def set_probqv_from_fq(fastq_filename):
"""Set probability and QVs from FASTQ, return probqv, log_info."""
assert isinstance(fastq_filename, str)
start_t = time.time()
probqv = ProbFromFastq(fastq_filename=fastq_filename)
msg = "Loading QVs from {f} took {t} sec.".\
format(f=fastq_filename, t=(time.time()-start_t))
return probqv, msg
def pickup_icec_job(pickle_filename, ccs_fofn, flnc_filename,
fasta_files_to_add, root_dir):
"""
Reconstruct an ICE object from a pickle file and restart this ICE job.
"""
log.info("Reading ICE pickle %s ....", pickle_filename)
icec_obj, icec_pickle_filename = ensure_pickle_goodness(
pickle_filename=pickle_filename,
root_dir=root_dir,
fasta_files_to_add=fasta_files_to_add)
c_fa = current_fasta(root_dir)
c_fq = current_fastq(root_dir)
log.info("Making current.fasta %s for ICE ....", c_fa)
make_current_fasta(icec_obj=icec_obj,
flnc_filename=flnc_filename,
root_dir=root_dir)
log.info("Loading prob QV information....")
probqv = None
if ccs_fofn is None:
logging.info("Loading probability from model (0.01,0.07,0.06)")
probqv = ProbFromModel(.01, .07, .06)
else:
#if use_finer_qv:
# probqv = ProbFromQV(input_fofn=ccs_fofn, fasta_filename=input_fasta)
# logging.info("Loading prob QVs from %s + %s took %s secs",
# ccs_fofn, input_fasta, time.time()-start_t)
logging.info("Converting %s to %s", c_fa, c_fq)
ice_fa2fq(c_fa, ccs_fofn, c_fq)
logging.info("Loading prob QVs from %s", c_fq)
probqv = ProbFromFastq(c_fq)
log.info("Starting ICE from pickle %s....", icec_pickle_filename)
icec = ice.IceIterative.from_pickle(icec_pickle_filename, probqv)
# first must RE-RUN gcon to get all the proper refs
icec.changes = set()
icec.refs = {}
icec.ccs_fofn = ccs_fofn
icec.all_fasta_filename = flnc_filename
todo = icec.uc.keys()
log.info("Re-run gcon for proper refs....")
icec.run_gcon_parallel(todo)
log.info("Re-calculating cluster prob, just to be safe....")
icec.calc_cluster_prob(True)
log.info("Sanity checking uc_refs now....")
icec.sanity_check_uc_refs()
log.info("Ensuring prob QV of new ids are consistent....")
icec.ensure_probQV_newid_consistency()
log.info("Sanity check done. Resuming ICE job.")
icec.run()
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,
tmp_dir=None):
"""
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.
tmp_dir - where to save intermediate files such as dazz files.
if None, writer dazz files to the same directory as query/target.
"""
input_fasta = realpath(input_fasta)
ref_fasta = realpath(ref_fasta)
out_pickle = realpath(out_pickle)
output_dir = op.dirname(out_pickle)
ice_opts = IceOptions()
ice_opts.detect_cDNA_size(ref_fasta)
# ice_partial is already being called through qsub, so run everything local!
runner = DalignerRunner(query_filename=input_fasta,
target_filename=ref_fasta,
is_FL=False,
same_strand_only=False,
query_converted=False,
target_converted=True,
dazz_dir=tmp_dir,
script_dir=op.join(output_dir, "script"),
use_sge=False,
sge_opts=None,
cpus=cpus)
runner.run(min_match_len=300,
output_dir=output_dir,
sensitive_mode=ice_opts.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 %s + %s took %s secs", ccs_fofn,
input_fasta,
time.time() - start_t)
else:
input_fastq = input_fasta[:input_fasta.rfind('.')] + '.fastq'
logging.info("Converting %s + %s --> %s", input_fasta,
ccs_fofn, input_fastq)
ice_fa2fq(input_fasta, ccs_fofn, input_fastq)
probqv = ProbFromFastq(input_fastq)
logging.info("Loading QVs from %s took %s secs", input_fastq,
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 la4ice_filename in runner.la4ice_filenames:
start_t = time.time()
hitItems = daligner_against_ref(
query_dazz_handler=runner.query_dazz_handler,
target_dazz_handler=runner.target_dazz_handler,
la4ice_filename=la4ice_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 %s took %s sec", la4ice_filename,
str(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 ContigSetReaderWrapper(input_fasta))
logging.info("Counting reads with no hit.")
nohit = allhits.difference(seen)
logging.info("Dumping uc to a pickle: %s.", out_pickle)
with open(out_pickle, 'w') as f:
if out_pickle.endswith(".pickle"):
dump({'partial_uc': partial_uc, 'nohit': nohit}, f)
elif out_pickle.endswith(".json"):
f.write(json.dumps({'partial_uc': partial_uc, 'nohit': nohit}))
else:
raise IOError("Unrecognized extension: %s" % out_pickle)
done_filename = realpath(done_filename) if done_filename is not None \
else out_pickle + '.DONE'
logging.debug("Creating %s.", done_filename)
touch(done_filename)
# remove all the .las and .las.out filenames
runner.clean_run()