def __init__(self, bam, sample_info_file,
sample_id, output_sh,
genome, walltime,
submit=False):
"""
Parameters
----------
Returns
-------
Raises
------
"""
self.sample_info_file = sample_info_file
if self.sample_info_file is not None:
sample_info = pd.read_table(self.sample_info_file, header=None)
self.bams = sample_info[0]
self.sample_ids = sample_info[1]
self.sh_files = ['{}.miso.sh'.format(bam) for bam in self.bams]
self.multiple_samples = True
else:
self.sample_ids = [sample_id]
self.bams = [bam]
self.sh_files = [output_sh]
self.multiple_samples = False
self.genome = genome
self.walltime = walltime
self.submit = submit
all_samples_commands = []
for bam, sample_id, sh_file in zip(self.bams, self.sample_ids,
self.sh_files):
self._write_single_sample(bam, sample_id, sh_file)
sh_command = 'bash {}'.format(sh_file)
if self.submit and not self.multiple_samples:
commands = [sh_command]
sub = Submitter(commands, job_name='miso',
sh_filename='{}.qsub.sh'.format(sh_file),
ppn=16, walltime=self.walltime)
sub.job(submit=self.submit)
if self.multiple_samples:
all_samples_commands.append(sh_command)
if self.multiple_samples:
sub = Submitter(all_samples_commands, job_name='miso',
sh_filename='miso.qsub.sh',
array=True, ppn=16, walltime=self.walltime)
sub.job(submit=self.submit)
def __init__(self, genomeDir, genomeFastaFiles, sjdb,
sjdbOverhang, job_name, out_sh=None, submit=True):
"""Any CamelCase here is directly copied from the STAR inputs for
complete compatibility
"""
# Make the directory if it's not there already
try:
os.mkdir(genomeDir)
except OSError:
pass
commands = []
commands.append('STAR --runMode genomeGenerate --genomeDir {0} '
'--genomeFastaFiles {1} --runThreadN 16 {2} '
'--sjdbOverhang {3}'.format(
genomeDir, genomeFastaFiles, sjdb, sjdbOverhang))
sub = Submitter(queue_type='PBS', sh_filename=out_sh,
commands=commands,
job_name=job_name, nodes=1, ppn=16, queue='home',
walltime='4:00:00')
sub.job(submit=submit)
def test_pbs(self):
"""Test PBS queue (TSCC)
"""
job_name = 'test_qtools_submitter_pbs'
submit_sh = '{}/{}.sh'.format(self.out_dir, job_name)
sub = Submitter(queue_type='PBS',
sh_filename=submit_sh,
commands=self.commands,
job_name=job_name,
nodes=1,
ppn=1,
queue='home-yeo',
walltime='0:01:00')
job_id = sub.job(submit=False)
true_result_string = '''#!/bin/bash
#PBS -N test_qtools_submitter_pbs
#PBS -o {0}/test_qtools_submitter_pbs.sh.out
#PBS -e {0}/test_qtools_submitter_pbs.sh.err
#PBS -V
#PBS -l walltime=0:01:00
#PBS -l nodes=1:ppn=1
#PBS -A yeo-group
#PBS -q home-yeo
# Go to the directory from which the script was called
cd $PBS_O_WORKDIR
date
echo testing
'''.format(self.out_dir)
true_result = true_result_string.split('\n')
# with open(submit_sh) as f:
# for x in f.readlines():
# print x,
for true, test in zip(true_result, open(submit_sh)):
self.assertEqual(true.strip().split(), test.strip().split())
# Make sure the job ID is a single (potentially multi-digit) integer
# But only do this if we're on TSCC or oolite
if ON_SERVER:
self.assertRegexpMatches(job_id, '^\d+$')
subprocess.Popen(["qdel", job_id], stdout=PIPE)
def test_pbs(self):
"""Test PBS queue (TSCC)
"""
job_name = 'test_qtools_submitter_pbs'
submit_sh = '{}/{}.sh'.format(self.out_dir, job_name)
sub = Submitter(queue_type='PBS', sh_filename=submit_sh,
commands=self.commands,
job_name=job_name, nodes=1, ppn=1,
queue='home-yeo', walltime='0:01:00'
)
job_id = sub.job(submit=False)
true_result_string = '''#!/bin/bash
#PBS -N test_qtools_submitter_pbs
#PBS -o {0}/test_qtools_submitter_pbs.sh.out
#PBS -e {0}/test_qtools_submitter_pbs.sh.err
#PBS -V
#PBS -l walltime=0:01:00
#PBS -l nodes=1:ppn=1
#PBS -A yeo-group
#PBS -q home-yeo
# Go to the directory from which the script was called
cd $PBS_O_WORKDIR
date
echo testing
'''.format(self.out_dir)
true_result = true_result_string.split('\n')
# with open(submit_sh) as f:
# for x in f.readlines():
# print x,
for true, test in zip(true_result, open(submit_sh)):
self.assertEqual(true.strip().split(), test.strip().split())
# Make sure the job ID is a single (potentially multi-digit) integer
# But only do this if we're on TSCC or oolite
if ON_SERVER:
self.assertRegexpMatches(job_id, '^\d+$')
