def __init__(self):
if self.task_name is None:
raise Exception("{}.task_name must by set".format(
self.__class__.__name__))
self.directory()
self.context = Context.instance()
self.sample = self.context.sample
self.config = self.context.config
self._prepare()
def _prepare(self):
self.mode = self.config['phigaro']['mode']
self.finder = V2Finder(
window_len=self.config['phigaro']['window_len'],
threshold_min=self.config['phigaro']['threshold_min_%s' %
self.mode],
threshold_max=self.config['phigaro']['threshold_max_%s' %
self.mode],
mode=self.mode)
self._print_vogs = self.config['phigaro'].get('print_vogs', False)
self._filename = self.config['phigaro'].get('filename', False)
self._no_html = self.config['phigaro'].get('no_html', False)
self._not_open = self.config['phigaro'].get('not_open', False)
self._save_fasta = self.config['phigaro'].get('save_fasta', False)
self._output = self.config['phigaro'].get('output', False)
self._uuid = self.config['phigaro'].get('uuid', False)
self.context = Context.instance()
def main():
default_config_path = join(os.getenv('HOME'), '.phigaro', 'config.yml')
parser = argparse.ArgumentParser(
prog='phigaro',
description=
'Phigaro is a scalable command-line tool for predictions phages and prophages '
'from nucleid acid sequences',
)
parser.add_argument(
'-V',
'--version',
action='version',
version='%(prog)s {version}'.format(version=__version__))
parser.add_argument(
'-f',
'--fasta-file',
help='Assembly scaffolds/contigs or full genomes, required',
required=True)
parser.add_argument('-c',
'--config',
default=default_config_path,
help='Config file, not required')
parser.add_argument('-v',
'--verbose',
action='store_true',
help=argparse.SUPPRESS)
parser.add_argument('-p',
'--print-vogs',
help='Print phage vogs for each region',
action='store_true')
parser.add_argument(
'-e',
'--extension',
default=['html'],
nargs='+',
help=
'Type of the output: html, txt or stdout. Default is html. You can specify several file formats with a space as a separator. Example: -e txt html stdout.'
)
parser.add_argument(
'-o',
'--output',
default=False,
help=
'Output filename for html and txt outputs. Required by default, but not required for stdout only output.'
)
parser.add_argument(
'--not-open',
help=
'Do not open html file automatically, if html output type is specified.',
action='store_true')
parser.add_argument('-t',
'--threads',
type=int,
default=multiprocessing.cpu_count(),
help='Num of threads ('
'default is num of CPUs={})'.format(
multiprocessing.cpu_count()))
parser.add_argument('--no-cleanup',
action='store_true',
help=argparse.SUPPRESS)
parser.add_argument('-S',
'--substitute-output',
action='append',
help=argparse.SUPPRESS)
args = parser.parse_args()
logging.basicConfig(level=logging.INFO if args.verbose else logging.WARN)
logging.getLogger('sh.command').setLevel(logging.WARN)
logger = logging.getLogger(__name__)
if not exists(args.config):
# TODO: pretty message
print('Please, create config file using phigaro-setup script')
exit(1)
args.extension = [atype.lower() for atype in args.extension]
if (not args.output) and (args.extension != ['stdout']):
print(
'Error! Argument -o/--output is required or change the type of the output to stdout.'
)
exit(1)
with open(args.config) as f:
logger.info('Using config file: {}'.format(args.config))
config = yaml.load(f, Loader=yaml.FullLoader)
config['phigaro']['print_vogs'] = args.print_vogs
config['phigaro']['filename'] = args.fasta_file
config['phigaro'][
'no_html'] = True if 'html' not in args.extension else False
config['phigaro']['not_open'] = args.not_open
config['phigaro']['output'] = args.output
config['phigaro']['uuid'] = uuid.uuid4().hex
filename = args.fasta_file
sample = '{}-{}'.format(sample_name(filename), config['phigaro']['uuid'])
Context.initialize(
sample=sample,
config=config,
threads=args.threads,
)
substitutions = parse_substitute_output(args.substitute_output)
preprocess_task = create_task(substitutions, PreprocessTask, filename)
prodigal_task = create_task(substitutions,
ProdigalTask,
preprocess_task=preprocess_task)
