal error occurred: %s" % (self.name, e))
os.kill(self.ppid, signal.SIGINT)
except (KeyboardInterrupt, SystemExit):
logger.debug("Process interrupted")
except Exception as e:
ex_type, ex, tb = sys.exc_info()
logger.error("\n".join(traceback.format_exception(ex_type, ex, tb)))
logger.error("An unhandled exception occurred")
os.kill(self.ppid, signal.SIGINT)
def warn(self, msg):
if logger.level() == logging.DEBUG:
name = self.name
else:
name = self.__class__.__name__
logger.warn("%-12s| %s" % (name, msg))
def RunNewbler(self):
#Code for running newbler
"""
Expects params keys:
PE1 and PE2 and/or SE
target_dir
-urt
"""
#Check for necessary params:
if not (('assembly_PE1' in self.params and 'assembly_PE2' in self.params) or 'assembly_SE' in self.params):
raise exceptions.FatalError('Missing self.params in RunNewbler.')
#Check for necessary files:
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params and not(os.path.exists(self.params['assembly_PE1']) or not(os.path.exists(self.params['assembly_PE2']))):
raise exceptions.FatalError('Missing PE files in RunNewbler.')
if 'assembly_SE' in self.params and not(os.path.exists(self.params['assembly_SE'])):
raise exceptions.FatalError('Missing SE file in RunNewbler.')
sample = self.params['sample']
target = self.params['target']
killed = False
failed = False
#determine whether to pipe output to a file or /dev/null
if self.params['verbose']:
out = open(os.path.join(self.params['target_dir'], "assembly.log"), 'w')
else:
out = open(os.devnull, 'w')
#Build args for newAssembly:
args = ['newAssembly', '-force']
if self.params['last_assembly'] and self.params['cdna']:
#only run with cdna switch on the final assembly
args += ['-cdna']
args += [os.path.join(self.params['target_dir'], 'assembly')]
logger.debug("Calling newAssembly for sample: %s target %s" % (sample, target))
logger.info(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
#Build args for addRun:
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params:
args = ['addRun', os.path.join(self.params['target_dir'], 'assembly')]
args += [self.params['assembly_PE1']]
logger.debug("Calling addRun for sample: %s target %s" % (sample, target))
logger.debug(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
args = ['addRun', os.path.join(self.params['target_dir'], 'assembly')]
args += [self.params['assembly_PE2']]
logger.debug("Calling addRun for sample: %s target %s" % (sample, target))
logger.debug(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
if 'assembly_SE' in self.params:
args = ['addRun', os.path.join(self.params['target_dir'], 'assembly')]
args += [self.params['assembly_SE']]
logger.debug("Calling addRun for sample: %s target %s" % (sample, target))
logger.debug(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
#Build args for runProject
args = ['runProject']
args += ['-cpu', '1']
if self.params['last_assembly'] and self.params['cdna']:
args += ['-noace']
else:
args += ['-nobig']
if self.params['urt'] and not self.params['last_assembly']:
#only run with the -urt switch when it isn't the final assembly
args += ['-urt']
if self.params['rip']:
args += ['-rip']
args += [os.path.join(self.params['target_dir'], 'assembly')]
try:
start = time.time()
logger.debug("Calling runProject for sample: %s target %s" % (sample, target))
logger.debug(" ".join(args))
ret = subprocess.Popen(args, stdout=out, stderr=out)
pid = ret.pid
while ret.poll() is None:
if time.time() - start > self.params['assemblytimeout']:
self.kill_process_children(pid)
logger.warn("Sample: %s target: %s iteration: %s Killing assembly after %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
killed = True
break
time.sleep(.5)
except Exception as exc:
txt = "Sample: %s, Target: %s: Unhandeled error running Newbler assembly" % (self.params['sample'], self.params['target'])
txt += '\n\t' + str(exc) + "".join(traceback.format_exception)
logger.warn(txt)
failed = True
pass
finally:
out.close()
#Sometimes newbler doesn't seem to exit completely:
self.kill_process_children(pid)
#if ret != 0:
#raise exceptions.RerunnableError("Newbler assembly failed.")
if not killed and ret.poll() != 0:
#raise exceptions.RerunnableError("Newbler assembly failed.")
failed = True
if failed:
logger.info("Sample: %s target: %s iteration: %s Assembly failed after %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"), 'w')
outf.write("assembly_failed\t" + str(time.time() - start))
outf.close()
if killed:
logger.info("Sample: %s target: %s iteration: %s Assembly killed after %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"), 'w')
outf.write("assembly_killed\t" + str(time.time() - start))
outf.close()
else:
#Run finished without error
logger.info("Sample: %s target: %s iteration: %s Assembly finished in %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"), 'w')
outf.write("assembly_complete\t" + str(time.time() - start))
outf.close()
def RunSpades(self):
"""
Several arguments can be passed to spades.py: -1 [PE1], -2 [PE2], -s [SE], and -o [target_dir]
"""
#Check that required params are available
if not (('assembly_PE1' in self.params and 'assembly_PE2' in self.params) or ('assembly_SE' in self.params)):
raise exceptions.FatalError('Missing self.params in RunSpades.')
