def test_read_number(self):
renamer = PairedEndRenamer("{id} read no. is: {rn}")
r1 = Sequence("theid cmtx", "ACGT")
r2 = Sequence("theid cmty", "ACGT")
info1 = ModificationInfo(r1)
info2 = ModificationInfo(r2)
renamed1, renamed2 = renamer(r1, r2, info1, info2)
assert renamed1.name == "theid read no. is: 1"
assert renamed2.name == "theid read no. is: 2"
def test_ids_not_identical(self):
renamer = PairedEndRenamer("{id} abc {comment} xyz")
r1 = Sequence("theid_a cmtx", "ACGT")
r2 = Sequence("theid_b cmty", "ACGT")
info1 = ModificationInfo(r1)
info2 = ModificationInfo(r2)
with pytest.raises(ValueError) as e:
renamer(r1, r2, info1, info2)
assert "not identical" in e.value.args[0]
def test_r2_comment(self):
renamer = PairedEndRenamer("{id} abc {r2.comment} xyz")
r1 = Sequence("theid cmtx", "ACGT")
r2 = Sequence("theid cmty", "ACGT")
info1 = ModificationInfo(r1)
info2 = ModificationInfo(r2)
renamed1, renamed2 = renamer(r1, r2, info1, info2)
assert renamed1.name == "theid abc cmty xyz"
assert renamed2.name == "theid abc cmty xyz"
def pipeline_from_parsed_args(args, paired, file_opener, adapters, adapters2) -> Pipeline:
"""
Setup a processing pipeline from parsed command-line arguments.
If there are any problems parsing the arguments, a CommandLineError is raised.
Return an instance of Pipeline (SingleEndPipeline or PairedEndPipeline)
"""
if args.action == 'none':
args.action = None
# Create the processing pipeline
if paired:
pair_filter_mode = 'any' if args.pair_filter is None else args.pair_filter
pipeline = PairedEndPipeline(
pair_filter_mode, file_opener
) # type: Any
else:
pipeline = SingleEndPipeline(file_opener)
# When adapters are being trimmed only in R1 or R2, override the pair filter mode
# as using the default of 'any' would regard all read pairs as untrimmed.
if isinstance(pipeline, PairedEndPipeline) and (not adapters2 or not adapters) and (
args.discard_untrimmed or args.untrimmed_output or args.untrimmed_paired_output):
pipeline.override_untrimmed_pair_filter = True
add_unconditional_cutters(pipeline, args.cut, args.cut2, paired)
pipeline_add = pipeline.add_both if paired else pipeline.add
if args.nextseq_trim is not None:
pipeline_add(NextseqQualityTrimmer(args.nextseq_trim, args.quality_base))
if args.quality_cutoff is not None:
cutoffs = parse_cutoffs(args.quality_cutoff)
pipeline_add(QualityTrimmer(cutoffs[0], cutoffs[1], args.quality_base))
add_adapter_cutter(
pipeline,
adapters,
adapters2,
paired,
args.pair_adapters,
args.action,
args.times,
args.reverse_complement,
not args.rename, # no "rc" suffix if --rename is used
args.index,
)
for modifier in modifiers_applying_to_both_ends_if_paired(args):
pipeline_add(modifier)
if args.rename and (args.prefix or args.suffix):
raise CommandLineError(
"Option --rename cannot be combined with --prefix (-x) or --suffix (-y)"
)
if args.rename and args.rename != "{header}":
try:
if paired:
pipeline.add_paired_modifier(PairedEndRenamer(args.rename))
else:
pipeline_add(Renamer(args.rename))
except InvalidTemplate as e:
raise CommandLineError(e)
# Set filtering parameters
# Minimum/maximum length
for attr in 'minimum_length', 'maximum_length':
param = getattr(args, attr)
if param is not None:
lengths = parse_lengths(param)
if not paired and len(lengths) == 2:
raise CommandLineError('Two minimum or maximum lengths given for single-end data')
if paired and len(lengths) == 1:
lengths = (lengths[0], lengths[0])
setattr(pipeline, attr, lengths)
pipeline.max_n = args.max_n
pipeline.max_expected_errors = args.max_expected_errors
pipeline.discard_casava = args.discard_casava
pipeline.discard_trimmed = args.discard_trimmed
pipeline.discard_untrimmed = args.discard_untrimmed
return pipeline
def test_invalid_template_variable(self):
with pytest.raises(InvalidTemplate):
PairedEndRenamer("{id} {invalid}")