def stipulate(args):
"""
REQUIRED TO CREATE ITERABLE FUNCTIONS TO RUN IN CUTADAPT 2.7. THIS FUNCTION IS CALLED ONLY ONE TIME.
"""
modifiers = []
pipeline_add = modifiers.append
adapter_parser = AdapterParser(
max_error_rate=args.error_rate,
min_overlap=args.overlap,
read_wildcards=args.match_read_wildcards,
adapter_wildcards=args.match_adapter_wildcards,
indels=args.indels,
)
adapters = adapter_parser.parse_multi(args.adapters)
warn_duplicate_adapters(adapters)
if args.nextseq_trim is not None:
pipeline_add(NextseqQualityTrimmer(args.nextseq_trim, args.phred64))
if args.quality_cutoff is not None:
cutoffs = parse_cutoffs(args.quality_cutoff)
pipeline_add(QualityTrimmer(cutoffs[0], cutoffs[1], args.phred64))
adapter_cutter = None
if adapters:
adapter_cutter = AdapterCutter(adapters, args.times, args.action)
pipeline_add(adapter_cutter)
if args.trim_n:
pipeline_add(NEndTrimmer())
add_unconditional_cutters(pipeline_add, args.cut)
print("modifiers (cutadapt):", modifiers)
return modifiers
def adapters_from_args(args) -> Tuple[List[Adapter], List[Adapter]]:
adapter_parser = AdapterParser(
max_errors=args.error_rate,
min_overlap=args.overlap,
read_wildcards=args.match_read_wildcards,
adapter_wildcards=args.match_adapter_wildcards,
indels=args.indels,
)
try:
adapters = adapter_parser.parse_multi(args.adapters)
adapters2 = adapter_parser.parse_multi(args.adapters2)
except (FileNotFoundError, ValueError, InvalidCharacter) as e:
raise CommandLineError(e)
warn_duplicate_adapters(adapters)
warn_duplicate_adapters(adapters2)
if args.debug == "trace":
for adapter in adapters + adapters2:
adapter.enable_debug()
return adapters, adapters2
def pipeline_from_parsed_args(args, paired, is_interleaved_output):
"""
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)
"""
check_arguments(args, paired, is_interleaved_output)
if args.action == 'none':
args.action = None
adapter_parser = AdapterParser(
max_error_rate=args.error_rate,
min_overlap=args.overlap,
read_wildcards=args.match_read_wildcards,
adapter_wildcards=args.match_adapter_wildcards,
indels=args.indels,
)
try:
adapters = adapter_parser.parse_multi(args.adapters)
adapters2 = adapter_parser.parse_multi(args.adapters2)
except (FileNotFoundError, ValueError) as e:
raise CommandLineError(e)
warn_duplicate_adapters(adapters)
warn_duplicate_adapters(adapters2)
if args.debug:
for adapter in adapters + adapters2:
adapter.enable_debug()
# 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)
else:
pipeline = SingleEndPipeline()
# 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 paired 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))
if args.pair_adapters:
try:
cutter = PairedAdapterCutter(adapters, adapters2, args.action)
except PairedAdapterCutterError as e:
raise CommandLineError("--pair-adapters: " + str(e))
pipeline.add_paired_modifier(cutter)
else:
adapter_cutter, adapter_cutter2 = None, None
if adapters:
adapter_cutter = AdapterCutter(adapters, args.times, args.action)
if adapters2:
adapter_cutter2 = AdapterCutter(adapters2, args.times, args.action)
if paired:
if adapter_cutter or adapter_cutter2:
pipeline.add(adapter_cutter, adapter_cutter2)
else:
if adapter_cutter:
pipeline.add(adapter_cutter)
for modifier in modifiers_applying_to_both_ends_if_paired(args):
pipeline_add(modifier)
# 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.discard_casava = args.discard_casava
pipeline.discard_trimmed = args.discard_trimmed
pipeline.discard_untrimmed = args.discard_untrimmed
return pipeline