def _combine_multiple_svcallers(samples):
"""
"""
by_bam = collections.OrderedDict()
for x in samples:
try:
by_bam[x[0]["align_bam"]].append(x[0])
except KeyError:
by_bam[x[0]["align_bam"]] = [x[0]]
highdepths = filter(
lambda x: x is not None,
list(
set([
tz.get_in(["config", "algorithm", "highdepth_regions"], x[0])
for x in samples
])))
out = []
for grouped_calls in by_bam.values():
def orig_svcaller_order(x):
return _get_svcallers(x).index(
x["config"]["algorithm"]["svcaller_active"])
sorted_svcalls = sorted([x for x in grouped_calls if "sv" in x],
key=orig_svcaller_order)
final_calls = reduce(operator.add, [x["sv"] for x in sorted_svcalls])
final = grouped_calls[0]
final_calls = ensemble.summarize(final_calls, final, highdepths)
final_calls = validate.evaluate(final, final_calls)
final["sv"] = final_calls
del final["config"]["algorithm"]["svcaller_active"]
out.append([final])
return out
def detect_sv(items, all_items=None, stage="standard"):
"""Top level parallel target for examining structural variation.
items = sample-sv_caller list, from one batch
"""
items = [utils.to_single_data(x) for x in items]
items = cwlutils.unpack_tarballs(items, items[0])
svcaller = items[0]["config"]["algorithm"].get("svcaller")
caller_fn = _get_callers(items, stage, special_cases=True).get(svcaller)
out = []
batch = dd.get_batch(items[0])
# no SV calling when just creating a PON for PureCN
if batch == "pon_build" and "purecn" in dd.get_svcaller(items[0]):
return out
if svcaller and caller_fn:
if (all_items and svcaller in _NEEDS_BACKGROUND
and not vcfutils.is_paired_analysis(
[x.get("align_bam") for x in items], items)):
names = set([dd.get_sample_name(x) for x in items])
background = [
x for x in all_items if dd.get_sample_name(x) not in names
]
for svdata in caller_fn(items, background):
out.append([svdata])
else:
for svdata in caller_fn(items):
out.append([svdata])
else:
for data in items:
out.append([data])
# Avoid nesting of callers for CWL runs for easier extraction
if cwlutils.is_cwl_run(items[0]):
out_cwl = []
for data in [utils.to_single_data(x) for x in out]:
# Run validation directly from CWL runs since we're single stage
data = validate.evaluate(data)
data["svvalidate"] = {
"summary": tz.get_in(["sv-validate", "csv"], data)
}
svs = data.get("sv")
if svs:
assert len(svs) == 1, svs
data["sv"] = svs[0]
else:
data["sv"] = {}
data = _add_supplemental(data)
out_cwl.append([data])
return out_cwl
return out
def finalize_sv(samples, config, initial_only=False):
"""Combine results from multiple sv callers into a single ordered 'sv' key.
Handles ensemble calling and plotting of results.
"""
by_bam = collections.OrderedDict()
for x in samples:
try:
by_bam[x["align_bam"]].append(x)
except KeyError:
by_bam[x["align_bam"]] = [x]
by_batch = collections.OrderedDict()
lead_batches = {}
for grouped_calls in by_bam.values():
def orig_svcaller_order(x):
return _get_svcallers(x).index(x["config"]["algorithm"]["svcaller_active"])
sorted_svcalls = sorted([x for x in grouped_calls if "sv" in x], key=orig_svcaller_order)
final = grouped_calls[0]
if len(sorted_svcalls) > 0:
final_calls = reduce(operator.add, [x["sv"] for x in sorted_svcalls])
if not initial_only:
for caller in (c for c in _get_svcallers(final) if c in _ENSEMBLE_CALLERS):
final_calls = _ENSEMBLE_CALLERS[caller](final_calls, final)
final_calls = ensemble.summarize(final_calls, final, grouped_calls)
final_calls = validate.evaluate(final, final_calls)
final["sv"] = final_calls
del final["config"]["algorithm"]["svcaller_active"]
batch = dd.get_batch(final) or dd.get_sample_name(final)
batches = batch if isinstance(batch, (list, tuple)) else [batch]
lead_batches[dd.get_sample_name(final)] = batches[0]
for batch in batches:
try:
by_batch[batch].append(final)
except KeyError:
by_batch[batch] = [final]
out = []
for batch, items in by_batch.items():
if any("svplots" in dd.get_tools_on(d) for d in items):
plot_items = plot.by_regions(items)
else:
plot_items = items
for data in plot_items:
if lead_batches[dd.get_sample_name(data)] == batch:
out.append([data])
return out
def finalize_sv(samples, config, initial_only=False):
"""Combine results from multiple sv callers into a single ordered 'sv' key.