hmmer_task = create_task(substitutions,
HmmerTask,
prodigal_task=prodigal_task)
run_phigaro_task = create_task(substitutions,
RunPhigaroTask,
prodigal_task=prodigal_task,
hmmer_task=hmmer_task)
tasks = [preprocess_task, prodigal_task, hmmer_task, run_phigaro_task]
task_output_file = run_tasks_chain(tasks)
if ('txt' in args.extension) or ('stdout' in args.extension):
with open(task_output_file) as f:
if 'txt' in args.extension:
out_f = open(args.output, 'w')
for line in f:
out_f.write(line)
if 'stdout' in args.extension:
out_f = sys.stdout
for line in f:
out_f.write(line)
out_f.close()
if not args.no_cleanup:
for t in tasks:
t.clean()
clean_fold()
def main():
default_config_path = join(os.getenv('HOME'), '.phigaro', 'config.yml')
parser = argparse.ArgumentParser(
prog='phigaro',
description=
'Phigaro is a scalable command-line tool for predictions phages and prophages '
'from nucleid acid sequences',
)
parser.add_argument(
'-V',
'--version',
action='version',
version='%(prog)s {version}'.format(version=__version__),
)
parser.add_argument(
'-f',
'--fasta-file',
help='Assembly scaffolds/contigs or full genomes, required',
required=True,
)
parser.add_argument(
'-c',
'--config',
default=default_config_path,
help='Path to the config file, not required. The deafult is %s' %
default_config_path,
)
parser.add_argument('-v',
'--verbose',
action='store_true',
help=argparse.SUPPRESS)
parser.add_argument(
'-p',
'--print-vogs',
help='Print phage vogs for each region',
action='store_true',
)
parser.add_argument(
'-e',
'--extension',
default=['html'],
nargs='+',
help=
'Type of the output: html, tsv, gff, bed or stdout. Default is html. You can specify several file formats with a space as a separator. Example: -e tsv html stdout.',
)
parser.add_argument(
'-o',
'--output',
default='',
help=
'Output filename for html and txt outputs. Required by default, but not required for stdout only output.',
)
parser.add_argument(
'--not-open',
help=
'Do not open html file automatically, if html output type is specified.',
action='store_true',
)
parser.add_argument(
'-t',
'--threads',
type=int,
default=multiprocessing.cpu_count(),
help='Num of threads ('
'default is num of CPUs={})'.format(multiprocessing.cpu_count()),
)
parser.add_argument(
'--no-cleanup',
action='store_true',
help=
"Do not delete any temporary files that was generated by Phigaro (HMMER & Prodigal outputs and some others)."
)
parser.add_argument(
'-S',
'--substitute-output',
action='append',
help=
'If you have precomputed prodigal and/or hmmer data you can provide paths to the files in the following format: program:address/to/the/file. In place of program you should write hmmer or prodigal. If you need to provide both files you should pass them separetely as two parametres.',
)
parser.add_argument(
'--save-fasta',
action='store_true',
help='Save all phage fasta sequences in a fasta file.',
)
parser.add_argument(
'-d',
'--delete-shorts',
action='store_true',
help='Exclude sequences with length < 20000 automatically.',
)
parser.add_argument(
'-m',
'--mode',
default='basic',
help=
'You can launch Phigaro at one of 3 modes: basic, abs, without_gc. Default is basic. Read more about modes at https://github.com/bobeobibo/phigaro/',
)
parser.add_argument('--wtp', action='store_true', help=argparse.SUPPRESS)
args = parser.parse_args()
logging.basicConfig(level=logging.INFO if args.verbose else logging.WARN)
logging.getLogger('sh.command').setLevel(logging.WARN)
logger = logging.getLogger(__name__)
if not exists(args.config):
# TODO: pretty message
print('Please, create config file using phigaro-setup script')
exit(1)
args.extension = [atype.lower() for atype in args.extension]
for ext in args.extension:
if ext not in ['html', 'gff', 'bed', 'tsv', 'stdout']:
print(
'Error! The unknown output format in -e/--extensionn parameter: %s. Please, choose one or several from the list: html, gff, bed, tsv, stdout'
% ext)
exit(1)
if (args.output == '') and (args.extension != ['stdout']):
print(
'Error! Argument -o/--output is required or change the type of the output to stdout.'
