def for_task(
i_length_cutoff_fn,
i_json_config_fn,
i_preads_fofn_fn,
i_raw_reads_db_fn, # maybe symlinked
o_json_fn,
):
"""This is used by HGAP4, run_report_preassembly_yield().
See pbfalcon.tusks
"""
tasks_dir = os.path.dirname(os.path.dirname(i_json_config_fn))
cfg = _get_cfg(i_json_config_fn, i_length_cutoff_fn)
genome_length = int(cfg.get('genome_size', 0)) # different name in falcon
length_cutoff = cfg['length_cutoff']
i_raw_reads_db_fn = os.path.realpath(i_raw_reads_db_fn)
report_dict = stats_preassembly.calc_dict(
i_preads_fofn_fn,
i_raw_reads_db_fn,
genome_length,
length_cutoff,
)
report = produce_report(**report_dict)
with open(o_json_fn, 'w') as ofs:
log.info("Writing report to {!r}:\n{}".format(o_json_fn, report))
ofs.write(report)
def for_task(
i_json_config_fn,
i_preads_fofn_fn,
i_raw_reads_db_fn, # maybe symlinked
o_json_fn,
):
"""See pbfalcon.tusks
"""
tasks_dir = os.path.dirname(os.path.dirname(i_json_config_fn))
i_length_cutoff_fn = os.path.join(tasks_dir, 'falcon_ns.tasks.task_falcon0_build_rdb-0', 'length_cutoff')
cfg = _get_cfg(i_json_config_fn, i_length_cutoff_fn)
genome_length = int(cfg.get('genome_size', 0)) # different name in falcon
length_cutoff = cfg['length_cutoff']
i_raw_reads_db_fn = os.path.realpath(i_raw_reads_db_fn)
report_dict = stats_preassembly.calc_dict(
i_preads_fofn_fn,
i_raw_reads_db_fn,
genome_length,
length_cutoff,
)
report = produce_report(**report_dict)
log.info('%r -> %r' %(report, o_json_fn))
with open(o_json_fn, 'w') as ofs:
log.info("Writing report to {!r}.".format(o_json_fn))
content = report.to_json()
ofs.write(content)
def task_report_pre_assembly(self):
# TODO(CD): Bashify this, in case it is slow.
i_raw_reads_db_fn = fn(self.raw_reads_db)
i_preads_fofn_fn = fn(self.preads_fofn)
i_length_cutoff_fn = fn(self.length_cutoff_fn)
o_json_fn = fn(self.pre_assembly_report)
cfg = self.parameters
genome_length = int(cfg.get('genome_size', 0)) # different name in falcon
length_cutoff = int(cfg['length_cutoff'])
length_cutoff = support.get_length_cutoff(length_cutoff,
i_length_cutoff_fn)
kwds = {
'i_raw_reads_db_fn': i_raw_reads_db_fn,
'i_preads_fofn_fn': i_preads_fofn_fn,
'genome_length': genome_length,
'length_cutoff': length_cutoff,
}
fc_run_logger.info('Report inputs: {}'.format(repr(kwds)))
report_dict = stats_preassembly.calc_dict(**kwds)
content = json.dumps(report_dict,
sort_keys=True,
indent=4,
separators=(',', ': '))
fc_run_logger.info('Report stats:\n{}'.format(content))
open(o_json_fn, 'w').write(content)
def for_task(
i_json_config_fn,
i_preads_fofn_fn,
i_raw_reads_db_fn, # maybe symlinked
o_json_fn,
):
"""This is used by HGAP4, run_report_preassembly_yield().
See pbfalcon.tusks
"""
tasks_dir = os.path.dirname(os.path.dirname(i_json_config_fn))
i_length_cutoff_fn = os.path.join(tasks_dir, 'falcon_ns.tasks.task_falcon0_build_rdb-0', 'length_cutoff')
cfg = _get_cfg(i_json_config_fn, i_length_cutoff_fn)
genome_length = int(cfg.get('genome_size', 0)) # different name in falcon
length_cutoff = cfg['length_cutoff']
i_raw_reads_db_fn = os.path.realpath(i_raw_reads_db_fn)
report_dict = stats_preassembly.calc_dict(
i_preads_fofn_fn,
i_raw_reads_db_fn,
genome_length,
length_cutoff,
)
report = produce_report(**report_dict)
with open(o_json_fn, 'w') as ofs:
log.info("Writing report to {!r}:\n{}".format(o_json_fn, report))
ofs.write(report)
def task_report_pre_assembly(self):
# TODO(CD): Bashify this, in case it is slow.
i_raw_reads_db_fn = fn(self.raw_reads_db)
i_preads_fofn_fn = fn(self.preads_fofn)
i_length_cutoff_fn = fn(self.length_cutoff_fn)
o_json_fn = fn(self.pre_assembly_report)
cfg = self.parameters
genome_length = int(cfg.get('genome_size', 0)) # different name in falcon
length_cutoff = int(cfg['length_cutoff'])
length_cutoff = support.get_length_cutoff(length_cutoff, i_length_cutoff_fn)
kwds = {
'i_raw_reads_db_fn': i_raw_reads_db_fn,
'i_preads_fofn_fn': i_preads_fofn_fn,
'genome_length': genome_length,
'length_cutoff': length_cutoff,
}
fc_run_logger.info('Report inputs: {}'.format(repr(kwds)))
report_dict = stats_preassembly.calc_dict(**kwds)
content = json.dumps(report_dict, sort_keys=True, indent=4, separators=(',', ': '))
fc_run_logger.info('Report stats:\n{}'.format(content))
open(o_json_fn, 'w').write(content)
