gcon_nproc=True,
arrow_nproc=False)
args = parser.parse_args()
# currently, the user is NOT allowed to set the definition for full missed start/end
# (i.e. how much of the query must be mapped to consider it a hit)
ice_opts = IceOptions2(
ece_penalty=args.ece_penalty,
ece_min_len=args.ece_min_len,
max_missed_start=args.max_missed_start,
max_missed_end=args.max_missed_end,
full_missed_start=30,
full_missed_end=20,
aligner_choice=args.aligner_choice,
)
unique_id = random.randint(1, 1000000)
sge_opts = SgeOptions2(unique_id,
use_sge=args.use_sge,
max_sge_jobs=args.max_sge_jobs,
blasr_nproc=args.blasr_nproc,
gcon_nproc=args.gcon_nproc,
sge_queue=args.sge_queue,
sge_env_name=args.sge_env_name,
qsub_extra=args.qsub_extra)
run_IceIterative2(args.fasta_filenames.split(','),
args.fastq_filenames.split(','), args.all_fasta_filename,
ice_opts, sge_opts, args.root_dir, args.init_uc_pickle)
def run(self):
"""Execute ice_partial.py all|one|split|i|merge."""
cmd = self.args.subCommand
logging.info("Running {f} {cmd} v{v}.".format(f=op.basename(__file__),
cmd=cmd,
v=get_version()))
cmd_str = ""
try:
args = self.args
obj = None
if cmd in ('all', 'one'):
# currently user NOT allowed to set full missed start/end
# (we also set it to 30/10 bp which is more stringent than in IceIterative2,
# which is hard set to 50/30)
ice_opts = IceOptions2(ece_penalty=args.ece_penalty,
ece_min_len=args.ece_min_len,
max_missed_start=args.max_missed_start,
max_missed_end=args.max_missed_end,
full_missed_start=30,
full_missed_end=10,
min_match_len=50,
aligner_choice=args.aligner_choice)
if cmd == "all":
sge_opts = SgeOptions2(unique_id=args.unique_id,
use_sge=args.use_sge,
max_sge_jobs=args.max_sge_jobs,
sge_queue=args.sge_queue,
sge_env_name=args.sge_env_name,
qsub_extra=args.qsub_extra)
obj = IceAllPartials2(
root_dir=args.root_dir,
fasta_filenames=args.fasta_filenames.split(','),
fastq_filenames=args.fastq_filenames.split(',')
if args.fastq_filenames is not None else None,
ref_fasta=args.ref_fasta,
out_pickle=args.out_pickle,
ice_opts=ice_opts,
sge_opts=sge_opts,
cpus=args.cpus,
tmp_dir=args.tmp_dir)
elif cmd == "one":
# Only assign nfl reads in the given input_fasta file to isoforms
# "one" is always run locally so no need for SGE option
obj = IcePartialOne2(input_fasta=args.input_fasta,
input_fastq=args.input_fastq,
ref_fasta=args.ref_fasta,
out_pickle=args.out_pickle,
done_filename=args.done_filename,
ice_opts=ice_opts,
cpus=args.cpus,
tmp_dir=args.tmp_dir)
elif cmd == "split":
obj = IcePartialSplit2(root_dir=args.root_dir,
nfl_fa=args.nfl_fa,
nfl_fq=args.nfl_fq,
N=args.N)
# elif cmd == "i":
# obj = IcePartialI(root_dir=args.root_dir, i=args.i,
# ccs_fofn=args.ccs_fofn,
# blasr_nproc=args.blasr_nproc,
# tmp_dir=args.tmp_dir)
elif cmd == "merge":
obj = IcePartialMerge(root_dir=args.root_dir, N=args.N)
else:
raise ValueError(
"Unknown command passed to {f}: {cmd}.".format(
f=op.basename(__file__), cmd=cmd))
cmd_str = obj.cmd_str()
logging.info("Running CMD: {cmd_str}".format(cmd_str=cmd_str))
obj.run()
except:
logging.exception("Exiting {cmd_str} with return code 1.".format(
cmd_str=cmd_str))
return 1
return 0
type=int,
help="ECE min len (default: 20)")
parser.add_argument("--max_missed_start",
default=100,
type=int,
help="Max missed 5'(default: 100 bp)")
parser.add_argument("--max_missed_end",
default=30,
type=int,
help="Max missed 3' (default: 30 bp)")
parser.add_argument("--cpus",
default=4,
help="Number of CPUs aligner uses (default: 4)")
args = parser.parse_args()
ice_opts = IceOptions2(ece_penalty=args.ece_penalty,
ece_min_len=args.ece_min_len,
max_missed_start=args.max_missed_start,
max_missed_end=args.max_missed_end,
full_missed_start=30,
full_missed_end=20)
sge_opts = SgeOptions2(unique_id=123, blasr_nproc=args.cpus)
# set min_match_len to (low_cDNA_size-max_missed_start-max_missed_end), rounded to lowest 100 bp
ice_opts.detect_cDNA_size(args.flnc_fa)
x = ice_opts.low_cDNA_size - ice_opts.max_missed_start - ice_opts.max_missed_end
ice_opts.min_match_len = max(50, (x / 100) * 100)
run_IceInit2(args.flnc_fa, args.output_pickle, ice_opts, sge_opts)
def run(self):
"""Execute ice_quiver.py all|i|merge|postprocess."""
