def combine_multiple_callers(samples):
"""Collapse together variant calls from multiple approaches into single data item with `variants`.
"""
by_bam = collections.OrderedDict()
for data in (x[0] for x in samples):
work_bam = tz.get_in(("combine", "work_bam", "out"), data, data.get("align_bam"))
jointcaller = tz.get_in(("config", "algorithm", "jointcaller"), data)
variantcaller = get_variantcaller(data)
key = (multi.get_batch_for_key(data), work_bam)
if key not in by_bam:
by_bam[key] = []
by_bam[key].append((variantcaller, jointcaller, data))
out = []
for callgroup in by_bam.values():
ready_calls = []
for variantcaller, jointcaller, data in callgroup:
if variantcaller:
cur = data.get("vrn_file_plus", {})
cur.update({"variantcaller": variantcaller,
"vrn_file": data.get("vrn_file_orig") if jointcaller else data.get("vrn_file"),
"vrn_file_batch": data.get("vrn_file_batch") if not jointcaller else None,
"vrn_stats": data.get("vrn_stats"),
"validate": data.get("validate") if not jointcaller else None})
if jointcaller:
cur["population"] = False
ready_calls.append(cur)
if jointcaller:
cur = {"variantcaller": jointcaller,
"vrn_file": data.get("vrn_file"),
"vrn_file_batch": data.get("vrn_file_batch"),
"validate": data.get("validate"),
"do_upload": False}
if not variantcaller:
cur["population"] = {"vcf": data.get("vrn_file")}
ready_calls.append(cur)
if not jointcaller and not variantcaller:
ready_calls.append({"variantcaller": "precalled",
"vrn_file": data.get("vrn_file"),
"validate": data.get("validate"),
"do_upload": False})
final = callgroup[0][-1]
def orig_variantcaller_order(x):
try:
return final["config"]["algorithm"]["orig_variantcaller"].index(x["variantcaller"])
except ValueError:
return final["config"]["algorithm"]["orig_jointcaller"].index(x["variantcaller"])
if len(ready_calls) > 1 and "orig_variantcaller" in final["config"]["algorithm"]:
final["variants"] = sorted(ready_calls, key=orig_variantcaller_order)
final["config"]["algorithm"]["variantcaller"] = final["config"]["algorithm"].pop("orig_variantcaller")
if "orig_jointcaller" in final["config"]["algorithm"]:
final["config"]["algorithm"]["jointcaller"] = final["config"]["algorithm"].pop("orig_jointcaller")
else:
final["variants"] = ready_calls
final.pop("vrn_file_batch", None)
final.pop("vrn_file_orig", None)
final.pop("vrn_file_plus", None)
final.pop("vrn_stats", None)
out.append([final])
return out
def combine_multiple_callers(samples):
"""Collapse together variant calls from multiple approaches into single data item with `variants`.
"""
by_bam = collections.OrderedDict()
for data in (x[0] for x in samples):
work_bam = tz.get_in(("combine", "work_bam", "out"), data, data.get("align_bam"))
jointcaller = tz.get_in(("config", "algorithm", "jointcaller"), data)
variantcaller = get_variantcaller(data)
key = (multi.get_batch_for_key(data), work_bam)
if key not in by_bam:
by_bam[key] = []
by_bam[key].append((variantcaller, jointcaller, data))
out = []
for callgroup in by_bam.values():
ready_calls = []
for variantcaller, jointcaller, data in callgroup:
if variantcaller:
cur = data.get("vrn_file_plus", {})
cur.update({"variantcaller": variantcaller,
"vrn_file": data.get("vrn_file_orig") if jointcaller else data.get("vrn_file"),
"vrn_file_batch": data.get("vrn_file_batch") if not jointcaller else None,
"vrn_stats": data.get("vrn_stats"),
"validate": data.get("validate") if not jointcaller else None})
if jointcaller:
cur["population"] = False
ready_calls.append(cur)
if jointcaller:
ready_calls.append({"variantcaller": jointcaller,
"vrn_file": data.get("vrn_file"),
"vrn_file_batch": data.get("vrn_file_batch"),
"validate": data.get("validate"),
"do_upload": False})
if not jointcaller and not variantcaller:
ready_calls.append({"variantcaller": "precalled",
"vrn_file": data.get("vrn_file"),
"validate": data.get("validate"),
"do_upload": False})
final = callgroup[0][-1]
def orig_variantcaller_order(x):
try:
return final["config"]["algorithm"]["orig_variantcaller"].index(x["variantcaller"])
except ValueError:
return final["config"]["algorithm"]["orig_jointcaller"].index(x["variantcaller"])
if len(ready_calls) > 1 and "orig_variantcaller" in final["config"]["algorithm"]:
final["variants"] = sorted(ready_calls, key=orig_variantcaller_order)
final["config"]["algorithm"]["variantcaller"] = final["config"]["algorithm"].pop("orig_variantcaller")
if "orig_jointcaller" in final["config"]["algorithm"]:
final["config"]["algorithm"]["jointcaller"] = final["config"]["algorithm"].pop("orig_jointcaller")
else:
final["variants"] = ready_calls
final.pop("vrn_file_batch", None)
final.pop("vrn_file_orig", None)
final.pop("vrn_file_plus", None)
final.pop("vrn_stats", None)
out.append([final])
return out
def _group_by_batches(samples, check_fn):
"""Group calls by batches, processing families together during ensemble calling.
"""
batch_groups = collections.defaultdict(list)
extras = []
for data in [x[0] for x in samples]:
if check_fn(data):
batch_groups[multi.get_batch_for_key(data)].append(data)
else:
extras.append([data])
return batch_groups, extras
def _group_by_batches(samples, check_fn):
"""Group calls by batches, processing families together during ensemble calling.
"""
batch_groups = collections.defaultdict(list)
extras = []
for data in [x[0] for x in samples]:
if check_fn(data):
batch_groups[multi.get_batch_for_key(data)].append(data)
else:
extras.append([data])
return batch_groups, extras
def _collapse_by_bam_variantcaller(samples):
"""Collapse regions to a single representative by BAM input, variant caller and batch.
"""
by_bam = collections.OrderedDict()
for data in (x[0] for x in samples):
work_bam = utils.get_in(data, ("combine", "work_bam", "out"), data.get("align_bam"))
variantcaller = get_variantcaller(data)
if isinstance(work_bam, list):
work_bam = tuple(work_bam)
key = (multi.get_batch_for_key(data), work_bam, variantcaller)
try:
by_bam[key].append(data)
except KeyError:
by_bam[key] = [data]
out = []
for grouped_data in by_bam.values():
cur = grouped_data[0]
cur.pop("region", None)
region_bams = cur.pop("region_bams", None)
if region_bams and len(region_bams[0]) > 1:
cur.pop("work_bam", None)
out.append([cur])
return out
def _collapse_by_bam_variantcaller(samples):
"""Collapse regions to a single representative by BAM input, variant caller and batch.
"""
by_bam = collections.OrderedDict()
for data in (x[0] for x in samples):
work_bam = utils.get_in(data, ("combine", "work_bam", "out"), data.get("align_bam"))
variantcaller = get_variantcaller(data)
if isinstance(work_bam, list):
work_bam = tuple(work_bam)
key = (multi.get_batch_for_key(data), work_bam, variantcaller)
try:
by_bam[key].append(data)
except KeyError:
by_bam[key] = [data]
out = []
for grouped_data in by_bam.values():
cur = grouped_data[0]
cur.pop("region", None)
region_bams = cur.pop("region_bams", None)
if region_bams and len(region_bams[0]) > 1:
cur.pop("work_bam", None)
out.append([cur])
return out
