def run(self):
"""Run"""
logging.debug("root_dir: {d}.".format(d=self.root_dir))
logging.debug("nfl_fa: {f}.".format(f=self.nfl_fa))
logging.debug("Total number of chunks: N={N}.".format(N=self.N))
# Validate input files,
(num_reads, reads_per_split, nfl_dir, splitted_fas_todo) = \
self.validate_inputs()
logging.info("Total number of reads is {n}.".format(n=num_reads))
logging.info("Splitting nfl_fa into chunks each " +
"containing {n} reads.".format(n=reads_per_split))
splitted_fas_done = splitFasta(input_fasta=real_ppath(self.nfl_fa),
reads_per_split=reads_per_split,
out_dir=nfl_dir,
out_prefix="input.split")
logging.info("Splitted files are: " + "\n".join(splitted_fas_done))
for fa in splitted_fas_todo:
if fa not in splitted_fas_done:
logging.info("touching {f}".format(f=fa))
touch(fa)
def run(self):
"""Run"""
logging.debug("root_dir: {d}.".format(d=self.root_dir))
logging.debug("nfl_fa: {f}.".format(f=self.nfl_fa))
logging.debug("Total number of chunks: N={N}.".format(N=self.N))
# Validate input files,
(num_reads, reads_per_split, nfl_dir, splitted_fas_todo) = self.validate_inputs()
logging.info("Total number of reads is {n}.".format(n=num_reads))
logging.info("Splitting nfl_fa into chunks each " + "containing {n} reads.".format(n=reads_per_split))
splitted_fas_done = splitFasta(
input_fasta=real_ppath(self.nfl_fa),
reads_per_split=reads_per_split,
out_dir=nfl_dir,
out_prefix="input.split",
)
logging.info("Splitted files are: " + "\n".join(splitted_fas_done))
for fa in splitted_fas_todo:
if fa not in splitted_fas_done:
logging.info("touching {f}".format(f=fa))
touch(fa)
def build_uc_from_partial_daligner(input_fastq, ref_fasta, out_pickle,
ccs_fofn=None,
done_filename=None, use_finer_qv=False, cpus=24, no_qv_or_aln_checking=True):
"""
Given an input_fasta file of non-full-length (partial) reads and
(unpolished) consensus isoforms sequences in ref_fasta, align reads to
consensus isoforms using BLASR, and then build up a mapping between
consensus isoforms and reads (i.e., assign reads to isoforms).
Finally, save
{isoform_id: [read_ids],
nohit: set(no_hit_read_ids)}
to an output pickle file.
ccs_fofn --- If None, assume no quality value is available,
otherwise, use QV from ccs_fofn.
"""
input_fastq = realpath(input_fastq)
input_fasta = input_fastq[:input_fastq.rfind('.')] + '.fasta'
ice_fq2fa(input_fastq, input_fasta)
ref_fasta = realpath(ref_fasta)
out_pickle = realpath(out_pickle)
output_dir = os.path.dirname(out_pickle)
daligner_sensitive_mode, _low, _high, _ignore5, _ignore3, _ece_min_len = get_daligner_sensitivity_setting(ref_fasta, is_fasta=True)
# DB should always be already converted
ref_obj = DazzIDHandler(ref_fasta, True)
input_obj = DazzIDHandler(input_fasta, False)
# ice_partial is already being called through qsub, so run everything local!
runner = DalignerRunner(input_fasta, ref_fasta, is_FL=False, same_strand_only=False, \
query_converted=True, db_converted=True, query_made=False, \
db_made=True, use_sge=False, cpus=cpus, sge_opts=None)
las_filenames, las_out_filenames = runner.runHPC(min_match_len=_low, output_dir=output_dir, sensitive_mode=daligner_sensitive_mode)
if no_qv_or_aln_checking:
# not using QVs or alignment checking!
# this probqv is just a DUMMY to pass to daligner_against_ref, which won't be used
logging.info("Not using QV for partial_uc. Loading dummy QV.")