)
exit(1)
with open(args.config) as f:
logger.info('Using config file: {}'.format(args.config))
config = yaml.load(f, Loader=yaml.FullLoader)
config['phigaro']['wtp'] = args.wtp
config['phigaro']['print_vogs'] = args.print_vogs
config['phigaro']['filename'] = args.fasta_file
config['phigaro']['no_html'] = (True
if 'html' not in args.extension else False)
config['phigaro']['not_open'] = args.not_open
config['phigaro']['output'] = (args.output + '/' + os.path.splitext(
os.path.basename(args.fasta_file))[0] + '.phigaro').replace('//', '/')
config['phigaro']['uuid'] = uuid.uuid4().hex
config['phigaro']['delete_shorts'] = args.delete_shorts
config['phigaro']['gff'] = True if ('gff' in args.extension) else False
config['phigaro']['bed'] = True if ('bed' in args.extension) else False
config['phigaro']['mode'] = args.mode
config['phigaro']['save_fasta'] = args.save_fasta
filename = args.fasta_file
sample = '{}-{}'.format(sample_name(filename), config['phigaro']['uuid'])
if args.wtp:
config['phigaro']['not_open'] = True
config['phigaro']['gff'] = True
config['phigaro']['bed'] = True
args.extension.append('tsv')
config['phigaro']['delete_shorts'] = True
config['phigaro']['print_vogs'] = True
config['phigaro']['output_wtp'] = args.output + '/phigaro.txt'
config['phigaro']['output'] = args.output + '/phigaro/phigaro'
config['phigaro']['save_fasta'] = True
if config['phigaro']['output'] != '':
fold = os.path.dirname(config['phigaro']['output'])
if fold and not os.path.isdir(fold):
os.makedirs(fold)
if args.wtp:
fold = os.path.dirname(config['phigaro']['output_wtp'])
if fold and not os.path.isdir(fold):
os.makedirs(fold)
Context.initialize(
sample=sample,
config=config,
threads=args.threads,
)
substitutions = parse_substitute_output(args.substitute_output)
preprocess_task = create_task(substitutions, PreprocessTask, filename)
prodigal_task = create_task(substitutions,
ProdigalTask,
preprocess_task=preprocess_task)
hmmer_task = create_task(substitutions,
HmmerTask,
prodigal_task=prodigal_task)
run_phigaro_task = create_task(
substitutions,
RunPhigaroTask,
prodigal_task=prodigal_task,
hmmer_task=hmmer_task,
)
tasks = [preprocess_task, prodigal_task, hmmer_task, run_phigaro_task]
task_output_file = run_tasks_chain(tasks)
if ('tsv' in args.extension) or ('stdout' in args.extension):
with open(task_output_file) as f:
f = list(f)
if 'tsv' in args.extension:
out_f = open(config['phigaro']['output'] + '.tsv', 'w')
for line in f:
out_f.write(line)
if 'stdout' in args.extension:
out_f = sys.stdout
for line in f:
out_f.write(line)
out_f.close()
if not args.no_cleanup:
for t in tasks:
t.clean()
clean_fold()
def __init__(self, input):
super().__init__()
self.input = input
self.context = Context.instance()
def main():
default_config_path = join(os.getenv('HOME'), '.phigaro', 'config.yml')
parser = argparse.ArgumentParser(
prog='phigaro',
description=
'Phigaro is a scalable command-line tool for predictions phages and prophages '
'from nucleid acid sequences',
)
parser.add_argument(
'-V',
'--version',
action='version',
version='%(prog)s {version}'.format(version=__version__))
parser.add_argument(
'-f',
'--fasta-file',
help='Assembly scaffolds/contigs or full genomes, required',
required=True)
parser.add_argument('-c',
'--config',
default=default_config_path,
help='Config file, not required')
parser.add_argument('-v',
'--verbose',
action='store_true',
help=argparse.SUPPRESS)
parser.add_argument('-o',
'--output',
help='Output file, not required, default is stdout')
parser.add_argument('-p',
'--print-vogs',
help='Print phage vogs for each region',
action='store_true')
parser.add_argument('-t',
'--threads',
type=int,
default=multiprocessing.cpu_count(),
help='Num of threads ('
'default is num of CPUs={})'.format(
multiprocessing.cpu_count()))
parser.add_argument('--no-cleanup',
action='store_true',
help=argparse.SUPPRESS)
parser.add_argument('-S',
'--substitute-output',
action='append',
help=argparse.SUPPRESS)
args = parser.parse_args()
logging.basicConfig(level=logging.INFO if args.verbose else logging.WARN)
logging.getLogger('sh.command').setLevel(logging.WARN)
logger = logging.getLogger(__name__)
if not exists(args.config):
# TODO: pretty message
print('Please create config file using phigaro-setup script')
exit(1)
with open(args.config) as f:
logger.info('Using config file: {}'.format(args.config))
config = yaml.load(f)
config['phigaro']['print_vogs'] = args.print_vogs
filename = args.fasta_file
sample = '{}-{}'.format(sample_name(filename), uuid.uuid4().hex)
Context.initialize(
sample=sample,
config=config,
threads=args.threads,
)
substitutions = parse_substitute_output(args.substitute_output)
gene_mark_task = create_task(substitutions, GeneMarkTask, filename)
hmmer_task = create_task(substitutions,
HmmerTask,
gene_mark_task=gene_mark_task)
run_phigaro_task = create_task(substitutions,
RunPhigaroTask,
gene_mark_task=gene_mark_task,
hmmer_task=hmmer_task)
tasks = [gene_mark_task, hmmer_task, run_phigaro_task]
task_output_file = run_tasks_chain(tasks)
with open(task_output_file) as f:
out_f = open(args.output, 'w') if args.output else sys.stdout
for line in f:
out_f.write(line)
if out_f is sys.stdout:
out_f.close()
if not args.no_cleanup:
for t in tasks:
t.clean()