def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
cur_name = "%s, %s" % (data["name"][-1], get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
effect_todo = effects.get_type(data)
if effect_todo:
if effect_todo == "snpeff":
ann_vrn_file = effects.snpeff_effects(data)
elif effect_todo == "vep":
ann_vrn_file = effects.run_vep(data)
else:
raise ValueError(
"Unexpected variant effects configuration: %s" %
effect_todo)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
logger.info("Filtering for %s" % cur_name)
data["vrn_file"] = variant_filtration(
data["vrn_file"], data["sam_ref"],
tz.get_in(("genome_resources", "variation"), data, {}), data)
logger.info("Prioritization for %s" % cur_name)
data["vrn_file"] = prioritize.handle_vcf_calls(data["vrn_file"], data)
return [[data]]
def finalize_genotyper(call_file, ref_file, config):
"""Perform SNP genotyping and analysis using GATK.
"""
vrn_files = configured_vrn_files(config, ref_file)
filter_snp = variant_filtration(call_file, ref_file, vrn_files, config)
_eval_genotyper(filter_snp, ref_file, vrn_files.dbsnp, config)
return filter_snp
def postprocess_variants(items):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
vrn_key = "vrn_file"
if not isinstance(items, dict):
items = [utils.to_single_data(x) for x in items]
if "vrn_file_joint" in items[0]:
vrn_key = "vrn_file_joint"
data, items = _get_batch_representative(items, vrn_key)
items = cwlutils.unpack_tarballs(items, data)
data = cwlutils.unpack_tarballs(data, data)
cur_name = "%s, %s" % (dd.get_sample_name(data), get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
orig_vrn_file = data.get(vrn_key)
data = _symlink_to_workdir(data, [vrn_key])
data = _symlink_to_workdir(data,
["config", "algorithm", "variant_regions"])
if data.get(vrn_key):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file, vrn_stats = effects.add_to_vcf(data[vrn_key], data)
if ann_vrn_file:
data[vrn_key] = ann_vrn_file
if vrn_stats:
data["vrn_stats"] = vrn_stats
orig_items = _get_orig_items(items)
logger.info("Annotate VCF file: %s" % cur_name)
data[vrn_key] = annotation.finalize_vcf(data[vrn_key],
get_variantcaller(data),
orig_items)
if dd.get_analysis(data).lower().find("rna-seq") >= 0:
logger.info("Annotate RNA editing sites")
ann_file = vcfanno.run_vcfanno(dd.get_vrn_file(data), ["rnaedit"],
data)
if ann_file:
data[vrn_key] = ann_file
if cwlutils.is_cwl_run(data):
logger.info("Annotate with population level variation data")
ann_file = population.run_vcfanno(dd.get_vrn_file(data), data,
population.do_db_build([data]))
if ann_file:
data[vrn_key] = ann_file
logger.info("Filtering for %s" % cur_name)
data[vrn_key] = variant_filtration(
data[vrn_key], dd.get_ref_file(data),
tz.get_in(("genome_resources", "variation"), data, {}), data,
orig_items)
logger.info("Prioritization for %s" % cur_name)
prio_vrn_file = prioritize.handle_vcf_calls(data[vrn_key], data,
orig_items)
if prio_vrn_file != data[vrn_key]:
data[vrn_key] = prio_vrn_file
logger.info("Germline extraction for %s" % cur_name)
data = germline.extract(data, orig_items)
if dd.get_align_bam(data):
data = damage.run_filter(data[vrn_key], dd.get_align_bam(data),
dd.get_ref_file(data), data, orig_items)
if orig_vrn_file and os.path.samefile(data[vrn_key], orig_vrn_file):
data[vrn_key] = orig_vrn_file
return [[data]]
def postprocess_variants(items):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
data = _get_batch_representative(items, "vrn_file")
cur_name = "%s, %s" % (dd.get_sample_name(data), get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
orig_vrn_file = data.get("vrn_file")
data = _symlink_to_workdir(data, ["vrn_file"])
data = _symlink_to_workdir(data,
["config", "algorithm", "variant_regions"])
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file, vrn_stats = effects.add_to_vcf(data["vrn_file"], data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
if vrn_stats:
data["vrn_stats"] = vrn_stats
orig_items = _get_orig_items(items)
logger.info("Annotate VCF file: %s" % cur_name)
data["vrn_file"] = annotation.finalize_vcf(data["vrn_file"],
get_variantcaller(data),
orig_items)
logger.info("Filtering for %s" % cur_name)
data["vrn_file"] = variant_filtration(
data["vrn_file"], dd.get_ref_file(data),
tz.get_in(("genome_resources", "variation"), data, {}), data,
orig_items)
logger.info("Prioritization for %s" % cur_name)
data["vrn_file"] = prioritize.handle_vcf_calls(data["vrn_file"], data,
orig_items)
logger.info("Germline extraction for %s" % cur_name)
data = germline.extract(data, orig_items)
if orig_vrn_file and os.path.samefile(data["vrn_file"], orig_vrn_file):
data["vrn_file"] = orig_vrn_file
return [[data]]
def postprocess_variants(items):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
data = _get_batch_representative(items, "vrn_file")
cur_name = "%s, %s" % (dd.get_sample_name(data), get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
orig_vrn_file = data.get("vrn_file")
data = _symlink_to_workdir(data, ["vrn_file"])
data = _symlink_to_workdir(data, ["config", "algorithm", "variant_regions"])
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file, vrn_stats = effects.add_to_vcf(data["vrn_file"], data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
if vrn_stats:
data["vrn_stats"] = vrn_stats
logger.info("Filtering for %s" % cur_name)
orig_items = _get_orig_items(items)
data["vrn_file"] = variant_filtration(data["vrn_file"], dd.get_ref_file(data),
tz.get_in(("genome_resources", "variation"), data, {}),
data, orig_items)
logger.info("Prioritization for %s" % cur_name)
data["vrn_file"] = prioritize.handle_vcf_calls(data["vrn_file"], data, orig_items)
logger.info("Germline extraction for %s" % cur_name)
data = germline.extract(data, orig_items)
if orig_vrn_file and os.path.samefile(data["vrn_file"], orig_vrn_file):
data["vrn_file"] = orig_vrn_file
return [[data]]
def finalize_genotyper(call_file, bam_file, ref_file, config):
"""Perform SNP genotyping and analysis.