probqv = ProbFromModel(.01, .07, .06)
else:
# if ccs_fofn is None:
# logging.info("Loading probability from model (0.01,0.07,0.06)")
# probqv = ProbFromModel(.01, .07, .06)
# else:
start_t = time.time()
probqv = ProbFromFastq(input_fastq)
logging.info("Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t))
# --------- comment out below since we are just using FASTQ / BAM
# if use_finer_qv:
# probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
# logging.info("Loading QVs from {i} + {f} took {s} secs".format(f=ccs_fofn, i=input_fasta,\
# s=time.time()-start_t))
# else:
# input_fastq = input_fasta[:input_fasta.rfind('.')] + '.fastq'
# logging.info("Converting {i} + {f} --> {fq}".format(i=input_fasta, f=ccs_fofn, fq=input_fastq))
# ice_fa2fq(input_fasta, ccs_fofn, input_fastq)
# probqv = ProbFromFastq(input_fastq)
# logging.info("Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t))
# print >> sys.stderr, "Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t)
logging.info("Calling dalign_against_ref ...")
partial_uc = {} # Maps each isoform (cluster) id to a list of reads
# which can map to the isoform
seen = set() # reads seen
logging.info("Building uc from DALIGNER hits.")
for las_out_filename in las_out_filenames:
start_t = time.time()
hitItems = dalign_against_ref(input_obj, ref_obj, las_out_filename,
is_FL=False,
sID_starts_with_c=True,
qver_get_func=probqv.get_smoothed,
qvmean_get_func=probqv.get_mean,
ece_penalty=1,
ece_min_len=_ece_min_len,
same_strand_only=False,
no_qv_or_aln_checking=no_qv_or_aln_checking,
max_missed_start=_ignore5,
max_missed_end=_ignore3)
for h in hitItems:
if h.ece_arr is not None:
if h.cID not in partial_uc:
partial_uc[h.cID] = set()
partial_uc[h.cID].add(h.qID)
seen.add(h.qID)
logging.info("processing {0} took {1} sec".format(las_out_filename, time.time()-start_t))
print >> sys.stderr, "processing {0} took {1} sec".format(las_out_filename, time.time()-start_t)
for k in partial_uc:
partial_uc[k] = list(partial_uc[k])
allhits = set(r.name.split()[0] for r in FastaReader(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: {f}.".format(f=out_pickle))
with open(out_pickle, 'w') as f:
dump({'partial_uc': partial_uc, 'nohit': nohit}, f)
done_filename = realpath(done_filename) if done_filename is not None \
else out_pickle + '.DONE'
logging.debug("Creating {f}.".format(f=done_filename))
touch(done_filename)
# remove all the .las and .las.out filenames
for file in las_filenames:
os.remove(file)
for file in las_out_filenames:
os.remove(file)
def build_uc_from_partial(input_fasta, ref_fasta, out_pickle,
sa_file=None, ccs_fofn=None,
done_filename=None, blasr_nproc=12, use_finer_qv=False):
"""
Given an input_fasta file of non-full-length (partial) reads and
(unpolished) consensus isoforms sequences in ref_fasta, align reads to
consensus isoforms using BLASR, and then build up a mapping between
consensus isoforms and reads (i.e., assign reads to isoforms).
Finally, save
{isoform_id: [read_ids],
nohit: set(no_hit_read_ids)}
to an output pickle file.
ccs_fofn --- If None, assume no quality value is available,
otherwise, use QV from ccs_fofn.
