def main(argv):
global parser
(opts, args) = parser.parse_args()
if valid_arguments(opts, args):
print usage
sys.exit(0)
signal.signal(signal.SIGINT, sigint_handler)
signal.signal(signal.SIGTERM, sigint_handler)
eprintf("COMMAND : %s\n", sys.argv[0] + ' ' + ' '.join(argv))
# initialize the input directory or file
input_fp = opts.input_fp
output_dir = path.abspath(opts.output_dir)
verbose = opts.verbose
print_only = opts.print_only
sample_subset = removeSuffix(opts.sample_subset)
run_type = opts.run_type.strip()
'''no need to remove the whole directory'''
# if run_type == 'overwrite':
# force_remove_dir=True
# else:
# force_remove_dir=False
if opts.config_file:
config_file = opts.config_file
else:
config_file = cmd_folder + PATHDELIM + metapaths_config
if opts.ncbi_header and opts.ncbi_sbt:
if not path.exists(opts.ncbi_header):
print "Could not open or missing NCBI header file " + opts.ncbi_header
print "Either disable option to CREATE_SEQUIN_FILE or provide a valid header file"
sys.exit(0)
if not path.exists(opts.ncbi_sbt):
print """You must must have a sbt file obtained from the NCBI \"Create Submission Template\" form \n
http://www.ncbi.nlm.nih.gov/WebSub/template.cgi """ + opts.ncbi_sbt
sys.exit(0)
ncbi_sequin_params = path.abspath(opts.ncbi_header)
ncbi_sequin_sbt = path.abspath(opts.ncbi_sbt)
else:
ncbi_sequin_params = None
ncbi_sequin_sbt = None
# try to load the parameter file
try:
if opts.parameter_fp:
parameter_fp = opts.parameter_fp
else:
parameter_fp = cmd_folder + PATHDELIM + metapaths_param
except IOError:
raise IOError, (
"Can't open parameters file (%s). Does it exist? Do you have read access?"
% opts.parameter_fp)
try:
if run_type in ['overlay', 'safe'] and not path.exists(output_dir):
makedirs(output_dir)
except OSError:
print ""
print "ERROR: Cannot create output directory \"" + output_dir + "\"\n"+\
" Perhaps directory \"" + output_dir + "\" already exists.\n" +\
" Please choose a different directory, or \n" +\
" run with the option \"-r overwrite\" to force overwrite it."
sys.exit(1)
if verbose:
status_update_callback = print_to_stdout
else:
status_update_callback = no_status_updates
command_line_params = {}
command_line_params['verbose'] = opts.verbose
params = parse_metapaths_parameters(parameter_fp)
""" load the sample inputs it expects either a fasta
file or a directory containing fasta and yaml file pairs
"""
globalerrorlogger = WorkflowLogger(generate_log_fp(
output_dir, basefile_name='global_errors_warnings'),
open_mode='w')
input_output_list = {}
if path.isfile(input_fp):
""" check if it is a file """
input_output_list = create_an_input_output_pair(
input_fp, output_dir, globalerrorlogger=globalerrorlogger)
else:
if path.exists(input_fp):
""" check if dir exists """
input_output_list = create_input_output_pairs(
input_fp, output_dir, globalerrorlogger=globalerrorlogger)
else:
""" must be an error """
eprintf(
"ERROR\tNo valid input sample file or directory containing samples exists .!"
