def test_parse_small_sv(one_sv_variant, case_obj):
parsed_variant = parse_variant(one_sv_variant, case_obj)
assert parsed_variant["category"] == "sv"
assert parsed_variant["sub_category"] == one_sv_variant.INFO[
"SVTYPE"].lower()
assert parsed_variant["position"] == int(one_sv_variant.POS)
def test_parse_small_sv(one_sv_variant, case_obj):
parsed_variant = parse_variant(one_sv_variant, case_obj)
assert parsed_variant['category'] == 'sv'
assert parsed_variant['sub_category'] == one_sv_variant.INFO[
'SVTYPE'].lower()
assert parsed_variant['position'] == int(one_sv_variant.POS)
def test_parse_with_header(one_variant, case_obj, rank_results_header):
"""docstring for test_parse_all_variants"""
parsed_variant = parse_variant(one_variant, case_obj,
rank_results_header=rank_results_header)
assert parsed_variant['chromosome'] == '1'
assert parsed_variant['rank_result']['Consequence'] == 1
def test_parse_minimal(one_variant, case_obj):
"""Test to parse a minimal variant"""
parsed_variant = parse_variant(one_variant,
case_obj,
variant_type="clinical")
assert parsed_variant["position"] == int(one_variant.POS)
assert parsed_variant["category"] == "snv"
def test_load_vep97_parsed_variant(one_vep97_annotated_variant,
real_populated_database, case_obj):
"""test first parsing and then loading a vep v97 annotated variant"""
# GIVEN a variant annotated using the following CSQ entry fields
csq_header = "Allele|Consequence|IMPACT|SYMBOL|Gene|Feature_type|Feature|BIOTYPE|EXON|INTRON|HGVSc|HGVSp|cDNA_position|CDS_position|Protein_position|Amino_acids|Codons|Existing_variation|DISTANCE|STRAND|FLAGS|SYMBOL_SOURCE|HGNC_ID|CANONICAL|TSL|APPRIS|CCDS|ENSP|SWISSPROT|TREMBL|UNIPARC|REFSEQ_MATCH|SOURCE|GIVEN_REF|USED_REF|BAM_EDIT|SIFT|PolyPhen|DOMAINS|HGVS_OFFSET|MOTIF_NAME|MOTIF_POS|HIGH_INF_POS|MOTIF_SCORE_CHANGE|MES-NCSS_downstream_acceptor|MES-NCSS_downstream_donor|MES-NCSS_upstream_acceptor|MES-NCSS_upstream_donor|MES-SWA_acceptor_alt|MES-SWA_acceptor_diff|MES-SWA_acceptor_ref|MES-SWA_acceptor_ref_comp|MES-SWA_donor_alt|MES-SWA_donor_diff|MES-SWA_donor_ref|MES-SWA_donor_ref_comp|MaxEntScan_alt|MaxEntScan_diff|MaxEntScan_ref|LoFtool|ExACpLI|GERP++_NR|GERP++_RS|REVEL_rankscore|phastCons100way_vertebrate|phyloP100way_vertebrate|CLINVAR|CLINVAR_CLNSIG|CLINVAR_CLNVID|CLINVAR_CLNREVSTAT|genomic_superdups_frac_match"
header = [word.upper() for word in csq_header.split("|")]
# WHEN parsed using
parsed_vep97_annotated_variant = parse_variant(
variant=one_vep97_annotated_variant, vep_header=header, case=case_obj)
# GIVEN a database without any variants
adapter = real_populated_database
assert adapter.variant_collection.find_one() is None
# WHEN loading the variant into the database
adapter.load_variant(variant_obj=parsed_vep97_annotated_variant)
# THEN the variant is loaded with the fields correctly parsed
# revel score
variant = adapter.variant_collection.find_one()
assert isinstance(variant["revel_score"], float)
# conservation fields
for key, value in variant["conservation"].items():
assert value == ["NotConserved"]
# clinvar fields
assert isinstance(variant["clnsig"][0]["accession"], int)
assert variant["clnsig"][0]["value"] in REV_CLINSIG_MAP # can be str or int
assert isinstance(variant["clnsig"][0]["revstat"], str) # str