blasr_nproc --- equivalent to blasr -nproc, number of CPUs to use
"""
input_fasta = realpath(input_fasta)
m5_file = input_fasta + ".blasr"
out_pickle = realpath(out_pickle)
if sa_file is None:
if op.exists(input_fasta + ".sa"):
sa_file = input_fasta + ".sa"
cmd = "blasr {i} ".format(i=real_upath(input_fasta)) + \
"{r} -bestn 5 ".format(r=real_upath(ref_fasta)) + \
"-nproc {n} -m 5 ".format(n=blasr_nproc) + \
"-maxScore -1000 -minPctIdentity 85 " + \
"-out {o} ".format(o=real_upath(m5_file))
if sa_file is not None and op.exists(sa_file):
cmd += "-sa {sa}".format(sa=real_upath(sa_file))
logging.info("CMD: {cmd}".format(cmd=cmd))
_out, _code, _msg = backticks(cmd)
if _code != 0:
errMsg = "Command failed: {cmd}\n{e}".format(cmd=cmd, e=_msg)
logging.error(errMsg)
raise RuntimeError(errMsg)
if ccs_fofn is None:
logging.info("Loading probability from model (0.01,0.07,0.06)")
probqv = ProbFromModel(.01, .07, .06)
else:
start_t = time.time()
if use_finer_qv:
logging.info("Loading QVs from {i} + {f} took {s} secs".format(f=ccs_fofn, i=input_fasta,\
s=time.time()-start_t))
probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
else:
input_fastq = input_fasta[:input_fasta.rfind('.')] + '.fastq'
logging.info("Converting {i} + {f} --> {fq}".format(i=input_fasta, f=ccs_fofn, fq=input_fastq))
ice_fa2fq(input_fasta, ccs_fofn, input_fastq)
logging.info("Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t))
probqv = ProbFromFastq(input_fastq)
logging.info("Calling blasr_against_ref ...")
hitItems = blasr_against_ref(output_filename=m5_file,
is_FL=False,
sID_starts_with_c=True,
qvmean_get_func=probqv.get_mean,
qver_get_func=probqv.get_smoothed,
ece_penalty=1,
ece_min_len=20,
same_strand_only=False,
max_missed_start=200,
max_missed_end=50)
partial_uc = {} # Maps each isoform (cluster) id to a list of reads
# which can map to the isoform
seen = set() # reads seen
logging.info("Building uc from BLASR hits.")
for h in hitItems:
if h.ece_arr is not None:
if h.cID not in partial_uc:
partial_uc[h.cID] = set()
partial_uc[h.cID].add(h.qID)
seen.add(h.qID)
for k in partial_uc:
partial_uc[k] = list(partial_uc[k])
allhits = set(r.name.split()[0] for r in FastaReader(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: {f}.".format(f=out_pickle))
with open(out_pickle, 'w') as f:
dump({'partial_uc': partial_uc, 'nohit': nohit}, f)
os.remove(m5_file)
done_filename = realpath(done_filename) if done_filename is not None \
else out_pickle + '.DONE'
logging.debug("Creating {f}.".format(f=done_filename))
touch(done_filename)
def build_uc_from_partial_daligner(input_fasta, ref_fasta, out_pickle,
ccs_fofn=None,
done_filename=None, use_finer_qv=False, cpus=24, no_qv_or_aln_checking=True):
"""
Given an input_fasta file of non-full-length (partial) reads and
(unpolished) consensus isoforms sequences in ref_fasta, align reads to
consensus isoforms using BLASR, and then build up a mapping between
consensus isoforms and reads (i.e., assign reads to isoforms).
Finally, save
{isoform_id: [read_ids],
nohit: set(no_hit_read_ids)}
to an output pickle file.
ccs_fofn --- If None, assume no quality value is available,
otherwise, use QV from ccs_fofn.
"""
input_fasta = realpath(input_fasta)
ref_fasta = realpath(ref_fasta)
out_pickle = realpath(out_pickle)
output_dir = os.path.dirname(out_pickle)
daligner_sensitive_mode, _low, _high = get_daligner_sensitivity_setting(ref_fasta)
# DB should always be already converted
ref_obj = DazzIDHandler(ref_fasta, True)
input_obj = DazzIDHandler(input_fasta, False)
# ice_partial is already being called through qsub, so run everything local!
runner = DalignerRunner(input_fasta, ref_fasta, is_FL=False, same_strand_only=False, \
query_converted=True, db_converted=True, query_made=False, \
db_made=True, use_sge=False, cpus=cpus, sge_opts=None)
las_filenames, las_out_filenames = runner.runHPC(min_match_len=300, output_dir=output_dir, sensitive_mode=daligner_sensitive_mode)
if no_qv_or_aln_checking:
# not using QVs or alignment checking!
# this probqv is just a DUMMY to pass to daligner_against_ref, which won't be used
logging.info("Not using QV for partial_uc. Loading dummy QV.")