#Check that the files actually exist
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params and not(os.path.exists(self.params['assembly_PE1']) or not(os.path.exists(self.params['assembly_PE2']))):
raise exceptions.FatalError('Missing PE files in RunSpades.')
if 'assembly_SE' in self.params and not(os.path.exists(self.params['assembly_SE'])):
raise exceptions.FatalError('Missing SE file in RunSpades.')
sample = self.params['sample']
target = self.params['target']
#Build args for assembler call
args = ['spades.py', '-t', '1']
if self.params['only-assembler'] and not self.params['last_assembly']:
args.append("--only-assembler")
if self.params['format'] == 'fasta':
args.append('--only-assembler') # spades errors on read correction if the input isn't fastq
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params:
args += ['-1', self.params['assembly_PE1'], '-2', self.params['assembly_PE2']]
if 'assembly_SE' in self.params:
args += ['-s', self.params['assembly_SE']]
args += ['-o', os.path.join(self.params['target_dir'], 'assembly')]
if self.params['verbose']:
out = open(os.path.join(self.params['target_dir'], "assembly.log"), 'w')
else:
out = open(os.devnull, 'w')
logger.debug("Sample: %s target: %s Running spades assembler." % (sample, target))
logger.info(" ".join(args))
killed = False
failed = False
start = time.time()
try:
#ret = subprocess.call(args, stderr=out, stdout=out)
ret = subprocess.Popen(args, stdout=out, stderr=out)
pid = ret.pid
while ret.poll() is None:
if time.time() - start > self.params['assemblytimeout']:
ret.kill()
killed = True
logger.warn("Sample: %s target: %s Assembly killed after %s seconds." % (sample, target, time.time() - start))
break
time.sleep(.5)
except Exception as exc:
txt = ("Sample: %s, Target: %s: Unhandeled error running Spades assembly" % (sample, target))
txt += '\n\t' + str(exc)
logger.warn(txt)
failed = True
pass
finally:
out.close()
#Ensure that assembler exits cleanly:
self.kill_process_children(pid)
if not killed and ret.poll() != 0:
failed = True
if failed:
logger.info("Sample: %s target: %s iteration: %s Assembly failed after %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"), 'w')
outf.write("assembly_failed")
outf.close()
elif killed:
logger.info("Sample: %s target: %s iteration: %s Assembly killed after %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"), 'w')
outf.write("assembly_killed")
outf.close()
else:
#Run finished without error
logger.info("Sample: %s target: %s iteration: %s Assembly finished in %s seconds" % (sample, target, self.params['iteration'], time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"), 'w')
outf.write("assembly_complete")
outf.close()
def RunNewbler(self):
#Code for running newbler
"""
Expects params keys:
PE1 and PE2 and/or SE
target_dir
-urt
"""
#Check for necessary params:
if not (
('assembly_PE1' in self.params and 'assembly_PE2' in self.params)
or 'assembly_SE' in self.params):
raise exceptions.FatalError('Missing self.params in RunNewbler.')
#Check for necessary files:
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params and not (
os.path.exists(self.params['assembly_PE1'])
or not (os.path.exists(self.params['assembly_PE2']))):
raise exceptions.FatalError('Missing PE files in RunNewbler.')
if 'assembly_SE' in self.params and not (os.path.exists(
self.params['assembly_SE'])):
raise exceptions.FatalError('Missing SE file in RunNewbler.')
sample = self.params['sample']
target = self.params['target']
killed = False
failed = False
#determine whether to pipe output to a file or /dev/null
if self.params['verbose']:
out = open(os.path.join(self.params['target_dir'], "assembly.log"),
'w')
else:
out = open(os.devnull, 'w')
#Build args for newAssembly:
args = ['newAssembly', '-force']
if self.params['last_assembly'] and self.params['cdna']:
#only run with cdna switch on the final assembly
args += ['-cdna']
args += [os.path.join(self.params['target_dir'], 'assembly')]
logger.debug("Calling newAssembly for sample: %s target %s" %
(sample, target))
logger.info(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
#Build args for addRun:
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params:
args = [
'addRun',
os.path.join(self.params['target_dir'], 'assembly')
]
args += [self.params['assembly_PE1']]
logger.debug("Calling addRun for sample: %s target %s" %
(sample, target))
logger.debug(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
args = [
'addRun',
os.path.join(self.params['target_dir'], 'assembly')
]
args += [self.params['assembly_PE2']]
logger.debug("Calling addRun for sample: %s target %s" %
(sample, target))
logger.debug(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
if 'assembly_SE' in self.params:
args = [
'addRun',
os.path.join(self.params['target_dir'], 'assembly')
]
args += [self.params['assembly_SE']]
logger.debug("Calling addRun for sample: %s target %s" %
(sample, target))
logger.debug(" ".join(args))
ret = subprocess.call(args, stdout=out, stderr=out)
#Build args for runProject
args = ['runProject']
args += ['-cpu', '1']
if self.params['last_assembly'] and self.params['cdna']:
args += ['-noace']
else:
args += ['-nobig']
if self.params['urt'] and not self.params['last_assembly']:
#only run with the -urt switch when it isn't the final assembly
args += ['-urt']
if self.params['rip']:
args += ['-rip']
args += [os.path.join(self.params['target_dir'], 'assembly')]
try:
start = time.time()
logger.debug("Calling runProject for sample: %s target %s" %
(sample, target))
logger.debug(" ".join(args))
ret = subprocess.Popen(args, stdout=out, stderr=out)
pid = ret.pid
while ret.poll() is None:
if time.time() - start > self.params['assemblytimeout']:
self.kill_process_children(pid)
logger.warn(
"Sample: %s target: %s iteration: %s Killing assembly after %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
killed = True
break
time.sleep(.5)
except Exception as exc:
txt = "Sample: %s, Target: %s: Unhandeled error running Newbler assembly" % (
self.params['sample'], self.params['target'])
txt += '\n\t' + str(exc) + "".join(traceback.format_exception)
logger.warn(txt)
failed = True
pass
finally:
out.close()
#Sometimes newbler doesn't seem to exit completely:
self.kill_process_children(pid)
#if ret != 0:
#raise exceptions.RerunnableError("Newbler assembly failed.")