def test_load_cancer_SV_variant(one_cancer_manta_SV_variant,
real_populated_database, cancer_case_obj):
""" Test loading a cancer SV variant into a mongo database """
# GIVEN a database containing one cancer case
adapter = real_populated_database
adapter.case_collection.insert_one(cancer_case_obj)
assert sum(1
for i in adapter.case_collection.find({"track": "cancer"})) == 1
# AND no variants
assert adapter.variant_collection.find_one() is None
# WHEN parsing a SV variant
parsed_cancer_SV_variant = parse_variant(
variant=one_cancer_manta_SV_variant, case=cancer_case_obj)
# WHEN loading the variant into the database
adapter.load_variant(variant_obj=parsed_cancer_SV_variant)
# THEN the variant should have been parsed correctly
variant = adapter.variant_collection.find_one()
assert variant["variant_type"] == "clinical"
assert variant["chromosome"]
assert variant["position"]
assert variant["end"]
assert isinstance(variant["somatic_score"], int)
def test_parse_minimal(one_variant, case_obj):
"""Test to parse a minimal variant"""
parsed_variant = parse_variant(one_variant,
case_obj,
variant_type='clinical')
assert parsed_variant['position'] == int(one_variant.POS)
assert parsed_variant['category'] == 'snv'
def test_parse_clinsig_vep97(one_vep97_annotated_variant,
real_populated_database, case_obj):
"""Test Clinsig parsing in a VEP97 formatted VCF"""
# GIVEN a variant annotated using the following CSQ entry fields
csq_header = "Allele|Consequence|IMPACT|SYMBOL|Gene|Feature_type|Feature|BIOTYPE|EXON|INTRON|HGVSc|HGVSp|cDNA_position|CDS_position|Protein_position|Amino_acids|Codons|Existing_variation|DISTANCE|STRAND|FLAGS|SYMBOL_SOURCE|HGNC_ID|CANONICAL|TSL|APPRIS|CCDS|ENSP|SWISSPROT|TREMBL|UNIPARC|REFSEQ_MATCH|SOURCE|GIVEN_REF|USED_REF|BAM_EDIT|SIFT|PolyPhen|DOMAINS|HGVS_OFFSET|MOTIF_NAME|MOTIF_POS|HIGH_INF_POS|MOTIF_SCORE_CHANGE|MES-NCSS_downstream_acceptor|MES-NCSS_downstream_donor|MES-NCSS_upstream_acceptor|MES-NCSS_upstream_donor|MES-SWA_acceptor_alt|MES-SWA_acceptor_diff|MES-SWA_acceptor_ref|MES-SWA_acceptor_ref_comp|MES-SWA_donor_alt|MES-SWA_donor_diff|MES-SWA_donor_ref|MES-SWA_donor_ref_comp|MaxEntScan_alt|MaxEntScan_diff|MaxEntScan_ref|GERP++_NR|GERP++_RS|REVEL_rankscore|phastCons100way_vertebrate|phyloP100way_vertebrate|LoFtool|ExACpLI|CLINVAR|CLINVAR_CLNSIG|CLINVAR_CLNVID|CLINVAR_CLNREVSTAT|genomic_superdups_frac_match"
header = [word.upper() for word in csq_header.split("|")]
# WHEN parsed using the parse_variant method
parsed_vep97_annotated_variant = parse_variant(
variant=one_vep97_annotated_variant, vep_header=header, case=case_obj)
# GIVEN a database without any variants
adapter = real_populated_database
assert adapter.variant_collection.find_one() is None
# WHEN loading the variant into the database
adapter.load_variant(variant_obj=parsed_vep97_annotated_variant)
# THEN the variant is loaded with the fields correctly parsed
variant = adapter.variant_collection.find_one()
# Clinvar fields shoud be correctly parsed:
first_clnsig = variant["clnsig"][0]
assert first_clnsig
# Clinvar accession should be a numberical value
assert isinstance(first_clnsig["accession"], int)
# Value field should be a string (i.e. pathogenic, benign,..)
assert isinstance(first_clnsig["value"], str)