"""
vrn_files = configured_vrn_files(config, ref_file)
variantcaller = config["algorithm"].get("variantcaller", "gatk")
if variantcaller in ["freebayes", "cortex", "samtools", "gatk-haplotype", "varscan"]:
call_file = annotation.annotate_nongatk_vcf(call_file, bam_file, vrn_files.dbsnp,
ref_file, config)
filter_snp = variant_filtration(call_file, ref_file, vrn_files, config)
return filter_snp
def finalize_genotyper(call_file, bam_file, ref_file, config):
"""Perform SNP genotyping and analysis.
"""
vrn_files = configured_vrn_files(config, ref_file)
variantcaller = config["algorithm"].get("variantcaller", "gatk")
if variantcaller in ["freebayes", "cortex"]:
call_file = freebayes.postcall_annotate(call_file, ref_file, vrn_files, config)
filter_snp = variant_filtration(call_file, ref_file, vrn_files, config)
phase_snp = phasing.read_backed_phasing(filter_snp, bam_file, ref_file, config)
_eval_genotyper(phase_snp, ref_file, vrn_files.dbsnp, config)
return phase_snp
def finalize_genotyper(call_file, bam_file, ref_file, config):
"""Perform SNP genotyping and analysis.
"""
vrn_files = configured_vrn_files(config, ref_file)
variantcaller = config["algorithm"].get("variantcaller", "gatk")
if variantcaller in ["freebayes", "cortex", "samtools"]:
call_file = freebayes.postcall_annotate(call_file, ref_file, vrn_files, config)
filter_snp = variant_filtration(call_file, ref_file, vrn_files, config)
phase_snp = phasing.read_backed_phasing(filter_snp, bam_file, ref_file, config)
_eval_genotyper(phase_snp, ref_file, vrn_files.dbsnp, config)
return phase_snp
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
logger.info("Finalizing variant calls: %s" % str(data["name"]))
if data["work_bam"] and data.get("vrn_file"):
data["vrn_file"] = variant_filtration(
data["vrn_file"], data["sam_ref"],
data["genome_resources"]["variation"], data["config"])
logger.info("Calculating variation effects for %s" % str(data["name"]))
ann_vrn_file = effects.snpeff_effects(data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
return [[data]]
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
logger.info("Finalizing variant calls: %s" % str(data["name"]))
if data["work_bam"] and data.get("vrn_file"):
vrn_files = configured_vrn_files(data["config"], data["sam_ref"])
data["vrn_file"] = variant_filtration(data["vrn_file"], data["sam_ref"], vrn_files,
data["config"])
logger.info("Calculating variation effects for %s" % str(data["name"]))
ann_vrn_file = effects.snpeff_effects(data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
return [[data]]
def finalize_genotyper(call_file, bam_file, ref_file, config):
"""Perform SNP genotyping and analysis.