)
eprintf("ERROR\tAs provided as arguments in the -in option.!\n")
exit_process(
"ERROR\tAs provided as arguments in the -in option.!\n")
""" these are the subset of sample to process if specified
in case of an empty subset process all the sample """
# remove all samples that are not specifed unless sample_subset is empty
remove_unspecified_samples(input_output_list,
sample_subset,
globalerrorlogger=globalerrorlogger)
# add check the config parameters
sorted_input_output_list = sorted(input_output_list.keys())
filetypes = check_file_types(sorted_input_output_list)
#stop on in valid samples
if not halt_on_invalid_input(input_output_list, filetypes, sample_subset):
globalerrorlogger.printf(
"ERROR\tInvalid inputs found. Check for file with bad format or characters!\n"
)
halt_process(opts.delay)
# make sure the sample files are found
report_missing_filenames(input_output_list,
sample_subset,
logger=globalerrorlogger)
#check the pipeline configuration
config_settings = read_pipeline_configuration(config_file,
globalerrorlogger)
parameter = Parameters()
if not staticDiagnose(config_settings, params, logger=globalerrorlogger):
eprintf(
"ERROR\tFailed to pass the test for required scripts and inputs before run\n"
)
globalerrorlogger.printf(
"ERROR\tFailed to pass the test for required scripts and inputs before run\n"
)
halt_process(opts.delay)
samplesData = {}
# PART1 before the blast
block_mode = opts.block_mode
runid = opts.runid
try:
# load the sample information
print "RUNNING MetaPathways version 2.5.2"
if len(input_output_list):
for input_file in sorted_input_output_list:
sample_output_dir = input_output_list[input_file]
algorithm = get_parameter(params,
'annotation',
'algorithm',
default='LAST').upper()
s = SampleData()
s.setInputOutput(inputFile=input_file,
sample_output_dir=sample_output_dir)
s.setParameter('algorithm', algorithm)
s.setParameter('ncbi_params_file', ncbi_sequin_params)
s.setParameter('ncbi_sequin_sbt', ncbi_sequin_sbt)
s.setParameter('FILE_TYPE', filetypes[input_file][0])
if params["INPUT"]['format'] in [
"gbk-annotated", "gff-annotated"
]:
s.setParameter('ANNOTATED', True)
else:
s.setParameter('ANNOTATED', False)
s.setParameter('SEQ_TYPE', filetypes[input_file][1])
s.clearJobs()
if run_type == 'overwrite' and path.exists(sample_output_dir):
shutil.rmtree(sample_output_dir)
makedirs(sample_output_dir)
if not path.exists(sample_output_dir):
makedirs(sample_output_dir)
s.prepareToRun()
samplesData[input_file] = s
# load the sample information
run_metapathways(samplesData,
sample_output_dir,
output_dir,
globallogger=globalerrorlogger,
command_line_params=command_line_params,
params=params,
metapaths_config=metapaths_config,
status_update_callback=status_update_callback,
config_file=config_file,
run_type=run_type,
config_settings=config_settings,
block_mode=block_mode,
runid=runid)
else:
eprintf(
"ERROR\tNo valid input files/Or no files specified to process in folder %s!\n",
sQuote(input_fp))
globalerrorlogger.printf(
"ERROR\tNo valid input files to process in folder %s!\n",
sQuote(input_fp))
# blast the files
blasting_system = get_parameter(params,
'metapaths_steps',
'BLAST_REFDB',
default='yes')
if blasting_system == 'grid':
# blasting the files files on the grids
input_files = sorted_input_output_list
blast_in_grid(
sampleData[input_file],
input_files,
path.abspath(opts.output_dir), #important to use opts.
params=params,
metapaths_config=metapaths_config,
config_file=config_file,
run_type=run_type,
runid=runid)
except:
exit_process(str(traceback.format_exc(10)), logger=globalerrorlogger)
eprintf(" *********** \n")
eprintf("INFO : FINISHED PROCESSING THE SAMPLES \n")
eprintf(" THE END \n")
eprintf(" *********** \n")
halt_process(opts.delay)
def main(argv):
global parser
(opts, args) = parser.parse_args()
if valid_arguments(opts, args):
print usage
sys.exit(0)
signal.signal(signal.SIGINT, sigint_handler)
signal.signal(signal.SIGTERM, sigint_handler)
eprintf("%-10s:%s\n" %('COMMAND', sys.argv[0] + ' ' + ' '.join(argv)) )
# initialize the input directory or file
input_fp = opts.input_fp
output_dir = path.abspath(opts.output_dir)
verbose = opts.verbose
print_only = opts.print_only
sample_subset = removeSuffix(opts.sample_subset)
run_type = opts.run_type.strip()
'''no need to remove the whole directory'''
# if run_type == 'overwrite':
# force_remove_dir=True
# else:
# force_remove_dir=False
if opts.config_file:
config_file= opts.config_file
else:
config_file = cmd_folder + PATHDELIM + metapaths_config
# try to load the parameter file
try:
if opts.parameter_fp:
parameter_fp= opts.parameter_fp
else:
parameter_fp = cmd_folder + PATHDELIM + metapaths_param
except IOError:
raise IOError, ( "Can't open parameters file (%s). Does it exist? Do you have read access?" % opts.parameter_fp )
try:
if run_type in ['overlay', 'safe'] and not path.exists(output_dir):
makedirs(output_dir)
except OSError:
print ""
print "ERROR: Cannot create output directory \"" + output_dir + "\"\n"+\
" Perhaps directory \"" + output_dir + "\" already exists.\n" +\
" Please choose a different directory, or \n" +\
" run with the option \"-r overwrite\" to force overwrite it."