# Revstat field should be also a string (i.e. criteria_provided, ..)
assert isinstance(first_clnsig["revstat"], str)
def test_parse_with_header(one_variant, case_obj, rank_results_header):
"""docstring for test_parse_all_variants"""
parsed_variant = parse_variant(one_variant,
case_obj,
rank_results_header=rank_results_header)
assert parsed_variant['chromosome'] == '1'
assert parsed_variant['rank_result']['Consequence'] == 1
def test_build_minimal(case_obj, cyvcf2_variant):
## GIVEN a variant with minimal information
variant = cyvcf2_variant
parsed_variant = parse_variant(variant, case_obj)
assert "ids" in parsed_variant
variant_obj = build_variant(parsed_variant, INSTITUTE_ID)
assert variant_obj["_id"] == parsed_variant["ids"]["document_id"]
def test_parse_with_header(one_variant, case_obj, rank_results_header):
"""docstring for test_parse_all_variants"""
parsed_variant = parse_variant(one_variant,
case_obj,
rank_results_header=rank_results_header)
assert parsed_variant["chromosome"] == "1"
assert parsed_variant["rank_result"]["Consequence"] == 1
def test_parse_cadd(variants, case_obj):
# GIVEN some parsed variant dicts
for variant in variants:
# WHEN score is present
if "CADD" in variant.INFO:
cadd_score = float(variant.INFO["CADD"])
parsed_variant = parse_variant(variant, case_obj)
# THEN make sure that the cadd score is parsed correct
assert parsed_variant["cadd_score"] == cadd_score
def test_parse_cadd(variants, case_obj):
# GIVEN some parsed variant dicts
for variant in variants:
# WHEN score is present
if 'CADD' in variant.INFO:
cadd_score = float(variant.INFO['CADD'])
parsed_variant = parse_variant(variant, case_obj)
# THEN make sure that the cadd score is parsed correct
assert parsed_variant['cadd_score'] == cadd_score
def parsed_sv_variants(request, sv_variants, case_obj):
"""Get a generator with parsed variants"""
print('')
individual_positions = {}
for i, ind in enumerate(sv_variants.samples):
individual_positions[ind] = i
return (parse_variant(variant, case_obj,
individual_positions=individual_positions)
for variant in sv_variants)
def test_parse_small_str(one_str_variant, case_obj):
parsed_variant = parse_variant(one_str_variant, case_obj, category="str")
assert parsed_variant["category"] == "str"
assert parsed_variant["str_status"] == one_str_variant.INFO["STR_STATUS"]
assert parsed_variant["str_normal_max"] == one_str_variant.INFO[
"STR_NORMAL_MAX"]
assert (parsed_variant["str_pathologic_min"] ==
one_str_variant.INFO["STR_PATHOLOGIC_MIN"])
assert parsed_variant["position"] == int(one_str_variant.POS)
def test_parse_hmtvar(cyvcf2_variant, case_obj):
"""Test parsing HmtVar value from variant annotated with HmtNote"""
# GIVEN a variant containing HmtVar key in the INFO field:
cyvcf2_variant.INFO["HmtVar"] = "39192"
# THEN make sure that it is parsed correctly
hmtvar_variant_id = int(cyvcf2_variant.INFO["HmtVar"])
parsed_variant = parse_variant(cyvcf2_variant, case_obj)
assert parsed_variant["hmtvar_variant_id"] == hmtvar_variant_id
def test_parse_many_svs(sv_variants, case_obj):
"""docstring for test_parse_all_variants"""
for variant in sv_variants:
try:
parsed_variant = parse_variant(variant, case_obj)
except VcfError:
for info in variant['info_dict']:
print(info, variant['info'])
assert False
assert parsed_variant['chromosome'] == variant.CHROM
def test_parse_many_svs(sv_variants, case_obj):
"""docstring for test_parse_all_variants"""
for variant in sv_variants:
try:
parsed_variant = parse_variant(variant, case_obj)
except VcfError:
for info in variant['info_dict']:
print(info, variant['info'])
assert False
assert parsed_variant['chromosome'] == variant.CHROM
