def convert_guppy_result(guppy_result):
guppy_barcode = guppy_result['barcode']
if guppy_barcode['name'] == 'unclassified':
guppy_barcode = None
guppy_adapter = None
else:
guppy_barcode = read_barcode(guppy_barcode)
guppy_adapter = AdapterLayout(
kit=BarcodeScannerGuppy.KIT_MAPPING_REV.get(guppy_result['kit'],
guppy_result['kit']),
sequence="ATGC",
barcode_set_1=None,
barcode_set_2=None,
description="")
result = build_return_dict(
best_barcode=guppy_barcode,
best_barcode_score=guppy_result['score'],
best_adapter=guppy_adapter,
best_adapter_end=90,
trim5p=guppy_result['trim5p'],
trim3p=guppy_result['trim3p'],
exit_status=guppy_result['exit_status'],
)
return result
def scan(self,
read_sequence,
read_qualities,
bc_adapter_templates,
nobc_adapter_templates,
qcat_config=config.qcatConfig()):
"""
Aligns all passed barcodes to the read sequence and chooses the one
with the highest alignment score. Simplest version of barcode detection
that is independent of the sequencing kit but has higher FN, FP and
incorrect rate for barcode assignments. Currently does not implement any
identity thresholds. Therefor, filtering by quality score is required.
:param read_sequence: Sequence of the read
:type read_sequence: str
:param read_qualities: Base qualities of the read in Sanger encoding
:type read_qualities: str
:param barcode_set: List of possible Barcode tuples
:type barcode_set: List
:param qcat_config: qcatConfig object
:type qcat_config: qcatConfig
:return: see build_return_dict
:rtype: Dictionary
"""
exit_status = 0
best_barcode, best_barcode_q_score, identity, barcode_end = \
find_highest_scoring_barcode(barcode_region_read=read_sequence,
barcode_set=self.barcodes,
qcat_config=qcat_config,
compute_identity=True)
# barcode_err_probe = 0.0
# if best_barcode:
# barcode_err_probe = compute_adapter_error_prob(read_qualities,
# barcode_end,
# len(
# best_barcode.sequence))
# If identity is too low or error prob too high, report as unclassified
if identity < self.min_quality:
return empty_return_dict()
return build_return_dict(best_barcode=best_barcode,
best_barcode_score=best_barcode_q_score,
best_adapter=None,
best_adapter_end=barcode_end,
exit_status=exit_status)
def detect_barcode(self,
read_sequence,
read_qualities=None,
qcat_config=config.qcatConfig()):
result = empty_return_dict()
if not self.override_kit_name:
detected_adapter, _ = self.scan_ends(read_sequence, qcat_config)
override_kit_name = detected_adapter.kit
else:
override_kit_name = self.override_kit_name
if override_kit_name in self.brill_models:
brill_model, brill_model_len = self.brill_models[override_kit_name]
if len(read_sequence) > brill_model_len:
bc, qscore = self.run_brill(brill_model, brill_model_len, read_sequence)
barcode = None
adapter = None
if bc and bc != 'none':
adapter = self.get_adapter(override_kit_name)
barcode = adapter.get_barcode_set()[int(bc) - 1]
if bc and qscore >= self.min_quality:
result = build_return_dict(
best_barcode=barcode,
best_barcode_score=qscore,
best_adapter=adapter,
best_adapter_end=90,
trim5p=90,
trim3p=90,
exit_status=0
)
return result
def scan(self,
read_sequence,
read_qualities,
bc_adapter_templates,
nobc_adapter_templates,
qcat_config=config.qcatConfig()):
"""
Detects sequencing adapter and identifies best matching barcode
:param read_sequence: Read sequence containing adapter
:type read_sequence: str
:param read_qualities: Base qualities of the read in Sanger encoding
:type read_qualities: str
:param adapter_templates: List of AdapterLayout objects
with Ns
:type adapter_templates: List
:param qcat_config: qcatConfig object
:type qcat_config: qcatConfig
:return: see build_return_dict
:rtype: Dictionary
"""
exit_status = 0
# Finding best barcoded adapters
ret = find_best_adapter_template(adapter_templates=bc_adapter_templates,
read_sequence=read_sequence,
qcat_config=qcat_config)
best_adapter_template_index, aligned_adapter_end, best_adapter_score = ret
# if barcoded adapter found
best_adapter_template = bc_adapter_templates[
best_adapter_template_index]
# Detect barcode
barcode_set_index = 0
# For high quality adapter alignments just use the barcode region,
# for low quality compare full adapter to the barcodes
# Do not scann full adapter is it contains double barcoding
# best_adapter_score > 90.0 or
if best_adapter_score > 90.0 or best_adapter_template.is_doulbe_barcode():
barcode_region_read = extract_barcode_region(
read_sequence=read_sequence,
adapter_template=best_adapter_template,
barcode_set_index=barcode_set_index,
alignment_stop_ref=aligned_adapter_end,
qcat_config=qcat_config)
else:
barcode_region_read = read_sequence[:qcat_config.max_align_length]
# First barcode
up_context = best_adapter_template.get_upstream_context(
qcat_config.barcode_context_length, barcode_set_index)
down_context = best_adapter_template.get_downstream_context(
qcat_config.barcode_context_length, barcode_set_index)
barcode_set = best_adapter_template.get_barcode_set(
barcode_set_index)
if self.barcodes:
barcode_set = self.barcodes
# Find best barcode
best_barcode, best_barcode_q_score, best_barcode_score, barcode_end = \
find_highest_scoring_barcode(
barcode_region_read=barcode_region_read,
barcode_set=barcode_set,
upstream_context=up_context,
downstream_context=down_context,
qcat_config=qcat_config)
# If double barcode adapter
barcode_set_index = 1
if best_adapter_template.get_barcode_set(barcode_set_index):
barcode_region_read = extract_barcode_region(
read_sequence=read_sequence,
adapter_template=best_adapter_template,
barcode_set_index=barcode_set_index,
alignment_stop_ref=aligned_adapter_end,
qcat_config=qcat_config)
# First barcode
up_context = best_adapter_template.get_upstream_context(
qcat_config.barcode_context_length, barcode_set_index)
down_context = best_adapter_template.get_downstream_context(
qcat_config.barcode_context_length, barcode_set_index)
barcode_set = best_adapter_template.get_barcode_set(
barcode_set_index)
if self.barcodes:
barcode_set = self.barcodes
# Find best barcode
best_barcode_2, best_barcode_q_score_2, best_barcode_score_2, barcode_end_2 = \
find_highest_scoring_barcode(
barcode_region_read=barcode_region_read,
barcode_set=barcode_set,
upstream_context=up_context,
downstream_context=down_context,
qcat_config=qcat_config)
else:
logging.debug("Adapter type does not have second barcode")
return build_return_dict(
best_barcode=best_barcode,
best_barcode_score=best_barcode_q_score,
best_adapter=best_adapter_template,
best_adapter_end=aligned_adapter_end,
exit_status=exit_status
)