probqv = ProbFromModel(.01, .07, .06)
else:
if ccs_fofn is None:
logging.info("Loading probability from model (0.01,0.07,0.06)")
probqv = ProbFromModel(.01, .07, .06)
else:
start_t = time.time()
if use_finer_qv:
probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
logging.info("Loading QVs from {i} + {f} took {s} secs".format(f=ccs_fofn, i=input_fasta,\
s=time.time()-start_t))
else:
input_fastq = input_fasta[:input_fasta.rfind('.')] + '.fastq'
logging.info("Converting {i} + {f} --> {fq}".format(i=input_fasta, f=ccs_fofn, fq=input_fastq))
ice_fa2fq(input_fasta, ccs_fofn, input_fastq)
probqv = ProbFromFastq(input_fastq)
logging.info("Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t))
print >> sys.stderr, "Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t)
logging.info("Calling dalign_against_ref ...")
partial_uc = {} # Maps each isoform (cluster) id to a list of reads
# which can map to the isoform
seen = set() # reads seen
logging.info("Building uc from DALIGNER hits.")
for las_out_filename in las_out_filenames:
start_t = time.time()
hitItems = dalign_against_ref(input_obj, ref_obj, las_out_filename,
is_FL=False,
sID_starts_with_c=True,
qver_get_func=probqv.get_smoothed,
qvmean_get_func=probqv.get_mean,
ece_penalty=1,
ece_min_len=20,
same_strand_only=False,
no_qv_or_aln_checking=no_qv_or_aln_checking)
for h in hitItems:
if h.ece_arr is not None:
if h.cID not in partial_uc:
partial_uc[h.cID] = set()
partial_uc[h.cID].add(h.qID)
seen.add(h.qID)
logging.info("processing {0} took {1} sec".format(las_out_filename, time.time()-start_t))
print >> sys.stderr, "processing {0} took {1} sec".format(las_out_filename, time.time()-start_t)
for k in partial_uc:
partial_uc[k] = list(partial_uc[k])
allhits = set(r.name.split()[0] for r in FastaReader(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: {f}.".format(f=out_pickle))
with open(out_pickle, 'w') as f:
dump({'partial_uc': partial_uc, 'nohit': nohit}, f)
done_filename = realpath(done_filename) if done_filename is not None \
else out_pickle + '.DONE'
logging.debug("Creating {f}.".format(f=done_filename))
touch(done_filename)
# remove all the .las and .las.out filenames
for file in las_filenames:
os.remove(file)
for file in las_out_filenames:
os.remove(file)
def build_uc_from_partial(input_fasta, ref_fasta, out_pickle,
sa_file=None, ccs_fofn=None,
done_filename=None, blasr_nproc=12, use_finer_qv=False):
"""
Given an input_fasta file of non-full-length (partial) reads and
(unpolished) consensus isoforms sequences in ref_fasta, align reads to
consensus isoforms using BLASR, and then build up a mapping between
consensus isoforms and reads (i.e., assign reads to isoforms).
Finally, save
{isoform_id: [read_ids],
nohit: set(no_hit_read_ids)}
to an output pickle file.
ccs_fofn --- If None, assume no quality value is available,
otherwise, use QV from ccs_fofn.