def test_parse_many_strs(str_variants, case_obj):
"""docstring for test_parse_many_strs"""
for variant in str_variants:
try:
parsed_variant = parse_variant(variant, case_obj, category="str")
except VcfError:
for info in variant["info_dict"]:
print(info, variant["info"])
assert False
assert parsed_variant["chromosome"] == variant.CHROM
def test_parse_mitomapassociateddiseases(cyvcf2_variant, case_obj):
"""Test parsing HmtVar value from variant annotated with HmtNote"""
# GIVEN a variant containing HmtVar key in the INFO field:
cyvcf2_variant.INFO["MitomapAssociatedDiseases"] = "LHON"
# THEN make sure that it is parsed correctly
mitomap_associated_diseases = cyvcf2_variant.INFO[
"MitomapAssociatedDiseases"]
parsed_variant = parse_variant(cyvcf2_variant, case_obj)
assert parsed_variant[
"mitomap_associated_diseases"] == mitomap_associated_diseases
def parsed_cancer_variant(request, cancer_variants, one_cancer_variant,
cancer_case_obj):
"""Return a parsed variant"""
individual_positions = {}
for i, ind in enumerate(cancer_variants.samples):
individual_positions[ind] = i
variant_dict = parse_variant(one_cancer_variant,
cancer_case_obj,
individual_positions=individual_positions)
return variant_dict
def test_parse_old_obs_archive_SV(case_obj, cyvcf2_variant):
"""Test parsing local_obs_old and local_obs_old_freq off a variant VCF file"""
nr_old_obs = 22
freq_old_obs = 0.3
# GIVEN a VCF variant containing old local observations stats
cyvcf2_variant.INFO["clinical_genomics_loqusObs"] = nr_old_obs
cyvcf2_variant.INFO["clinical_genomics_loqusFrq"] = freq_old_obs
# WHEN parsing the variant
parsed_var = parse_variant(cyvcf2_variant, case_obj)
# THEN the parsed variant should contain these values
assert parsed_var["local_obs_old"] == nr_old_obs
assert parsed_var["local_obs_old_freq"] == freq_old_obs
def test_parse_revel(cyvcf2_variant, case_obj):
## GIVEN a variant with REVEL score in the CSQ entry
csq_header = "ALLELE|CONSEQUENCE|REVEL_rankscore"
csq_entry = "C|missense_variant|0.75,C|missense_variant|0.75" # mimic a variant with transcripts
cyvcf2_variant.INFO["CSQ"] = csq_entry
header = [word.upper() for word in csq_header.split("|")]
# WHEN the variant is parsed
parsed_variant = parse_variant(variant=cyvcf2_variant,
case=case_obj,
vep_header=header)
# THEN the REVEL score should be parsed correctly
assert parsed_variant["revel_score"] == 0.75
def test_build_minimal(case_obj):
## GIVEN a variant with minimal information
class Cyvcf2Variant(object):
def __init__(self):
self.CHROM = '1'
self.REF = 'A'
self.ALT = ['C']
self.POS = 10
self.end = 11
self.FILTER = None
self.ID = '.'
self.QUAL = None
self.var_type = 'snp'
self.INFO = {}
variant = Cyvcf2Variant()
parsed_variant = parse_variant(variant, case_obj)
assert 'ids' in parsed_variant
variant_obj = build_variant(parsed_variant, INSTITUTE_ID)
assert variant_obj['_id'] == parsed_variant['ids']['document_id']
def test_build_minimal(case_obj):
## GIVEN a variant with minimal information
class Cyvcf2Variant(object):
def __init__(self):
self.CHROM = '1'
self.REF = 'A'
self.ALT = ['C']
self.POS = 10
self.end = 11
self.FILTER = None
self.ID = '.'
self.QUAL = None
self.var_type = 'snp'
self.INFO = {}
variant = Cyvcf2Variant()
parsed_variant = parse_variant(variant, case_obj)
assert 'ids' in parsed_variant
variant_obj = build_variant(parsed_variant, INSTITUTE_ID)
assert variant_obj['_id'] == parsed_variant['ids']['document_id']
def test_build_minimal(case_obj):
## GIVEN a variant with minimal information
class Cyvcf2Variant(object):
def __init__(self):
self.CHROM = "1"
self.REF = "A"
self.ALT = ["C"]
self.POS = 10
self.end = 11
self.FILTER = None
self.ID = "."
