dataSet.filters.addFilter(**filters)
log.info("{i} other filters added".format(i=len(filters)))
rlen = sanitize_read_length(read_length)
if rlen:
dataSet.filters.addRequirement(length=[('>=', rlen)])
if rlen or other_filters:
dataSet.updateCounts()
dataSet.write(out_file)
return 0
@registry("filterdataset",
"0.1.0",
FileTypes.DS_SUBREADS,
subreads_file_type,
is_distributed=True,
nproc=1,
options={
"read_length": rl_opt,
"other_filters": filters_opt
})
def run_filterDataSet(rtc):
return run_filter_dataset(
rtc.task.input_files[0], rtc.task.output_files[0],
rtc.task.options["pbcoretools.task_options.read_length"],
rtc.task.options["pbcoretools.task_options.other_filters"])
if __name__ == '__main__':
sys.exit(registry_runner(registry, sys.argv[1:]))
# This is just for test purposes
log.info("Running example main with {i} {o} kw:{k}".format(i=input_file,
o=output_file,
k=kwargs))
# write mock output files, otherwise the End-to-End test will fail when
# run within testkit
with open(output_file, 'w') as f:
f.write("MOCK TEST DATA")
return 0
@registry("dev_mk_example_txt", "0.2.2", (FileTypes.TXT, ), (FileTypes.TXT, ), nproc=1, options=dict(alpha=1234))
def run_rtc(rtc):
"""Example Task for testing. Takes a txt file as input and writes a txt file with mock data"""
# The above docstring will be used as the Task/ToolContract Description
log.info("Got RTC task options {t}".format(t=rtc.task.options))
log.info("Got nproc {n}".format(n=rtc.task.nproc))
# The Task options are now accessible via global identifier
alpha = rtc.task.options['pbsmrtpipe_examples.task_options.alpha']
return _example_main(rtc.task.input_files[0], rtc.task.output_files[0], nproc=rtc.task.nproc, alpha=alpha)
if __name__ == '__main__':
sys.exit(registry_runner(registry, sys.argv[1:]))
DRIVER_BASE = 'python -m pbinternal2.analysis_tools '
registry = registry_builder(TOOL_NAMESPACE, DRIVER_BASE)
@registry("hello_world_analysis", "0.1.0", FileTypes.FASTA, FileTypes.REPORT, is_distributed=True, nproc=1)
def run_rtc(rtc):
return run_hello_world(rtc.task.input_files[0], rtc.task.output_files[0])
@registry("cond_to_report", "0.2.0", FileTypes.COND_RESEQ, FileTypes.REPORT, nproc=1, is_distributed=False)
def run_rtc(rtc):
return run_cond_to_report(rtc.task.input_files[0], rtc.task.output_files[0])
@registry("cond_to_alignmentsets_report", "0.2.0", FileTypes.COND_RESEQ, FileTypes.REPORT, nproc=1, is_distributed=True)
def run_rtc(rtc):
return run_cond_alignmentsets_to_report(rtc.task.input_files[0], rtc.task.output_files[0])
if __name__ == '__main__':
default_log_level = logging.DEBUG
sys.exit(registry_runner(registry,
sys.argv[1:],
default_log_level=default_log_level))
if __name__ == '__main__':
sys.exit(registry_runner(registry, sys.argv[1:]))
label = "label_" + file_type.file_type_id
desc = "File {f}".format(f=file_type)
return OutputFileType(file_type.file_type_id, label, repr(file_type), desc,
default_name)
def _to_outputs(file_types):
return [_to_output(i, ft) for i, ft in enumerate(file_types)]
@registry("dev_txt_custom_outs",
"0.1.0",
FileTypes.TXT,
_to_outputs((FileTypes.TXT, FileTypes.TXT)),
name="Custom Txt Task")
def run_rtc(rtc):
"""Test for using OutputFileTypes as outputs
Output types can be specified as FileType, or OutputFileType instances
"""
return _example_main(rtc.task.input_files, rtc.task.output_files)
if __name__ == '__main__':
default_log_level = logging.DEBUG
sys.exit(
registry_runner(registry,
sys.argv[1:],
default_log_level=default_log_level,
version=__version__))
DRIVER_BASE = "python -m pbinternal2.tasks.loading "
LOADING_DIST_ID = "loading_dist"
LOADING_DIST = 20
def _get_id(base_opt):
return "{n}.task_options.{o}".format(n=TOOL_NAMESPACE,
o=base_opt)
registry = registry_builder(TOOL_NAMESPACE, Constants.DRIVER_BASE)
@registry("loading_vs_poisson_report", "0.1.0",
FileTypes.STS_H5,
FileTypes.REPORT,
is_distributed=True,
nproc=1,
options={Constants.LOADING_DIST_ID: Constants.LOADING_DIST})
def task_loading_vs_poisson_report(rtc):
return loading_vs_poisson(
rtc.task.input_files[0],
rtc.task.output_files[0],
rtc.task.nproc,
rtc.task.options[_get_id(Constants.LOADING_DIST_ID)])
if __name__ == '__main__':
sys.exit(registry_runner(registry, sys.argv[1:],
default_log_level=logging.INFO))
@registry("dev_txt_hello", "0.1.0", FileTypes.TXT, (FileTypes.TXT, FileTypes.TXT), nproc=3, is_distributed=False)
def run_rtc(rtc):
return _example_main(rtc.task.input_files, rtc.task.output_files)
def _to_output(i, file_type):
default_name = "_".join([file_type.file_type_id, file_type.base_name + "_" + str(i)])
label = "label_" + file_type.file_type_id
desc = "File {f}".format(f=file_type)
return OutputFileType(file_type.file_type_id, label, repr(file_type), desc, default_name)
def _to_outputs(file_types):
return [_to_output(i, ft) for i, ft in enumerate(file_types)]
@registry(
"dev_txt_custom_outs", "0.1.0", FileTypes.TXT, _to_outputs((FileTypes.TXT, FileTypes.TXT)), name="Custom Txt Task"
)
def run_rtc(rtc):
"""Test for using OutputFileTypes as outputs
Output types can be specified as FileType, or OutputFileType instances
"""
return _example_main(rtc.task.input_files, rtc.task.output_files)
if __name__ == "__main__":
default_log_level = logging.DEBUG
sys.exit(registry_runner(registry, sys.argv[1:], default_log_level=default_log_level, version=__version__))
@registry("dev_txt_hello", "0.1.0", FileTypes.TXT, (FileTypes.TXT, FileTypes.TXT), nproc=3, is_distributed=False)
def run_rtc(rtc):
return _example_main(rtc.task.input_files, rtc.task.output_files)
def _to_output(i, file_type):
default_name = "_".join([file_type.file_type_id, file_type.base_name + "_" + str(i)])
label = "label_" + file_type.file_type_id
desc = "File {f}".format(f=file_type)
return OutputFileType(file_type.file_type_id, label, repr(file_type), desc, default_name)
def _to_outputs(file_types):
return [_to_output(i, ft) for i, ft in enumerate(file_types)]
@registry("dev_txt_custom_outs", "0.1.0", FileTypes.TXT, _to_outputs((FileTypes.TXT, FileTypes.TXT)), name="Custom Txt Task")
def run_rtc(rtc):
"""Test for using OutputFileTypes as outputs
Output types can be specified as FileType, or OutputFileType instances
"""
return _example_main(rtc.task.input_files, rtc.task.output_files)
if __name__ == '__main__':
default_log_level = logging.DEBUG
sys.exit(registry_runner(registry,
sys.argv[1:],
default_log_level=default_log_level))