subprocess.Popen(["qdel", job_id],
stdout=PIPE)
def test_sge(self):
"""Test SGE queue (oolite)
"""
job_name = 'test_qtools_submitter_sge'
submit_sh = '{}/{}.sh'.format(self.out_dir, job_name)
sub = Submitter(queue_type='SGE',
sh_filename=submit_sh,
commands=self.commands,
job_name=job_name,
nodes=1,
ppn=1,
queue='home-yeo',
walltime='0:01:00')
job_id = sub.job(submit=False)
true_result_string = '''#!/bin/bash
#$ -N test_qtools_submitter_sge
#$ -o {0}/test_qtools_submitter_sge.sh.out
#$ -e {0}/test_qtools_submitter_sge.sh.err
#$ -V
#$ -S /bin/bash
#$ -cwd
#$ -l bigmem
#$ -l h_vmem=16G
date
echo testing
'''.format(self.out_dir)
true_result = true_result_string.split('\n')
# with open(submit_sh) as f:
# for x in f.readlines():
# print x,
for true, test in zip(true_result, open(submit_sh)):
self.assertEqual(true.strip().split(), test.strip().split())
# Make sure the job ID is a single (potentially multi-digit) integer
# But only do this if we're on TSCC or oolite
if ON_SERVER:
self.assertRegexpMatches(job_id, '^\d+$')
subprocess.Popen(["qdel", job_id], stdout=PIPE)
def test_sge(self):
"""Test SGE queue (oolite)
"""
job_name = 'test_qtools_submitter_sge'
submit_sh = '{}/{}.sh'.format(self.out_dir, job_name)
sub = Submitter(queue_type='SGE', sh_filename=submit_sh,
commands=self.commands,
job_name=job_name, nodes=1, ppn=1,
queue='home-yeo', walltime='0:01:00'
)
job_id = sub.job(submit=False)
true_result_string = '''#!/bin/bash
#$ -N test_qtools_submitter_sge
#$ -o {0}/test_qtools_submitter_sge.sh.out
#$ -e {0}/test_qtools_submitter_sge.sh.err
#$ -V
#$ -S /bin/bash
#$ -cwd
#$ -l bigmem
#$ -l h_vmem=16G
date
echo testing
'''.format(self.out_dir)
true_result = true_result_string.split('\n')
# with open(submit_sh) as f:
# for x in f.readlines():
# print x,
for true, test in zip(true_result, open(submit_sh)):
self.assertEqual(true.strip().split(), test.strip().split())
# Make sure the job ID is a single (potentially multi-digit) integer
# But only do this if we're on TSCC or oolite
if ON_SERVER:
self.assertRegexpMatches(job_id, '^\d+$')
subprocess.Popen(["qdel", job_id],
stdout=PIPE)
def __init__(self, bam, sample_info_file,
sample_id, output_sh,
genome, walltime, nodes=1, ppn=16,
submit=False, read_length=None):
"""
Parameters
----------
Returns
-------
Raises
------
"""
self.sample_info_file = sample_info_file
if self.sample_info_file is not None:
sample_info = pd.read_table(self.sample_info_file, header=None)
self.bams = sample_info[0]
self.sample_ids = sample_info[1]
self.sh_files = ['{}.miso.sh'.format(bam) for bam in self.bams]
self.multiple_samples = True
else:
self.sample_ids = [sample_id]
self.bams = [bam]
self.sh_files = [output_sh]
self.multiple_samples = False
self.genome = genome
self.walltime = walltime
self.submit = submit
self.nodes = nodes
self.ppn = ppn
self.read_length = self.read_length
all_samples_commands = []
for bam, sample_id, sh_file in zip(self.bams, self.sample_ids,
self.sh_files):
self._write_single_sample(bam, sample_id, sh_file)
sh_command = 'bash {}'.format(sh_file)
if self.submit and not self.multiple_samples:
commands = [sh_command]
sub = Submitter(commands, job_name='miso',
sh_filename='{}.qsub.sh'.format(sh_file),
ppn=self.ppn, nodes=self.nodes,
walltime=self.walltime)
sub.job(submit=self.submit)
if self.multiple_samples:
all_samples_commands.append(sh_command)
if self.multiple_samples:
sub = Submitter(all_samples_commands, job_name='miso',
sh_filename='miso.qsub.sh',
array=True,
ppn=self.ppn, nodes=self.nodes,
walltime=self.walltime)
sub.job(submit=self.submit)
print species
except:
print "usage: submit_parse_oldsplice.py <species>"
assert (species != None) and (len(species) > 0)
for filename in files:
filenames.append(filename)
sample = filename.replace(".splices", "").replace(".flip", "_flip")
samples.append(sample)
from gscripts.qtools import Submitter
sub = Submitter()
cmd = "parse_oldsplice.py --species %s" %species
for filename, sample in zip(filenames, samples):
cmd += " --sample %s %s " %(filename, sample)
#print cmd
cmd = [cmd]
sub.job(command_list=cmd, array=False, sh_file="parse.sh", job_name="parse", submit=True, queue="home", ppn=1)