Handles ensemble calling and plotting of results.
"""
by_bam = collections.OrderedDict()
for x in samples:
try:
by_bam[x["align_bam"]].append(x)
except KeyError:
by_bam[x["align_bam"]] = [x]
by_batch = collections.OrderedDict()
lead_batches = {}
for grouped_calls in by_bam.values():
def orig_svcaller_order(x):
return _get_svcallers(x).index(x["config"]["algorithm"]["svcaller_active"])
sorted_svcalls = sorted([x for x in grouped_calls if "sv" in x],
key=orig_svcaller_order)
final = grouped_calls[0]
if len(sorted_svcalls) > 0:
final_calls = reduce(operator.add, [x["sv"] for x in sorted_svcalls])
if not initial_only:
for caller in (c for c in _get_svcallers(final) if c in _ENSEMBLE_CALLERS):
final_calls = _ENSEMBLE_CALLERS[caller](final_calls, final)
final_calls = ensemble.summarize(final_calls, final, grouped_calls)
final_calls = validate.evaluate(final, final_calls)
final["sv"] = final_calls
del final["config"]["algorithm"]["svcaller_active"]
batch = dd.get_batch(final) or dd.get_sample_name(final)
batches = batch if isinstance(batch, (list, tuple)) else [batch]
lead_batches[dd.get_sample_name(final)] = batches[0]
for batch in batches:
try:
by_batch[batch].append(final)
except KeyError:
by_batch[batch] = [final]
out = []
for batch, items in by_batch.items():
if any("svplots" in dd.get_tools_on(d) for d in items):
plot_items = plot.by_regions(items)
else:
plot_items = items
for data in plot_items:
if lead_batches[dd.get_sample_name(data)] == batch:
out.append([data])
return out
def finalize_sv(samples, config):
"""Combine results from multiple sv callers into a single ordered 'sv' key.
Handles ensemble calling and plotting of results.
"""
by_bam = collections.OrderedDict()
for x in samples:
try:
by_bam[x["align_bam"]].append(x)
except KeyError:
by_bam[x["align_bam"]] = [x]
highdepths = filter(
lambda x: x is not None,
list(set([tz.get_in(["config", "algorithm", "highdepth_regions"], x) for x in samples])),
)
by_batch = collections.OrderedDict()
lead_batches = {}
for grouped_calls in by_bam.values():
def orig_svcaller_order(x):
return _get_svcallers(x).index(x["config"]["algorithm"]["svcaller_active"])
sorted_svcalls = sorted([x for x in grouped_calls if "sv" in x], key=orig_svcaller_order)
final = grouped_calls[0]
if len(sorted_svcalls) > 0:
final_calls = reduce(operator.add, [x["sv"] for x in sorted_svcalls])
final_calls = ensemble.summarize(final_calls, final, highdepths)
final_calls = validate.evaluate(final, final_calls)
final["sv"] = final_calls
del final["config"]["algorithm"]["svcaller_active"]
batch = dd.get_batch(final) or dd.get_sample_name(final)
batches = batch if isinstance(batch, (list, tuple)) else [batch]
lead_batches[dd.get_sample_name(final)] = batches[0]
for batch in batches:
try:
by_batch[batch].append(final)
except KeyError:
by_batch[batch] = [final]
out = []
for batch, items in by_batch.items():
plot_items = plot.by_regions(items)
for data in plot_items:
if lead_batches[dd.get_sample_name(data)] == batch:
out.append([data])
return out
def detect_sv(items, all_items=None, stage="standard"):
"""Top level parallel target for examining structural variation.