blasr_nproc --- equivalent to blasr -nproc, number of CPUs to use
"""
input_fasta = realpath(input_fasta)
m5_file = input_fasta + ".blasr"
out_pickle = realpath(out_pickle)
if sa_file is None:
if op.exists(input_fasta + ".sa"):
sa_file = input_fasta + ".sa"
cmd = "blasr {i} ".format(i=real_upath(input_fasta)) + \
"{r} -bestn 5 ".format(r=real_upath(ref_fasta)) + \
"-nproc {n} -m 5 ".format(n=blasr_nproc) + \
"-maxScore -1000 -minPctIdentity 85 " + \
"-out {o} ".format(o=real_upath(m5_file))
if sa_file is not None and op.exists(sa_file):
cmd += "-sa {sa}".format(sa=real_upath(sa_file))
logging.info("CMD: {cmd}".format(cmd=cmd))
_out, _code, _msg = backticks(cmd)
if _code != 0:
errMsg = "Command failed: {cmd}\n{e}".format(cmd=cmd, e=_msg)
logging.error(errMsg)
raise RuntimeError(errMsg)
if ccs_fofn is None:
logging.info("Loading probability from model (0.01,0.07,0.06)")
probqv = ProbFromModel(.01, .07, .06)
else:
start_t = time.time()
if use_finer_qv:
logging.info("Loading QVs from {i} + {f} took {s} secs".format(f=ccs_fofn, i=input_fasta,\
s=time.time()-start_t))
probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
else:
input_fastq = input_fasta[:input_fasta.rfind('.')] + '.fastq'
logging.info("Converting {i} + {f} --> {fq}".format(i=input_fasta, f=ccs_fofn, fq=input_fastq))
ice_fa2fq(input_fasta, ccs_fofn, input_fastq)
logging.info("Loading QVs from {fq} took {s} secs".format(fq=input_fastq, s=time.time()-start_t))
probqv = ProbFromFastq(input_fastq)
logging.info("Calling blasr_against_ref ...")
hitItems = blasr_against_ref(output_filename=m5_file,
is_FL=False,
sID_starts_with_c=True,
qvmean_get_func=probqv.get_mean,
qver_get_func=probqv.get_smoothed,
ece_penalty=1,
ece_min_len=10,
same_strand_only=False)
partial_uc = {} # Maps each isoform (cluster) id to a list of reads
# which can map to the isoform
seen = set() # reads seen
logging.info("Building uc from BLASR hits.")
for h in hitItems:
if h.ece_arr is not None:
if h.cID not in partial_uc:
partial_uc[h.cID] = set()
partial_uc[h.cID].add(h.qID)
seen.add(h.qID)
for k in partial_uc:
partial_uc[k] = list(partial_uc[k])
allhits = set(r.name.split()[0] for r in FastaReader(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: {f}.".format(f=out_pickle))
with open(out_pickle, 'w') as f:
dump({'partial_uc': partial_uc, 'nohit': nohit}, f)
os.remove(m5_file)
done_filename = realpath(done_filename) if done_filename is not None \
else out_pickle + '.DONE'
logging.debug("Creating {f}.".format(f=done_filename))
touch(done_filename)
def build_uc_from_partial(input_fasta,
ref_fasta,
out_pickle,
sa_file=None,
ccs_fofn=None,
done_filename=None,
blasr_nproc=12):
"""Align consensus isoforms in ref_fasta and reads in input_fasta,
and save mappings between isoforms and reads to out_pickle.
ccs_fofn --- If None, assume no quality value is available,
otherwise, use QV from ccs_fofn.
blasr_nproc --- equivalent to blasr -nproc, number of CPUs to use
"""
input_fasta = realpath(input_fasta)
m5_file = input_fasta + ".blasr"
out_pickle = realpath(out_pickle)
if sa_file is None:
if op.exists(input_fasta + ".sa"):
sa_file = input_fasta + ".sa"
cmd = "blasr {i} ".format(i=input_fasta) + \
"{r} -bestn 5 ".format(r=ref_fasta) + \
"-nproc {n} -m 5 ".format(n=blasr_nproc) + \
"-maxScore -1000 -minPctIdentity 85 -out {o} ".format(o=m5_file)
if sa_file is not None and op.exists(sa_file):
cmd += "-sa {sa}".format(sa=sa_file)
logging.info("CMD: {cmd}".format(cmd=cmd))
_out, _code, _msg = backticks(cmd)
if _code != 0:
errMsg = "Command failed: {cmd}\n{e}".format(cmd=cmd, e=_msg)
logging.error(errMsg)
raise RuntimeError(errMsg)
if ccs_fofn is None:
logging.info("Loading probability from model")
probqv = ProbFromModel(.01, .07, .06)
else:
logging.info("Loading probability from QV in {f}".format(f=ccs_fofn))
probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
logging.info("Calling blasr_against_ref ...")
