def test_is_pathogenic_VEP97_conflicting(one_vep97_annotated_variant):
## WHEN checking if variants should be loaded
pathogenic = is_pathogenic(one_vep97_annotated_variant)
## THEN assert that the variant should be loaded
assert pathogenic is True
def test_is_pathogenic_no_annotation(cyvcf2_variant):
## GIVEN a variant without clinvar annotations
## WHEN checking if variants should be loaded
pathogenic = is_pathogenic(cyvcf2_variant)
## THEN assert that The variant should be loaded
assert pathogenic is False
def test_is_pathogenic_classic_pathogenic(cyvcf2_variant):
## GIVEN a variant with classic clinvar annotations
acc_nr = "RCV000014440.17|RCV000014441.25|RCV000014442.25|RCV000014443.17|RCV000184011.1|RCV000188658.1"
clnsig = "5|4|3|2|1|0"
revstat = "conf|single|single|single|conf|conf"
cyvcf2_variant.INFO["CLNVID"] = acc_nr
cyvcf2_variant.INFO["CLNSIG"] = clnsig
cyvcf2_variant.INFO["CLNREVSTAT"] = revstat
## WHEN checking if variants should be loaded
pathogenic = is_pathogenic(cyvcf2_variant)
## THEN assert that The variant should be loaded
assert pathogenic is True
def test_is_pathogenic_benign(cyvcf2_variant):
## GIVEN a variant with classic clinvar annotations
acc_nr = "265359"
clnsig = "Likely_benign"
revstat = "criteria_provided,_multiple_submitters,_no_conflicts"
cyvcf2_variant.INFO["CLNVID"] = acc_nr
cyvcf2_variant.INFO["CLNSIG"] = clnsig
cyvcf2_variant.INFO["CLNREVSTAT"] = revstat
## WHEN checking if variants should be loaded
pathogenic = is_pathogenic(cyvcf2_variant)
## THEN assert that The variant should be loaded
assert pathogenic is False
def _load_variants(
self,
variants,
variant_type,
case_obj,
individual_positions,
rank_threshold,
institute_id,
build=None,
rank_results_header=None,
vep_header=None,
category="snv",
sample_info=None,
):
"""Perform the loading of variants
This is the function that loops over the variants, parse them and build the variant
objects so they are ready to be inserted into the database.
Args:
variants(iterable(cyvcf2.Variant))
variant_type(str): ['clinical', 'research']
case_obj(dict)
individual_positions(dict): How individuals are positioned in vcf
rank_treshold(int): Only load variants with a rank score > than this
institute_id(str)
build(str): Genome build
rank_results_header(list): Rank score categories
vep_header(list)
category(str): ['snv','sv','cancer','str']
sample_info(dict): A dictionary with info about samples.
Strictly for cancer to tell which is tumor
Returns:
nr_inserted(int)
"""
build = build or "37"
genes = [gene_obj for gene_obj in self.all_genes(build=build)]
gene_to_panels = self.gene_to_panels(case_obj)
hgncid_to_gene = self.hgncid_to_gene(genes=genes, build=build)
genomic_intervals = self.get_coding_intervals(genes=genes)
LOG.info("Start inserting {0} {1} variants into database".format(
variant_type, category))
start_insertion = datetime.now()
start_five_thousand = datetime.now()
# These are the number of parsed varaints
nr_variants = 0
# These are the number of variants that meet the criteria and gets inserted
nr_inserted = 0
# This is to keep track of blocks of inserted variants
inserted = 1
nr_bulks = 0
# We want to load batches of variants to reduce the number of network round trips
bulk = {}
current_region = None
for nr_variants, variant in enumerate(variants):
# All MT variants are loaded
mt_variant = "MT" in variant.CHROM
rank_score = parse_rank_score(variant.INFO.get("RankScore"),
case_obj["_id"])
pathogenic = is_pathogenic(variant)
# Check if the variant should be loaded at all
# if rank score is None means there are no rank scores annotated, all variants will be loaded
# Otherwise we load all variants above a rank score treshold
# Except for MT variants where we load all variants
if ((rank_score is None) or (rank_score > rank_threshold)
or mt_variant or pathogenic):
nr_inserted += 1
# Parse the vcf variant
parsed_variant = parse_variant(
variant=variant,
case=case_obj,
variant_type=variant_type,
rank_results_header=rank_results_header,
vep_header=vep_header,
individual_positions=individual_positions,
category=category,
)
# Build the variant object
variant_obj = build_variant(
variant=parsed_variant,
institute_id=institute_id,
gene_to_panels=gene_to_panels,
hgncid_to_gene=hgncid_to_gene,
sample_info=sample_info,
)
# Check if the variant is in a genomic region
var_chrom = variant_obj["chromosome"]
var_start = variant_obj["position"]
# We need to make sure that the interval has a length > 0
var_end = variant_obj["end"] + 1
var_id = variant_obj["_id"]
# If the bulk should be loaded or not
load = True
new_region = None
intervals = genomic_intervals.get(var_chrom, IntervalTree())
genomic_regions = intervals.overlap(var_start, var_end)
# If the variant is in a coding region
if genomic_regions:
# We know there is data here so get the interval id
new_region = genomic_regions.pop().data
# If the variant is in the same region as previous
# we add it to the same bulk
if new_region == current_region:
load = False
# This is the case where the variant is intergenic
else:
# If the previous variant was also intergenic we add the variant to the bulk
if not current_region:
load = False
# We need to have a max size of the bulk
if len(bulk) > 10000:
load = True
# Load the variant object
if load:
# If the variant bulk contains coding variants we want to update the compounds
if current_region:
self.update_compounds(bulk)
try:
# Load the variants
self.load_variant_bulk(list(bulk.values()))
nr_bulks += 1
except IntegrityError as error:
pass
bulk = {}
current_region = new_region
bulk[var_id] = variant_obj
if nr_variants != 0 and nr_variants % 5000 == 0:
LOG.info("%s variants parsed", str(nr_variants))
LOG.info(
"Time to parse variants: %s",
(datetime.now() - start_five_thousand),
)
start_five_thousand = datetime.now()
if (nr_inserted != 0
and (nr_inserted * inserted) % (1000 * inserted) == 0):
LOG.info("%s variants inserted", nr_inserted)
inserted += 1
# If the variants are in a coding region we update the compounds
if current_region:
self.update_compounds(bulk)
# Load the final variant bulk
self.load_variant_bulk(list(bulk.values()))
nr_bulks += 1
LOG.info("All variants inserted, time to insert variants: {0}".format(
datetime.now() - start_insertion))
if nr_variants:
nr_variants += 1
LOG.info("Nr variants parsed: %s", nr_variants)
LOG.info("Nr variants inserted: %s", nr_inserted)
LOG.debug("Nr bulks inserted: %s", nr_bulks)
return nr_inserted