"""
vrn_files = configured_vrn_files(config, ref_file)
variantcaller = config["algorithm"].get("variantcaller", "gatk")
if variantcaller in [
"freebayes", "cortex", "samtools", "gatk-haplotype", "varscan"
]:
call_file = annotation.annotate_nongatk_vcf(call_file, bam_file,
vrn_files.dbsnp, ref_file,
config)
filter_snp = variant_filtration(call_file, ref_file, vrn_files, config)
return filter_snp
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
cur_name = "%s, %s" % (data["name"][-1], get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
if data["work_bam"] and data.get("vrn_file"):
data["vrn_file"] = variant_filtration(data["vrn_file"], data["sam_ref"],
data["genome_resources"]["variation"],
data["config"])
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file = effects.snpeff_effects(data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
return [[data]]
def postprocess_variants(items):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
vrn_key = "vrn_file"
if not isinstance(items, dict):
items = [utils.to_single_data(x) for x in items]
if "vrn_file_joint" in items[0]:
vrn_key = "vrn_file_joint"
data, items = _get_batch_representative(items, vrn_key)
items = cwlutils.unpack_tarballs(items, data)
data = cwlutils.unpack_tarballs(data, data)
cur_name = "%s, %s" % (dd.get_sample_name(data), get_variantcaller(data, require_bam=False))
logger.info("Finalizing variant calls: %s" % cur_name)
orig_vrn_file = data.get(vrn_key)
data = _symlink_to_workdir(data, [vrn_key])
data = _symlink_to_workdir(data, ["config", "algorithm", "variant_regions"])
if data.get(vrn_key):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file, vrn_stats = effects.add_to_vcf(data[vrn_key], data)
if ann_vrn_file:
data[vrn_key] = ann_vrn_file
if vrn_stats:
data["vrn_stats"] = vrn_stats
orig_items = _get_orig_items(items)
logger.info("Annotate VCF file: %s" % cur_name)
data[vrn_key] = annotation.finalize_vcf(data[vrn_key], get_variantcaller(data, require_bam=False),
orig_items)
if cwlutils.is_cwl_run(data):
logger.info("Annotate with population level variation data")
ann_file = population.run_vcfanno(data[vrn_key], data)
if ann_file:
data[vrn_key] = ann_file
logger.info("Filtering for %s" % cur_name)
data[vrn_key] = variant_filtration(data[vrn_key], dd.get_ref_file(data),
tz.get_in(("genome_resources", "variation"), data, {}),
data, orig_items)
logger.info("Prioritization for %s" % cur_name)
prio_vrn_file = prioritize.handle_vcf_calls(data[vrn_key], data, orig_items)
if prio_vrn_file != data[vrn_key]:
data[vrn_key] = prio_vrn_file
logger.info("Germline extraction for %s" % cur_name)
data = germline.extract(data, orig_items)
if dd.get_align_bam(data):
data = damage.run_filter(data[vrn_key], dd.get_align_bam(data), dd.get_ref_file(data),
data, orig_items)
if orig_vrn_file and os.path.samefile(data[vrn_key], orig_vrn_file):
data[vrn_key] = orig_vrn_file
return [[data]]
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
cur_name = "%s, %s" % (data["name"][-1], get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file = effects.snpeff_effects(data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
logger.info("Filtering for %s" % cur_name)
data["vrn_file"] = variant_filtration(
data["vrn_file"], data["sam_ref"],
data["genome_resources"]["variation"], data)
return [[data]]
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
cur_name = "%s, %s" % (data["name"][-1], get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file = effects.add_to_vcf(data["vrn_file"], data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
logger.info("Filtering for %s" % cur_name)
data["vrn_file"] = variant_filtration(data["vrn_file"], data["sam_ref"],
tz.get_in(("genome_resources", "variation"), data, {}),
data)
logger.info("Prioritization for %s" % cur_name)
data["vrn_file"] = prioritize.handle_vcf_calls(data["vrn_file"], data)
return [[data]]
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
cur_name = "%s, %s" % (data["name"][-1], get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
ann_vrn_file, vrn_stats = effects.add_to_vcf(data["vrn_file"], data)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
if vrn_stats:
data["vrn_stats"] = vrn_stats
logger.info("Filtering for %s" % cur_name)
data["vrn_file"] = variant_filtration(
data["vrn_file"], data["sam_ref"],
tz.get_in(("genome_resources", "variation"), data, {}), data)
logger.info("Prioritization for %s" % cur_name)
data["vrn_file"] = prioritize.handle_vcf_calls(data["vrn_file"], data)
return [[data]]
def postprocess_variants(data):
"""Provide post-processing of variant calls: filtering and effects annotation.
"""
cur_name = "%s, %s" % (data["name"][-1], get_variantcaller(data))
logger.info("Finalizing variant calls: %s" % cur_name)
if data.get("align_bam") and data.get("vrn_file"):
logger.info("Calculating variation effects for %s" % cur_name)
effect_todo = tz.get_in(("config", "algorithm", "effects"), data, "snpeff")
if effect_todo:
if effect_todo == "snpeff":
ann_vrn_file = effects.snpeff_effects(data)
elif effect_todo == "vep":
ann_vrn_file = effects.run_vep(data)
else:
raise ValueError("Unexpected variant effects configuration: %s" % effect_todo)
if ann_vrn_file:
data["vrn_file"] = ann_vrn_file
logger.info("Filtering for %s" % cur_name)
data["vrn_file"] = variant_filtration(data["vrn_file"], data["sam_ref"],
tz.get_in(("genome_resources", "variation"), data, {}),
data)
return [[data]]