hitItems = blasr_against_ref(output_filename=m5_file,
is_FL=False,
sID_starts_with_c=True,
qver_get_func=probqv.get_smoothed,
ece_penalty=1,
ece_min_len=10,
same_strand_only=False)
partial_uc = {} # Maps each isoform (cluster) id to a list of reads
# which can map to the isoform
seen = set() # reads seen
logging.info("Building uc from BLASR hits.")
for h in hitItems:
if h.ece_arr is not None:
if h.cID not in partial_uc:
partial_uc[h.cID] = []
partial_uc[h.cID].append(h.qID)
seen.add(h.qID)
allhits = set(r.name.split()[0] for r in FastaReader(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: {f}.".format(f=out_pickle))
with open(out_pickle, 'w') as f:
dump({'partial_uc': partial_uc, 'nohit': nohit}, f)
os.remove(m5_file)
done_filename = realpath(done_filename) if done_filename is not None \
else out_pickle + '.DONE'
logging.debug("Creating {f}.".format(f=done_filename))
touch(done_filename)
def build_uc_from_partial(
input_fasta, ref_fasta, out_pickle, sa_file=None, ccs_fofn=None, done_filename=None, blasr_nproc=12
):
"""Align consensus isoforms in ref_fasta and reads in input_fasta,
and save mappings between isoforms and reads to out_pickle.
ccs_fofn --- If None, assume no quality value is available,
otherwise, use QV from ccs_fofn.
blasr_nproc --- equivalent to blasr -nproc, number of CPUs to use
"""
input_fasta = realpath(input_fasta)
m5_file = input_fasta + ".blasr"
out_pickle = realpath(out_pickle)
if sa_file is None:
if op.exists(input_fasta + ".sa"):
sa_file = input_fasta + ".sa"
cmd = (
"blasr {i} ".format(i=input_fasta)
+ "{r} -bestn 5 ".format(r=ref_fasta)
+ "-nproc {n} -m 5 ".format(n=blasr_nproc)
+ "-maxScore -1000 -minPctIdentity 85 -out {o} ".format(o=m5_file)
)
if sa_file is not None and op.exists(sa_file):
cmd += "-sa {sa}".format(sa=sa_file)
logging.info("CMD: {cmd}".format(cmd=cmd))
_out, _code, _msg = backticks(cmd)
if _code != 0:
errMsg = "Command failed: {cmd}\n{e}".format(cmd=cmd, e=_msg)
logging.error(errMsg)
raise RuntimeError(errMsg)
if ccs_fofn is None:
logging.info("Loading probability from model")
probqv = ProbFromModel(0.01, 0.07, 0.06)
else:
logging.info("Loading probability from QV in {f}".format(f=ccs_fofn))
probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
logging.info("Calling blasr_against_ref ...")
hitItems = blasr_against_ref(
output_filename=m5_file,
is_FL=False,
sID_starts_with_c=True,
qver_get_func=probqv.get_smoothed,
ece_penalty=1,
ece_min_len=10,
same_strand_only=False,
)
partial_uc = {} # Maps each isoform (cluster) id to a list of reads
# which can map to the isoform
seen = set() # reads seen
logging.info("Building uc from BLASR hits.")
for h in hitItems:
if h.ece_arr is not None:
if h.cID not in partial_uc:
partial_uc[h.cID] = []
partial_uc[h.cID].append(h.qID)
seen.add(h.qID)
allhits = set(r.name.split()[0] for r in FastaReader(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: {f}.".format(f=out_pickle))
with open(out_pickle, "w") as f:
dump({"partial_uc": partial_uc, "nohit": nohit}, f)
os.remove(m5_file)
done_filename = realpath(done_filename) if done_filename is not None else out_pickle + ".DONE"
logging.debug("Creating {f}.".format(f=done_filename))
touch(done_filename)