def run_falcon_asm(input_files, output_files):
i_json_config_fn, i_fofn_fn = input_files[0:2]
o_fasta_fn = output_files[0]
cwd = os.getcwd()
pread_dir = os.path.dirname(i_fofn_fn)
preads4falcon_fasta_fn = os.path.join(pread_dir, "preads4falcon.fasta")
db_file = os.path.join(pread_dir, "preads.db")
job_done = os.path.join(cwd, "job_done")
config = _get_config_from_json_fileobj(open(i_json_config_fn))
script_fn = os.path.join(cwd, "run_falcon_asm.sh")
args = {
"las_fofn_fn": i_fofn_fn,
"preads4falcon_fasta_fn": preads4falcon_fasta_fn,
"db_file_fn": db_file,
"config": config,
"job_done": job_done,
"script_fn": script_fn,
}
assert_nonzero(i_fofn_fn)
assert_nonzero(preads4falcon_fasta_fn)
support.run_falcon_asm(**args)
run_cmd("bash %s" % script_fn, sys.stdout, sys.stderr, shell=False)
p_ctg = "p_ctg.fa"
assert_nonzero(p_ctg)
n_records = _linewrap_fasta(p_ctg, o_fasta_fn)
if n_records == 0:
# We already checked 0-length, but maybe this is still possible.
# Really, we want to detect 0 base-length, but I do not know how yet.
raise Exception("No records found in primary contigs: '%s'" % os.path.abspath(p_ctg))
say("Finished run_falcon_asm(%s, %s)" % (repr(input_files), repr(output_files)))
return 0
def run_falcon_asm(input_files, output_files):
i_json_config_fn, i_fofn_fn = input_files
o_fasta_fn, = output_files
cwd = os.getcwd()
pread_dir = os.path.dirname(i_fofn_fn)
preads4falcon_fasta_fn = os.path.join(pread_dir, 'preads4falcon.fasta')
db_file = os.path.join(pread_dir, 'preads.db')
job_done = os.path.join(cwd, 'job_done')
config = _get_config_from_json_fileobj(open(i_json_config_fn))
script_fn = os.path.join(cwd, "run_falcon_asm.sh")
args = {
'las_fofn_fn': i_fofn_fn,
'preads4falcon_fasta_fn': preads4falcon_fasta_fn,
'db_file_fn': db_file,
'config': config,
'job_done': job_done,
'script_fn': script_fn,
}
assert_nonzero(i_fofn_fn)
assert_nonzero(preads4falcon_fasta_fn)
support.run_falcon_asm(**args)
run_cmd('bash %s' % script_fn, sys.stdout, sys.stderr, shell=False)
p_ctg = 'p_ctg.fa'
assert_nonzero(p_ctg)
n_records = _linewrap_fasta(p_ctg, o_fasta_fn)
if n_records == 0:
# We already checked 0-length, but maybe this is still possible.
# Really, we want to detect 0 base-length, but I do not know how yet.
raise Exception("No records found in primary contigs: '%s'" %
os.path.abspath(p_ctg))
say('Finished run_falcon_asm(%s, %s)' %
(repr(input_files), repr(output_files)))
return 0
def run_rm_las(input_files, output_files, prefix):
""" Delete all intermediate las files. """
cmd = "pwd && find .. -type f -name '%s*.las' -delete -print" % prefix
say(cmd)
run_cmd(cmd, sys.stdout, sys.stderr)
with open(output_files[0], 'w') as writer:
writer.write("#%s" % cmd)
return 0
def run_rm_las(input_files, output_files, prefix):
""" Delete all intermediate las files. """
cmd = "pwd && find .. -type f -name '%s*.las' -delete -print" % prefix
say(cmd)
run_cmd(cmd, sys.stdout, sys.stderr)
with open(output_files[0], "w") as writer:
writer.write("#%s" % cmd)
return 0
def run_falcon_config_get_fasta(input_files, output_files):
i_config_fn, = input_files
o_fofn_fn, = output_files
config = _get_config(i_config_fn)
i_fofn_fn = config['input_fofn']
if not os.path.isabs(i_fofn_fn):
i_fofn_fn = os.path.join(os.path.dirname(i_config_fn), i_fofn_fn)
msg = '%r -> %r' % (i_fofn_fn, o_fofn_fn)
say(msg)
with cd(os.path.dirname(i_fofn_fn)):
return support.make_fofn_abs(i_fofn_fn, o_fofn_fn)
return 0
def run_falcon_config_get_fasta(input_files, output_files):
i_config_fn, = input_files
o_fofn_fn, = output_files
config = _get_config(i_config_fn)
i_fofn_fn = config["input_fofn"]
if not os.path.isabs(i_fofn_fn):
i_fofn_fn = os.path.join(os.path.dirname(i_config_fn), i_fofn_fn)
msg = "%r -> %r" % (i_fofn_fn, o_fofn_fn)
say(msg)
with cd(os.path.dirname(i_fofn_fn)):
return support.make_fofn_abs(i_fofn_fn, o_fofn_fn)
return 0
def run_falcon_make_fofn_abs(input_files, output_files):
i_json_fn, = input_files
o_fofn_fn, = output_files
config = _get_config_from_json_fileobj(open(i_json_fn))
i_fofn_fn = config["input_fofn"]
if not i_fofn_fn.startswith("/"):
# i_fofn_fn can be relative to the location of the config file.
original_config_fn = config["original_self"]
i_fofn_fn = os.path.join(os.path.dirname(original_config_fn), i_fofn_fn)
msg = "run_falcon_make_fofn_abs(%r -> %r)" % (i_fofn_fn, o_fofn_fn)
say(msg)
with cd(os.path.dirname(i_fofn_fn)):
return support.make_fofn_abs(i_fofn_fn, o_fofn_fn)
return 0
def run_falcon_make_fofn_abs(input_files, output_files):
i_json_fn, = input_files
o_fofn_fn, = output_files
config = _get_config_from_json_fileobj(open(i_json_fn))
i_fofn_fn = config['input_fofn']
if not i_fofn_fn.startswith('/'):
# i_fofn_fn can be relative to the location of the config file.
original_config_fn = config['original_self']
i_fofn_fn = os.path.join(os.path.dirname(original_config_fn),
i_fofn_fn)
msg = 'run_falcon_make_fofn_abs(%r -> %r)' % (i_fofn_fn, o_fofn_fn)
say(msg)
with cd(os.path.dirname(i_fofn_fn)):
return support.make_fofn_abs(i_fofn_fn, o_fofn_fn)
return 0
def _get_config_from_json_fileobj(ifs_json):
i_json = ifs_json.read()
say('JSON=\n%s' % i_json[:1024]) # truncated
return json.loads(i_json)
def run_cmd(cmd, *args, **kwds):
say('RUN: %s' % repr(cmd))
rc = pb_run_cmd(cmd, *args, **kwds)
say(' RC: %s' % repr(rc))
if rc.exit_code:
raise Exception(repr(rc))
def _get_config_from_json_fileobj(ifs_json):
i_json = ifs_json.read()
say("JSON=\n%s" % i_json[:1024]) # truncated
return json.loads(i_json)
def run_cmd(cmd, *args, **kwds):
say("RUN: %s" % repr(cmd))
rc = pb_run_cmd(cmd, *args, **kwds)
say(" RC: %s" % repr(rc))
if rc.exit_code:
raise Exception(repr(rc))