sys.exit(2)
if verbose:
status_update_callback = print_to_stdout
else:
status_update_callback = no_status_updates
command_line_params={}
command_line_params['verbose']= opts.verbose
if not path.exists(parameter_fp):
eprintf("%-10s: No parameters file %s found!\n" %('WARNING', parameter_fp))
eprintf("%-10s: Creating a parameters file %s found!\n" %('INFO', parameter_fp))
create_metapaths_parameters(parameter_fp, cmd_folder)
params=parse_metapaths_parameters(parameter_fp)
""" load the sample inputs it expects either a fasta
file or a directory containing fasta and yaml file pairs
"""
globalerrorlogger = WorkflowLogger(generate_log_fp(output_dir, basefile_name= 'global_errors_warnings'), open_mode='w')
input_output_list = {}
if path.isfile(input_fp):
""" check if it is a file """
input_output_list = create_an_input_output_pair(input_fp, output_dir, globalerrorlogger=globalerrorlogger)
else:
if path.exists(input_fp):
""" check if dir exists """
input_output_list = create_input_output_pairs(input_fp, output_dir, globalerrorlogger=globalerrorlogger)
else:
""" must be an error """
eprintf("ERROR\tNo valid input sample file or directory containing samples exists .!")
eprintf("ERROR\tAs provided as arguments in the -in option.!\n")
exit_process("ERROR\tAs provided as arguments in the -in option.!\n")
""" these are the subset of sample to process if specified
in case of an empty subset process all the sample """
# remove all samples that are not specifed unless sample_subset is empty
remove_unspecified_samples(input_output_list, sample_subset, globalerrorlogger = globalerrorlogger)
# add check the config parameters
sorted_input_output_list = sorted(input_output_list.keys())
filetypes = check_file_types(sorted_input_output_list)
#stop on in valid samples
if not halt_on_invalid_input(input_output_list, filetypes, sample_subset):
globalerrorlogger.printf("ERROR\tInvalid inputs found. Check for file with bad format or characters!\n")
halt_process(opts.delay)
# make sure the sample files are found
report_missing_filenames(input_output_list, sample_subset, logger=globalerrorlogger)
#check the pipeline configuration
print 'config'
if not path.exists(config_file):
eprintf("%-10s: No config file %s found!\n" %('WARNING', config_file))
eprintf("%-10s: Creating a config file %s!\n" %('INFO', config_file))
if not environment_variables_defined():
sys.exit(0)
create_metapaths_configuration(config_file, cmd_folder)
config_settings = read_pipeline_configuration(config_file, globalerrorlogger)
parameter = Parameters()
if not staticDiagnose(config_settings, params, logger = globalerrorlogger):
eprintf("ERROR\tFailed to pass the test for required scripts and inputs before run\n")
globalerrorlogger.printf("ERROR\tFailed to pass the test for required scripts and inputs before run\n")
return
samplesData = {}
# PART1 before the blast
block_mode = opts.block_mode
runid = opts.runid
try:
# load the sample information
print "RUNNING MetaPathways version FogDog 3.0"
if len(input_output_list):
for input_file in sorted_input_output_list:
sample_output_dir = input_output_list[input_file]
algorithm = get_parameter(params, 'annotation', 'algorithm', default='LAST').upper()
s = SampleData()
s.setInputOutput(inputFile = input_file, sample_output_dir = sample_output_dir)
s.setParameter('algorithm', algorithm)
s.setParameter('FILE_TYPE', filetypes[input_file][0])
s.setParameter('SEQ_TYPE', filetypes[input_file][1])
s.clearJobs()
if run_type=='overwrite' and path.exists(sample_output_dir):
shutil.rmtree(sample_output_dir)
makedirs(sample_output_dir)
if not path.exists(sample_output_dir):
makedirs(sample_output_dir)
s.prepareToRun()
samplesData[input_file] = s
# load the sample information
run_metapathways(
samplesData,
sample_output_dir,
output_dir,
globallogger = globalerrorlogger,
command_line_params=command_line_params,
params=params,
metapaths_config=metapaths_config,
status_update_callback=status_update_callback,
config_file=config_file,
run_type = run_type,
config_settings = config_settings,
block_mode = block_mode,
runid = runid
)
else:
eprintf("ERROR\tNo valid input files/Or no files specified to process in folder %s!\n",sQuote(input_fp) )
globalerrorlogger.printf("ERROR\tNo valid input files to process in folder %s!\n",sQuote(input_fp) )
except:
exit_process(str(traceback.format_exc(10)), logger= globalerrorlogger )
eprintf(" *********** \n")
eprintf("INFO : FINISHED PROCESSING THE SAMPLES \n")
eprintf(" THE END \n")
eprintf(" *********** \n")