def get_quality_filtered_genotypes(variant, quality_filter):
"""
Returns:
A list of (indiv_id, genotype) tuples for genotypes that pass quality filter, or an empty list if none do
"""
ret = []
num_het = 0
num_hom_alt = 0
num_genotypes = variant.num_genotypes()
for indiv_id, genotype in variant.get_genotypes():
if genotype.num_alt > 0:
if passes_genotype_filter(genotype, quality_filter):
if genotype.num_alt == 1:
num_het += 1
if genotype.num_alt == 2:
num_hom_alt += 1
ret.append((indiv_id, genotype))
if 'het_ratio' in quality_filter:
if float(num_het)*100 / num_genotypes > quality_filter['het_ratio']:
return []
if 'hom_alt_ratio' in quality_filter:
if float(num_hom_alt)*100 / num_genotypes > quality_filter['hom_alt_ratio']:
return []
return ret
def get_quality_filtered_genotypes(variant, quality_filter):
"""
Returns:
A list of (indiv_id, genotype) tuples for genotypes that pass quality filter, or an empty list if none do
"""
ret = []
num_het = 0
num_hom_alt = 0
num_genotypes = variant.num_genotypes()
for indiv_id, genotype in variant.get_genotypes():
if genotype.num_alt > 0:
if passes_genotype_filter(genotype, quality_filter):
if genotype.num_alt == 1:
num_het += 1
if genotype.num_alt == 2:
num_hom_alt += 1
ret.append((indiv_id, genotype))
if 'het_ratio' in quality_filter:
if float(num_het)*100 / num_genotypes > quality_filter['het_ratio']:
return []
if 'hom_alt_ratio' in quality_filter:
if float(num_hom_alt)*100 / num_genotypes > quality_filter['hom_alt_ratio']:
return []
return ret
def passes_quality_filter(variant, quality_filter, indivs_to_consider):
"""
Does variant pass given the items in quality_filter?
Return True or False
TODO: this is weird
"""
for indiv_id in indivs_to_consider:
genotype = variant.get_genotype(indiv_id)
if not passes_genotype_filter(genotype, quality_filter):
return False
return True
def passes_quality_filter(variant, quality_filter, indivs_to_consider):
"""
Does variant pass given the items in quality_filter?
Return True or False
TODO: this is weird
"""
for indiv_id in indivs_to_consider:
genotype = variant.get_genotype(indiv_id)
if not passes_genotype_filter(genotype, quality_filter):
return False
return True
def get_de_novo_variants(datastore,
reference,
family,
variant_filter=None,
quality_filter=None):
"""
Returns variants that follow homozygous recessive inheritance in family
"""
de_novo_filter = inheritance.get_de_novo_filter(family)
db_query = datastore._make_db_query(de_novo_filter, variant_filter)
collection = datastore._get_family_collection(family.project_id,
family.family_id)
if not collection:
raise ValueError(
"Error: mongodb collection not found for project %s family %s " %
(family.project_id, family.family_id))
MONGO_QUERY_RESULTS_LIMIT = 5000
variant_iter = collection.find(db_query).sort('xpos').limit(
MONGO_QUERY_RESULTS_LIMIT + 5)
# get ids of parents in this family
valid_ids = set(indiv_id for indiv_id in family.individuals)
paternal_ids = set(i.paternal_id for i in family.get_individuals()
if i.paternal_id in valid_ids)
maternal_ids = set(i.maternal_id for i in family.get_individuals()
if i.maternal_id in valid_ids)
parental_ids = paternal_ids | maternal_ids
# loop over all variants returned
for i, variant_dict in enumerate(variant_iter):
if i > MONGO_QUERY_RESULTS_LIMIT:
raise Exception(
"MONGO_QUERY_RESULTS_LIMIT of %s exceeded for query: %s" %
(MONGO_QUERY_RESULTS_LIMIT, db_query))
variant = Variant.fromJSON(variant_dict)
datastore.add_annotations_to_variant(variant, family.project_id)
if not passes_variant_filter(variant, variant_filter)[0]:
continue
# handle genotype filters
if len(parental_ids) != 2:
# ordinary filters for non-trios
for indiv_id in de_novo_filter.keys():
genotype = variant.get_genotype(indiv_id)
if not passes_genotype_filter(genotype, quality_filter):
break
else:
yield variant
else:
# for trios use Mark's recommended filters for de-novo variants:
# Hard-coded thresholds:
# 1) Child must have > 10% of combined Parental Read Depth
# 2) MinimumChildGQscore >= 20
# 3) MaximumParentAlleleBalance <= 5%
# Adjustable filters:
# Variants should PASS
# Child AB should be >= 20
# compute parental read depth for filter 1
total_parental_read_depth = 0
for indiv_id in parental_ids:
genotype = variant.get_genotype(indiv_id)
if genotype.extras and 'dp' in genotype.extras and genotype.extras[
'dp'] != '.':
total_parental_read_depth += int(genotype.extras['dp'])
else:
total_parental_read_depth = None # both parents must have DP to use the parental_read_depth filters
break
for indiv_id in de_novo_filter.keys():
quality_filter_temp = quality_filter.copy(
) # copy before modifying
if indiv_id in parental_ids:
# handle one of the parents
quality_filter_temp['max_ab'] = 5
else:
# handle child
quality_filter_temp['min_gq'] = 20
if total_parental_read_depth is not None:
quality_filter_temp[
'min_dp'] = total_parental_read_depth * 0.1
genotype = variant.get_genotype(indiv_id)
if not passes_genotype_filter(genotype, quality_filter_temp):
#print("%s: %s " % (variant.chr, variant.pos))
break
else:
yield variant
def get_de_novo_variants(datastore, reference, family, variant_filter=None, quality_filter=None):
"""
Returns variants that follow homozygous recessive inheritance in family
"""
de_novo_filter = inheritance.get_de_novo_filter(family)
db_query = datastore._make_db_query(de_novo_filter, variant_filter)
collection = datastore._get_family_collection(family.project_id, family.family_id)
if not collection:
raise ValueError("Error: mongodb collection not found for project %s family %s " % (family.project_id, family.family_id))
variant_iter = collection.find(db_query).sort('xpos')
# get ids of parents in this family
valid_ids = set(indiv_id for indiv_id in family.individuals)
paternal_ids = set(i.paternal_id for i in family.get_individuals() if i.paternal_id in valid_ids)
maternal_ids = set(i.maternal_id for i in family.get_individuals() if i.maternal_id in valid_ids)
parental_ids = paternal_ids | maternal_ids
# loop over all variants returned
for variant_dict in variant_iter:
variant = Variant.fromJSON(variant_dict)
datastore.add_annotations_to_variant(variant, family.project_id)
if not passes_variant_filter(variant, variant_filter)[0]:
continue
# handle genotype filters
if len(parental_ids) != 2:
# ordinary filters for non-trios
for indiv_id in de_novo_filter.keys():
genotype = variant.get_genotype(indiv_id)
if not passes_genotype_filter(genotype, quality_filter):
break
else:
yield variant
else:
# for trios use Mark's recommended filters for de-novo variants:
# Hard-coded thresholds:
# 1) Child must have > 10% of combined Parental Read Depth
# 2) MinimumChildGQscore >= 20
# 3) MaximumParentAlleleBalance <= 5%
# Adjustable filters:
# Variants should PASS
# Child AB should be >= 20
# compute parental read depth for filter 1
total_parental_read_depth = 0
for indiv_id in parental_ids:
genotype = variant.get_genotype(indiv_id)
if genotype.extras and 'dp' in genotype.extras:
total_parental_read_depth += int(genotype.extras['dp'])
else:
total_parental_read_depth = None # both parents must have DP to use the parental_read_depth filters
break
for indiv_id in de_novo_filter.keys():
quality_filter_temp = quality_filter.copy() # copy before modifying
if indiv_id in parental_ids:
# handle one of the parents
quality_filter_temp['max_ab'] = 5
else:
# handle child
quality_filter_temp['min_gq'] = 20
if total_parental_read_depth is not None:
quality_filter_temp['min_dp'] = total_parental_read_depth * 0.1
genotype = variant.get_genotype(indiv_id)
if not passes_genotype_filter(genotype, quality_filter_temp):
#print("%s: %s " % (variant.chr, variant.pos))
break
else:
yield variant