self.QUAL = None
self.var_type = "snp"
self.INFO = {}
variant = Cyvcf2Variant()
parsed_variant = parse_variant(variant, case_obj)
assert "ids" in parsed_variant
variant_obj = build_variant(parsed_variant, INSTITUTE_ID)
assert variant_obj["_id"] == parsed_variant["ids"]["document_id"]
def parsed_str_variant(request, one_str_variant, case_obj):
"""Return a parsed variant"""
print("")
variant_dict = parse_variant(one_str_variant, case_obj, category="str")
return variant_dict
def parsed_sv_variant(request, one_sv_variant, case_obj):
"""Return a parsed variant"""
print('')
variant_dict = parse_variant(one_sv_variant, case_obj)
return variant_dict
def test_parse_small_sv(one_sv_variant, case_obj):
parsed_variant = parse_variant(one_sv_variant, case_obj)
assert parsed_variant['category'] == 'sv'
assert parsed_variant['sub_category'] == one_sv_variant.INFO['SVTYPE'].lower()
assert parsed_variant['position'] == int(one_sv_variant.POS)
def test_parse_many_snvs(variants, case_obj):
"""docstring for test_parse_all_variants"""
for variant in variants:
parsed_variant = parse_variant(variant, case_obj)
assert parsed_variant['chromosome'] == variant.CHROM
def test_parse_customannotation(one_variant_customannotation, case_obj):
"""Test parsing of custom annotations"""
parsed_variant = parse_variant(one_variant_customannotation, case_obj)
assert parsed_variant["custom"] == [["key1", "val1"], ["key2", "val2"]]
def load_variants(adapter, variant_file, case_obj, variant_type='clinical',
category='snv', rank_threshold=5, chrom=None, start=None,
end=None):
"""Load all variant in variants
Args:
adapter(MongoAdapter)
variant_file(str): Path to variant file
case(Case)
variant_type(str)
category(str): 'snv' or 'sv'
rank_threshold(int)
chrom(str)
start(int)
end(int)
"""
institute_obj = adapter.institute(institute_id=case_obj['owner'])
if not institute_obj:
raise IntegrityError("Institute {0} does not exist in"
" database.".format(case_obj['owner']))
gene_to_panels = adapter.gene_to_panels()
hgncid_to_gene = adapter.hgncid_to_gene()
coordinates = {}
vcf_obj = VCF(variant_file)
rank_results_header = parse_rank_results_header(vcf_obj)
vep_header = parse_vep_header(vcf_obj)
# This is a dictionary to tell where ind are in vcf
individual_positions = {}
for i,ind in enumerate(vcf_obj.samples):
individual_positions[ind] = i
logger.info("Start inserting variants into database")
start_insertion = datetime.now()
start_five_thousand = datetime.now()
nr_variants = 0
nr_inserted = 0
inserted = 1
coordinates = False
if chrom:
coordinates = {
'chrom': chrom,
'start': start,
'end': end
}
try:
for nr_variants, variant in enumerate(vcf_obj):
rank_score = parse_rank_score(
variant.INFO.get('RankScore'),
case_obj['display_name']
)
variant_obj = None
add_variant = False
if coordinates or (rank_score > rank_threshold):
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
)
add_variant = True
# If there are coordinates the variant should be loaded
if coordinates:
if not check_coordinates(parsed_variant, coordinates):
add_variant = False
if add_variant:
variant_obj = build_variant(
variant=parsed_variant,
institute_id=institute_obj['_id'],
gene_to_panels=gene_to_panels,
hgncid_to_gene=hgncid_to_gene,
)
try:
load_variant(adapter, variant_obj)
nr_inserted += 1
except IntegrityError as error:
pass
if (nr_variants != 0 and nr_variants % 5000 == 0):
logger.info("%s variants parsed" % str(nr_variants))
logger.info("Time to parse variants: {} ".format(
datetime.now() - start_five_thousand))
start_five_thousand = datetime.now()
if (nr_inserted != 0 and (nr_inserted * inserted) % (1000 * inserted) == 0):
logger.info("%s variants inserted" % nr_inserted)
inserted += 1
except Exception as error:
if not coordinates:
logger.warning("Deleting inserted variants")
delete_variants(adapter, case_obj, variant_type)
raise error
logger.info("All variants inserted.")
logger.info("Number of variants in file: {0}".format(nr_variants + 1))
logger.info("Number of variants inserted: {0}".format(nr_inserted))
logger.info("Time to insert variants:{0}".format(datetime.now() - start_insertion))
def test_compounds_region(real_populated_database, case_obj,
variant_clinical_file):
"""When loading the variants not all variants will be loaded, only the ones that
have a rank score above a treshold.
This implies that some compounds will have the status 'not_loaded'=True.
When loading all variants for a region then all variants should
have status 'not_loaded'=False.