if not killed and ret.poll() != 0:
#raise exceptions.RerunnableError("Newbler assembly failed.")
failed = True
if failed:
logger.info(
"Sample: %s target: %s iteration: %s Assembly failed after %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"),
'w')
outf.write("assembly_failed\t" + str(time.time() - start))
outf.close()
if killed:
logger.info(
"Sample: %s target: %s iteration: %s Assembly killed after %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"),
'w')
outf.write("assembly_killed\t" + str(time.time() - start))
outf.close()
else:
#Run finished without error
logger.info(
"Sample: %s target: %s iteration: %s Assembly finished in %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"),
'w')
outf.write("assembly_complete\t" + str(time.time() - start))
outf.close()
def RunSpades(self):
"""
Several arguments can be passed to spades.py: -1 [PE1], -2 [PE2], -s [SE], and -o [target_dir]
"""
#Check that required params are available
if not (('assembly_PE1' in self.params
and 'assembly_PE2' in self.params) or
('assembly_SE' in self.params)):
raise exceptions.FatalError('Missing self.params in RunSpades.')
#Check that the files actually exist
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params and not (
os.path.exists(self.params['assembly_PE1'])
or not (os.path.exists(self.params['assembly_PE2']))):
raise exceptions.FatalError('Missing PE files in RunSpades.')
if 'assembly_SE' in self.params and not (os.path.exists(
self.params['assembly_SE'])):
raise exceptions.FatalError('Missing SE file in RunSpades.')
sample = self.params['sample']
target = self.params['target']
#Build args for assembler call
args = ['spades.py', '-t', '1']
if self.params['only-assembler'] and not self.params['last_assembly']:
args.append("--only-assembler")
if self.params['format'] == 'fasta':
args.append(
'--only-assembler'
) # spades errors on read correction if the input isn't fastq
if 'assembly_PE1' in self.params and 'assembly_PE2' in self.params:
args += [
'-1', self.params['assembly_PE1'], '-2',
self.params['assembly_PE2']
]
if 'assembly_SE' in self.params:
args += ['-s', self.params['assembly_SE']]
args += ['-o', os.path.join(self.params['target_dir'], 'assembly')]
if self.params['verbose']:
out = open(os.path.join(self.params['target_dir'], "assembly.log"),
'w')
else:
out = open(os.devnull, 'w')
logger.debug("Sample: %s target: %s Running spades assembler." %
(sample, target))
logger.info(" ".join(args))
killed = False
failed = False
start = time.time()
try:
#ret = subprocess.call(args, stderr=out, stdout=out)
ret = subprocess.Popen(args, stdout=out, stderr=out)
pid = ret.pid
while ret.poll() is None:
if time.time() - start > self.params['assemblytimeout']:
ret.kill()
killed = True
logger.warn(
"Sample: %s target: %s Assembly killed after %s seconds."
% (sample, target, time.time() - start))
break
time.sleep(.5)
except Exception as exc:
txt = (
"Sample: %s, Target: %s: Unhandeled error running Spades assembly"
% (sample, target))
txt += '\n\t' + str(exc)
logger.warn(txt)
failed = True
pass
finally:
out.close()
#Ensure that assembler exits cleanly:
self.kill_process_children(pid)
if not killed and ret.poll() != 0:
failed = True
if failed:
logger.info(
"Sample: %s target: %s iteration: %s Assembly failed after %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"),
'w')
outf.write("assembly_failed")
outf.close()
elif killed:
logger.info(
"Sample: %s target: %s iteration: %s Assembly killed after %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"),
'w')
outf.write("assembly_killed")
outf.close()
else:
#Run finished without error
logger.info(
"Sample: %s target: %s iteration: %s Assembly finished in %s seconds"
% (sample, target, self.params['iteration'],
time.time() - start))
outf = open(os.path.join(self.params['target_dir'], "finished"),
'w')
outf.write("assembly_complete")
outf.close()