"""
items = [utils.to_single_data(x) for x in items]
svcaller = items[0]["config"]["algorithm"].get("svcaller")
caller_fn = _get_callers(items, stage, special_cases=True).get(svcaller)
out = []
if svcaller and caller_fn:
if (all_items and svcaller in _NEEDS_BACKGROUND
and not vcfutils.is_paired_analysis(
[x.get("align_bam") for x in items], items)):
names = set([dd.get_sample_name(x) for x in items])
background = [
x for x in all_items if dd.get_sample_name(x) not in names
]
for svdata in caller_fn(items, background):
out.append([svdata])
else:
for svdata in caller_fn(items):
out.append([svdata])
else:
for data in items:
out.append([data])
# Avoid nesting of callers for CWL runs for easier extraction
if "cwl_keys" in items[0]:
out_cwl = []
for data in [utils.to_single_data(x) for x in out]:
# Run validation directly from CWL runs since we're single stage
data = validate.evaluate(data)
data["svvalidate"] = {
"summary": tz.get_in(["sv-validate", "csv"], data)
}
svs = data.get("sv")
if svs:
assert len(svs) == 1, svs
data["sv"] = svs[0]
out_cwl.append([data])
return out_cwl
return out
def detect_sv(items, all_items=None, stage="standard"):
"""Top level parallel target for examining structural variation.
"""
items = [utils.to_single_data(x) for x in items]
items = cwlutils.unpack_tarballs(items, items[0])
svcaller = items[0]["config"]["algorithm"].get("svcaller")
caller_fn = _get_callers(items, stage, special_cases=True).get(svcaller)
out = []
if svcaller and caller_fn:
if (all_items and svcaller in _NEEDS_BACKGROUND and
not vcfutils.is_paired_analysis([x.get("align_bam") for x in items], items)):
names = set([dd.get_sample_name(x) for x in items])
background = [x for x in all_items if dd.get_sample_name(x) not in names]
for svdata in caller_fn(items, background):
out.append([svdata])
else:
for svdata in caller_fn(items):
out.append([svdata])
else:
for data in items:
out.append([data])
# Avoid nesting of callers for CWL runs for easier extraction
if cwlutils.is_cwl_run(items[0]):
out_cwl = []
for data in [utils.to_single_data(x) for x in out]:
# Run validation directly from CWL runs since we're single stage
data = validate.evaluate(data)
data["svvalidate"] = {"summary": tz.get_in(["sv-validate", "csv"], data)}
svs = data.get("sv")
if svs:
assert len(svs) == 1, svs
data["sv"] = svs[0]
else:
data["sv"] = {}
data = _add_supplemental(data)
out_cwl.append([data])
return out_cwl
return out
def _combine_multiple_svcallers(samples):
"""
"""
by_bam = collections.OrderedDict()
for x in samples:
try:
by_bam[x[0]["align_bam"]].append(x[0])
except KeyError:
by_bam[x[0]["align_bam"]] = [x[0]]
out = []
for grouped_calls in by_bam.values():
def orig_svcaller_order(x):
return _get_svcallers(x).index(x["config"]["algorithm"]["svcaller_active"])
sorted_svcalls = sorted([x for x in grouped_calls if "sv" in x],
key=orig_svcaller_order)
final_calls = reduce(operator.add, [x["sv"] for x in sorted_svcalls])
final = grouped_calls[0]
final_calls = ensemble.summarize(final_calls, final)
final_calls = validate.evaluate(final, final_calls)
final["sv"] = final_calls
del final["config"]["algorithm"]["svcaller_active"]
out.append([final])
return out
def validate_sv(data):
"""Validate structural variant calls for a sample.
"""
return [[validate.evaluate(data)]]
def validate_sv(data):
"""Validate structural variant calls for a sample.
"""
return [[validate.evaluate(data)]]