cmd = self.args.subCommand
logging.info("Running {f} {cmd} v{v}.".format(f=op.basename(__file__),
cmd=cmd,
v=get_version()))
cmd_str = ""
try:
args = self.args
obj = None
if cmd == "all":
sge_opts = SgeOptions2(unique_id=args.unique_id,
use_sge=args.use_sge,
max_sge_jobs=args.max_sge_jobs,
blasr_nproc=args.blasr_nproc,
arrow_nproc=args.arrow_nproc,
sge_env_name=args.sge_env_name,
sge_queue=args.sge_queue,
qsub_extra=args.qsub_extra)
ipq_opts = IceArrowHQLQOptions2(
hq_isoforms_fa=args.hq_isoforms_fa,
hq_isoforms_fq=args.hq_isoforms_fq,
lq_isoforms_fa=args.lq_isoforms_fa,
lq_isoforms_fq=args.lq_isoforms_fq,
qv_trim_5=args.qv_trim_5,
qv_trim_3=args.qv_trim_3,
hq_arrow_min_accuracy=args.hq_arrow_min_accuracy)
obj = IceArrowAll2(root_dir=args.root_dir,
subread_xml=args.subread_xml,
sge_opts=sge_opts,
ipq_opts=ipq_opts,
tmp_dir=args.tmp_dir)
# elif cmd == "i":
# sge_opts = SgeOptions(unique_id=args.unique_id,
# use_sge=args.use_sge,
# max_sge_jobs=args.max_sge_jobs,
# blasr_nproc=args.blasr_nproc,
# arrow_nproc=args.arrow_nproc)
# obj = IceQuiverI(root_dir=args.root_dir, i=args.i, N=args.N,
# bas_fofn=args.bas_fofn,
# fasta_fofn=None,
# sge_opts=sge_opts,
# tmp_dir=args.tmp_dir)
#elif cmd == "merge":
# obj = IceQuiverMerge(root_dir=args.root_dir, N=args.N)
elif cmd == "postprocess":
ipq_opts = IceArrowHQLQOptions2(
hq_isoforms_fa=args.hq_isoforms_fa,
hq_isoforms_fq=args.hq_isoforms_fq,
lq_isoforms_fa=args.lq_isoforms_fa,
lq_isoforms_fq=args.lq_isoforms_fq,
qv_trim_5=args.qv_trim_5,
qv_trim_3=args.qv_trim_3,
hq_arrow_min_accuracy=args.hq_arrow_min_accuracy,
hq_min_full_length_reads=args.hq_min_full_length_reads)
obj = IceArrowPostProcess2(
root_dir=args.root_dir,
ipq_opts=ipq_opts,
quit_if_not_done=args.quit_if_not_done,
summary_fn=args.summary_fn,
report_fn=args.report_fn)
else:
raise ValueError(
"Unknown command passed to {f}: {cmd}.".format(
f=op.basename(__file__), cmd=cmd))
cmd_str = obj.cmd_str()
logging.info("Running CMD: {cmd_str}".format(cmd_str=cmd_str))
obj.run()
except:
logging.exception("Exiting {cmd_str} with return code 1.".format(
cmd_str=cmd_str))
return 1
return 0