def test_run(self):
"""Test function run()."""
inFN = op.join(self.testDir, "data/test_subset.fa")
outFN = op.join(self.testDir, "out/test_subset_unit.fa")
stdoutFN = op.join(self.testDir, "stdout/test_subset_unit.fa")
rules = SubsetRules(1, 1) # Full-length, non-chimeric
obj = ReadsSubsetExtractor(inFN, outFN, rules, True)
obj.run()
self.assertTrue(filecmp.cmp(outFN, stdoutFN))
def test_satisfy(self):
"""Test function satisfy()."""
inFN = op.join(self.testDir, "data/test_subset.fa")
reads = []
with FastaReader(inFN) as reader:
reads = [x for x in reader]
rules = SubsetRules(1, 1) # Full-length, non-chimeric
obj = ReadsSubsetExtractor("in", "out", rules, True)
ans = [ReadAnnotation.fromString(r.name) for r in reads]
res = [obj.satisfy(an, rules) for an in ans]
expected = [1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 1, 1]
self.assertTrue(res == expected)
def run(self):
"""Run classify, cluster, polish or subset."""
cmd = self.args.subCommand
try:
if cmd == 'classify':
opts = ChimeraDetectionOptions(
min_seq_len=self.args.min_seq_len,
min_score=self.args.min_score,
min_dist_from_end=self.args.min_dist_from_end,
max_adjacent_hit_dist=self.args.max_adjacent_hit_dist,
primer_search_window=self.args.primer_search_window,
detect_chimera_nfl=self.args.detect_chimera_nfl)
obj = Classifier(reads_fn=self.args.readsFN,
out_dir=self.args.outDir,
out_reads_fn=self.args.outReadsFN,
primer_fn=self.args.primerFN,
primer_report_fn=self.args.primerReportFN,
summary_fn=self.args.summary_fn,
cpus=self.args.cpus,
change_read_id=True,
opts=opts,
out_flnc_fn=self.args.flnc_fa,
out_nfl_fn=self.args.nfl_fa,
ignore_polyA=self.args.ignore_polyA,
reuse_dom=self.args.reuse_dom)
obj.run()
elif cmd == 'cluster':
ice_opts = IceOptions(cDNA_size=self.args.cDNA_size,
quiver=self.args.quiver,
use_finer_qv=self.args.use_finer_qv)
sge_opts = SgeOptions(unique_id=self.args.unique_id,
use_sge=self.args.use_sge,
max_sge_jobs=self.args.max_sge_jobs,
blasr_nproc=self.args.blasr_nproc,
quiver_nproc=self.args.quiver_nproc)
ipq_opts = IceQuiverHQLQOptions(qv_trim_5=self.args.qv_trim_5,
qv_trim_3=self.args.qv_trim_3,
hq_quiver_min_accuracy=self.args.hq_quiver_min_accuracy,
hq_isoforms_fa=self.args.hq_isoforms_fa,
hq_isoforms_fq=self.args.hq_isoforms_fq,
lq_isoforms_fa=self.args.lq_isoforms_fa,
lq_isoforms_fq=self.args.lq_isoforms_fq)
obj = Cluster(root_dir=self.args.root_dir,
flnc_fa=self.args.flnc_fa,
nfl_fa=self.args.nfl_fa,
bas_fofn=self.args.bas_fofn,
ccs_fofn=self.args.ccs_fofn,
fasta_fofn=self.args.fasta_fofn,
out_fa=self.args.consensusFa,
sge_opts=sge_opts,
ice_opts=ice_opts,
ipq_opts=ipq_opts,
report_fn=self.args.report_fn,
summary_fn=self.args.summary_fn,
nfl_reads_per_split=self.args.nfl_reads_per_split)
obj.run()
elif cmd == 'subset':
rules = SubsetRules(FL=self.args.FL,
nonChimeric=self.args.nonChimeric)
obj = ReadsSubsetExtractor(inFN=self.args.readsFN,
outFN=self.args.outFN,
rules=rules,
ignore_polyA=self.args.ignore_polyA,
printReadLengthOnly=self.args.printReadLengthOnly)
obj.run()
else:
raise PBTranscriptException(cmd,
"Unknown command passed to pbtranscript.py:"
+ self.args.subName)
except Exception:
logging.exception("Exiting pbtranscript with return code 1.")
return 1
return 0
def run(self):
"""Run classify, cluster, polish or subset."""
cmd = self.args.subCommand
try:
if cmd == 'classify':
opts = ChimeraDetectionOptions(
min_seq_len=self.args.min_seq_len,
min_score=self.args.min_score,
min_dist_from_end=self.args.min_dist_from_end,
max_adjacent_hit_dist=self.args.max_adjacent_hit_dist,
primer_search_window=self.args.primer_search_window)
obj = Classifier(reads_fn=self.args.readsFN,
out_dir=self.args.outDir,
out_reads_fn=self.args.outReadsFN,
primer_fn=self.args.primerFN,
primer_report_fn=self.args.primerReportFN,
summary_fn=self.args.summary_fn,
cpus=self.args.cpus,
change_read_id=True,
opts=opts,
out_flnc_fn=self.args.flnc_fa,
out_nfl_fn=self.args.nfl_fa,
ignore_polyA=self.args.ignore_polyA)
obj.run()
elif cmd == 'cluster':
ice_opts = IceOptions(cDNA_size=self.args.cDNA_size,
quiver=self.args.quiver)
sge_opts = SgeOptions(unique_id=self.args.unique_id,
use_sge=self.args.use_sge,
max_sge_jobs=self.args.max_sge_jobs,
blasr_nproc=self.args.blasr_nproc,
quiver_nproc=self.args.quiver_nproc)
obj = Cluster(root_dir=self.args.root_dir,
flnc_fa=self.args.flnc_fa,
nfl_fa=self.args.nfl_fa,
bas_fofn=self.args.bas_fofn,
ccs_fofn=self.args.ccs_fofn,
out_fa=self.args.consensusFa,
sge_opts=sge_opts,
ice_opts=ice_opts,
hq_isoforms_fa=self.args.hq_isoforms_fa,
hq_isoforms_fq=self.args.hq_isoforms_fq,
lq_isoforms_fa=self.args.lq_isoforms_fa,
lq_isoforms_fq=self.args.lq_isoforms_fq,
report_fn=self.args.report_fn,
summary_fn=self.args.summary_fn)
obj.run()
elif cmd == 'subset':
rules = SubsetRules(FL=self.args.FL,
nonChimeric=self.args.nonChimeric)
obj = ReadsSubsetExtractor(
inFN=self.args.readsFN,
outFN=self.args.outFN,
rules=rules,
ignore_polyA=self.args.ignore_polyA,
printReadLengthOnly=self.args.printReadLengthOnly)
obj.run()
else:
raise PBTranscriptException(
cmd, "Unknown command passed to pbtranscript.py:" +
self.args.subName)
except Exception as err:
logging.error(str(err))
return 1
return 0