def check_dependencies():
try:
# sb.check_call(["samtools", "tview"], stdout=sb.DEVNULL, stderr=sb.DEVNULL)
sb.check_call(["bzip2", "--help"], stdout=sb.DEVNULL, stderr=sb.DEVNULL)
except Exception as e:
error('Program not installed or not present in the system path\n'+str(e),
init_new_line=True, exit=True)
def extract_markers(database, clade, output_dir):
info('\tGenerating DB markers FASTA...', init_new_line=True)
fasta_markers = generate_markers_fasta(database, output_dir)
info('\tDone.', init_new_line=True)
info('\tLoading MetaPhlan ' + __version__ + ' database...',
init_new_line=True)
db = pickle.load(bz2.BZ2File(database))
info('\tDone.', init_new_line=True)
markers = set([])
for marker in db['markers']:
species = db['markers'][marker]['clade']
if clade == species:
markers.add(marker)
if len(markers) == 0:
error("No markers were found for the clade \"" + clade +
"\" in the database",
exit=True,
init_new_line=True)
info('\tNumber of markers for the clade \"' + clade + "\": " +
str(len(markers)),
init_new_line=True)
output_file = output_dir + clade + ".fna"
info('\tExporting markers...', init_new_line=True)
with open(output_file, 'w') as ofile:
for rec in SeqIO.parse(open(fasta_markers, 'r'), 'fasta'):
if rec.name in markers:
SeqIO.write(rec, ofile, 'fasta')
info('\tDone.', init_new_line=True)
os.remove(fasta_markers)
return output_file
def check_dependencies():
try:
sb.check_call("bowtie2-inspect", stdout=sb.DEVNULL, stderr=sb.DEVNULL)
except Exception as e:
error('Program not installed or not present in the system path\n' +
str(e),
init_new_line=True,
exit=True)
def check_input_files(input, input_format):
for s in input:
_, extension = os.path.splitext(s)
if not os.path.exists(s):
error('The input file \"'+s+'\" does not exist', exit=True,
init_new_line=True)
elif not input_format.lower() == extension[1:].lower():
error('The the input file \"'+s+'\" must be in \"'+
input_format.upper()+'\" format',
exit=True, init_new_line=True)
return True
def samples_to_markers(input, sorted, input_format, output_dir, breath_threshold, nprocs):
tmp_dir = output_dir+'tmp/'
try:
os.mkdir(tmp_dir)
except Exception as e:
error('Folder \"'+tmp_dir+'\" already exists!\n'+str(e), exit=True,
init_new_line=True)
input = convert_inputs(input, sorted, input_format, tmp_dir, nprocs)
execute_cmseq(input, output_dir, breath_threshold, nprocs)
shutil.rmtree(tmp_dir, ignore_errors=False, onerror=None)
def execute_pool(args, nprocs):
terminating = Event()
with Pool(initializer=init_terminating,
initargs=(terminating, ),
processes=nprocs) as pool:
try:
return [
_ for _ in pool.imap_unordered(
parallel_execution, args, chunksize=CHUNKSIZE)
]
except Exception as e:
error('Parallel execution fails: ' + str(e),
init_new_line=True,
exit=True)
def execute(cmd):
inp_f = None
out_f = sb.DEVNULL
if cmd['stdin']:
inp_f = open(cmd['stdin'], 'r')
if cmd['stdout']:
out_f = open(cmd['stdout'], 'w')
exec_res = sb.run(cmd['command_line'], stdin=inp_f, stdout=out_f)
if exec_res.returncode == 1:
error("An error was ocurred executing a external tool, exiting...",
init_new_line=True,
exit=True)
if cmd['stdin']:
inp_f.close()
if cmd['stdout']:
out_f.close()
def decompress_from_bz2(input, tmp_dir, nprocs):
decompressed = []
decompressed_format = []
results = execute_pool(((decompress_bz2_file, i, tmp_dir) for i in input),
nprocs)
for r in results:
decompressed.append(r[0])
decompressed_format.append(r[1])
if decompressed_format[1:] == decompressed_format[:-1]:
if decompressed_format[0][1:].lower() == "sam":
return decompressed, "sam"
elif decompressed_format[0][1:].lower() == "bam":
return decompressed, "bam"
else:
error("Decompressed files are not in SAM or BAM format",
exit=True, init_new_line=True)
else:
error("Decompressed files have different formats",
exit=True, init_new_line=True)
def check_params(args):
if not args.tree and not args.dist:
error('-t (or --tree) must be specified', exit=True,
init_new_line=True)
if not args.output_dir:
error('-o (or --output_dir) must be specified', exit=True,
init_new_line=True)
elif not os.path.exists(args.output_dir):
error('The directory {} does not exist'.format(args.output_dir), exit=True,
init_new_line=True)
if not args.metadata:
error('-m (or --metadata) must be specified', exit=True,
init_new_line=True)
def check_params(args):
if not args.input:
error('-i (or --input) must be specified', exit=True,
init_new_line=True)
elif not args.input_format:
error('-f (or --input_format) must be specified', exit=True,
init_new_line=True)
elif not args.output_dir:
error('-o (or --output_dir) must be specified', exit=True,
init_new_line=True)
elif args.input_format.lower() != "bam" and args.input_format.lower() != "sam" and args.input_format.lower() != "bz2":