"""
adapter = real_populated_database
variant_type = "clinical"
category = "snv"
## GIVEN a database without any variants
assert adapter.variant_collection.find_one() is None
institute_obj = adapter.institute_collection.find_one()
institute_id = institute_obj["_id"]
## WHEN loading variants into the database without updating compound information
vcf_obj = VCF(variant_clinical_file)
rank_results_header = parse_rank_results_header(vcf_obj)
vep_header = parse_vep_header(vcf_obj)
individual_positions = {}
for i, ind in enumerate(vcf_obj.samples):
individual_positions[ind] = i
variants = []
for i, variant in enumerate(vcf_obj):
parsed_variant = parse_variant(
variant=variant,
case=case_obj,
variant_type="clinical",
rank_results_header=rank_results_header,
vep_header=vep_header,
individual_positions=individual_positions,
category="snv",
)
variant_obj = build_variant(variant=parsed_variant,
institute_id=institute_id)
variants.append(variant_obj)
# Load all variants
adapter.variant_collection.insert_many(variants)
print("Nr variants: {0}".format(len(variants)))
## THEN assert that the variants does not have updated compound information
nr_compounds = 0
for var in adapter.variant_collection.find():
if not var.get("compounds"):
continue
for comp in var["compounds"]:
if "genes" in comp:
assert False
if "not_loaded" in comp:
assert False
nr_compounds += 1
assert nr_compounds > 0
## WHEN updating all compounds for a case
adapter.update_case_compounds(case_obj)
hgnc_ids = set([gene["hgnc_id"] for gene in adapter.all_genes()])
nr_compounds = 0
## THEN assert that all compounds (within the gene defenition) are updated
for var in adapter.variant_collection.find():
cont = False
for hgnc_id in var["hgnc_ids"]:
if hgnc_id not in hgnc_ids:
cont = True
if cont:
continue
if not var.get("compounds"):
continue
for comp in var["compounds"]:
nr_compounds += 1
if not "genes" in comp:
# pp(var)
assert False
if not "not_loaded" in comp:
assert False
assert nr_compounds > 0
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.
"""
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)
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'])
# 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:
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
genomic_regions = genomic_intervals.get(var_chrom, IntervalTree()).search(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
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
def test_parse_many_snvs(variants, case_obj):
"""docstring for test_parse_all_variants"""
for variant in variants:
parsed_variant = parse_variant(variant, case_obj)
assert parsed_variant["chromosome"] == variant.CHROM
def test_compounds_region(real_populated_database, case_obj, variant_clinical_file):
"""When loading the variants not all variants will be loaded, only the ones that
have a rank score above a treshold.
This implies that some compounds will have the status 'not_loaded'=True.
When loading all variants for a region then all variants should
have status 'not_loaded'=False.
"""
adapter = real_populated_database
variant_type = 'clinical'
category = 'snv'
## GIVEN a database without any variants
assert adapter.variant_collection.find().count() == 0
institute_obj = adapter.institute_collection.find_one()
institute_id = institute_obj['_id']
## WHEN loading variants into the database without updating compound information
vcf_obj = VCF(variant_clinical_file)
rank_results_header = parse_rank_results_header(vcf_obj)
vep_header = parse_vep_header(vcf_obj)
individual_positions = {}
for i, ind in enumerate(vcf_obj.samples):
individual_positions[ind] = i
variants = []
for i,variant in enumerate(vcf_obj):
parsed_variant = parse_variant(
variant=variant,
case=case_obj,
variant_type='clinical',
rank_results_header=rank_results_header,
vep_header=vep_header,
individual_positions=individual_positions,
category='snv',
)
variant_obj = build_variant(
variant=parsed_variant,
institute_id=institute_id,
)
variants.append(variant_obj)
# Load all variants
adapter.variant_collection.insert_many(variants)
print("Nr variants: {0}".format(len(variants)))
## THEN assert that the variants does not have updated compound information
nr_compounds = 0
for var in adapter.variant_collection.find():
if not var.get('compounds'):
continue
for comp in var['compounds']:
if 'genes' in comp:
assert False
if 'not_loaded' in comp:
assert False
nr_compounds += 1
assert nr_compounds > 0
## WHEN updating all compounds for a case
adapter.update_case_compounds(case_obj)
hgnc_ids = set([gene['hgnc_id'] for gene in adapter.all_genes()])
nr_compounds = 0
## THEN assert that all compounds (within the gene defenition) are updated
for var in adapter.variant_collection.find():
cont = False
for hgnc_id in var['hgnc_ids']:
if hgnc_id not in hgnc_ids:
cont = True
if cont:
continue
if not var.get('compounds'):
continue
for comp in var['compounds']:
nr_compounds += 1
if not 'genes' in comp:
# pp(var)
assert False
if not 'not_loaded' in comp:
assert False
assert nr_compounds > 0
def test_parse_minimal(one_variant, case_obj):
"""Test to parse a minimal variant"""
parsed_variant = parse_variant(one_variant, case_obj, variant_type='clinical')
assert parsed_variant['position'] == int(one_variant.POS)
assert parsed_variant['category'] == 'snv'