error('The input format must be SAM, BAM, or compressed in BZ2 format',
exit=True, init_new_line=True)
else:
check_input_files(args.input, args.input_format)
if not args.output_dir.endswith('/'):
args.output_dir += '/'
return args
def check_params(args):
if not args.clade:
error('-c (or --clade) must be specified',
exit=True,
init_new_line=True)
elif not args.output_dir:
error('-o (or --output_dir) must be specified',
exit=True,
init_new_line=True)
elif not os.path.exists(args.output_dir):
error('The directory {} does not exist'.format(args.output_dir),
exit=True,
init_new_line=True)
elif not os.path.exists(args.database):
error('The database does not exist', exit=True, init_new_line=True)
if not args.output_dir.endswith('/'):
args.output_dir += '/'
return args
def check_params(args):
if not args.tree and not args.dist:
error('-t (or --tree) must be specified',
exit=True,
init_new_line=True)
if not args.output_dir:
error('-o (or --output_dir) must be specified',
exit=True,
init_new_line=True)
if not args.metadata:
error('-m (or --metadata) must be specified',
exit=True,
init_new_line=True)
def compose_command(params,
check=False,
input_file=None,
database=None,
output_path=None,
output_file=None,
nproc=1):
program_name = None
stdin = None
stdout = None
environment = os.environ.copy()
r_output_path = None
r_output_file = None
command_line = params['command_line']
if 'program_name' in list(params):
command_line = command_line.replace('#program_name#',
params['program_name'])
program_name = params['program_name']
else:
error('Error: something wrong... ' + program_name + ' not found!',
exit=True)
if check:
command_line = program_name
if 'version' in list(params):
command_line = '{} {}'.format(program_name, params['version'])
else:
if 'params' in list(params):
command_line = command_line.replace('#params#', params['params'])
if 'threads' in list(params):
command_line = command_line.replace(
'#threads#', '{} {}'.format(params['threads'], nproc))
if output_path:
r_output_path = output_path
if 'output_path' in list(params):
command_line = command_line.replace(
'#output_path#', '{} {}'.format(params['output_path'],
output_path))
else:
output_file = os.path.join(output_path, output_file)
if input_file:
inp = input_file
if 'input' in list(params):
inp = '{} {}'.format(params['input'], input_file)
if '<' in command_line:
command_line = command_line.replace('<', '')
command_line = command_line.replace('#input#', '')
stdin = inp
else:
command_line = command_line.replace('#input#', inp)
if database and ('database' in list(params)):
command_line = command_line.replace(
'#database#', '{} {}'.format(params['database'], database))
if output_file:
out = output_file
r_output_file = output_file
if 'output' in list(params):
out = '{} {}'.format(params['output'], output_file)
if '>' in command_line:
command_line = command_line.replace('>', '')
command_line = command_line.replace('#output#', '')
stdout = out
else:
command_line = command_line.replace('#output#', out)
if 'environment' in list(params):
new_environment = dict([
(var.strip(), val.strip()) for var, val in [
a.strip().split('=')
for a in params['environment'].split(',')
]
])
environment.update(new_environment)
# find string sourrunded with " and make them as one string
quotes = [j for j, e in enumerate(command_line) if e == '"']
for s, e in zip(quotes[0::2], quotes[1::2]):
command_line = command_line.replace(
command_line[s + 1:e], command_line[s + 1:e].replace(' ', '#'))
return {
'command_line': [
str(a).replace('#', ' ') for a in re.sub(
' +', ' ', command_line.replace('"', '')).split(' ') if a
],
'stdin':
stdin,
'stdout':
stdout,
'env':
environment,
'output_path':
r_output_path,
'output_file':
r_output_file
}
'Duy Tin Truong ([email protected]), ' 'Francesco Asnicar ([email protected]), ' 'Moreno Zolfo ([email protected]), ' 'Francesco Beghini ([email protected])') __version__ = '3.0.11' __date__ = '07 Jul 2021' import sys try: from .util_fun import info, error except ImportError: from util_fun import info, error if sys.version_info[0] < 3: error("StrainPhlAn " + __version__ + " requires Python 3, your current Python version is {}.{}.{}".format( sys.version_info[0], sys.version_info[1], sys.version_info[2]), exit=True) import pickle, bz2, os, time import subprocess as sb import argparse as ap from Bio import SeqIO, Seq, SeqRecord try: from .external_exec import generate_markers_fasta except ImportError: from external_exec import generate_markers_fasta # get the directory that contains this script metaphlan_script_install_folder = os.path.dirname(os.path.abspath(__file__)) DEFAULT_DB_FOLDER = os.path.join(metaphlan_script_install_folder, "../metaphlan_databases")
