def read_variant(variant_id: str, db: MongoAdapter = Depends(database)):
variant = db.get_variant({"_id": variant_id})
if not variant:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
detail=f"Variant {variant_id} not found")
variant["total"] = db.nr_cases(snv_cases=True, sv_cases=False)
return variant
def insert_sv_variants(adapter: MongoAdapter, case_obj: Case) -> None:
"""Build sv_variant documents and insert them into database on the fly, one at a time"""
for variant in VCF(case_obj.vcf_sv_path, threads=settings.cyvcf_threads):
variant_id = get_variant_id(variant=variant)
ref = variant.REF
alt = variant.ALT[0]
coordinates = get_coords(variant)
chrom = coordinates["chrom"]
pos = coordinates["pos"]
variant_obj = Variant(
variant_id=variant_id,
chrom=chrom,
pos=pos,
end=coordinates["end"],
ref=ref,
alt=alt,
end_chrom=coordinates["end_chrom"],
sv_type=coordinates["sv_type"],
sv_len=coordinates["sv_length"],
case_id=case_obj.case_id,
homozygote=0,
hemizygote=0,
is_sv=True,
id_column=variant.ID,
)
adapter.add_structural_variant(variant=variant_obj, max_window=settings.load_sv_window)
def read_case(case_id: str, db: MongoAdapter = Depends(database)):
"""Return a specific case given petname ID"""
case = db.case({"case_id": case_id})
if not case:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
detail=f"Case {case_id} not found")
return JSONResponse(jsonable_encoder(Case(**case)),
status_code=status.HTTP_200_OK)
def read_cases(db: MongoAdapter = Depends(database)):
"""Return counts of SNV and SV variants in database"""
nr_cases_snvs = db.nr_cases(snv_cases=True, sv_cases=False)
nr_cases_svs = db.nr_cases(snv_cases=False, sv_cases=True)
return dict(
nr_cases_snvs=nr_cases_snvs,
nr_cases_svs=nr_cases_svs,
)
def insert_snv_variants(adapter: MongoAdapter, case_obj: Case) -> None:
"""Build variant documents and bulk insert them into database"""
variants = []
for variant in VCF(case_obj.vcf_path, threads=settings.cyvcf_threads):
variant_id = get_variant_id(variant=variant)
ref = variant.REF
alt = variant.ALT[0]
coordinates = get_coords(variant)
chrom = coordinates["chrom"]
pos = coordinates["pos"]
found_homozygote = 0
found_hemizygote = 0
for ind_obj in case_obj.individuals:
ind_pos = ind_obj["ind_index"]
if int(variant.gt_quals[ind_pos]) < settings.load_gq_threshold:
continue
genotype = GENOTYPE_MAP[variant.gt_types[ind_pos]]
if genotype not in ["het", "hom_alt"]:
continue
if genotype == "hom_alt":
found_homozygote = 1
if (
chrom in ["X", "Y"]
and ind_obj["sex"] == 1
and not check_par(chrom, pos, genome_build=settings.genome_build)
):
found_hemizygote = 1
variant_obj = Variant(
variant_id=variant_id,
chrom=chrom,
pos=pos,
end=coordinates["end"],
ref=ref,
alt=alt,
end_chrom=coordinates["end_chrom"],
sv_type=coordinates["sv_type"],
sv_len=coordinates["sv_length"],
case_id=case_obj.case_id,
homozygote=found_homozygote,
hemizygote=found_hemizygote,
is_sv=False,
id_column=variant.ID,
)
variants.append(variant_obj)
adapter.add_variants(variants=variants)
def build_case_object(
adapter: MongoAdapter,
case_id: str,
profile_path: Union[Path, str],
vcf_path: Union[Path, str],
vcf_sv_path: Union[Path, str] = None,
) -> Case:
"""Build case document and insert into the database, return resulting document"""
# Parse MAF profiles from profile files and save in the case object
profiles: dict = get_profiles(adapter=adapter, vcf_file=profile_path)
# Check if profiles have any duplicates in the database
check_profile_duplicates(adapter=adapter, profiles=profiles)
# CHeck that SNV file has GQ field
check_vcf_gq_field(vcf_path=vcf_path)
# CHeck that SNV file doesnt have SV variants
check_snv_variant_types(vcf_path=vcf_path)
individuals = {
sample: Individual(
ind_id=sample,
case_id=case_id,
ind_index=sample_index,
profile=profile,
)
for sample_index, (sample, profile) in enumerate(profiles.items())
}
individuals_list: List = list(individuals.values())
case_object = Case(
case_id=case_id,
profile_path=profile_path,
vcf_path=vcf_path,
vcf_sv_path=vcf_sv_path,
nr_sv_variants=0,
nr_variants=get_vcf_variant_count(vcf_path=vcf_path),
individuals=individuals_list,
inds=individuals,
id=case_id,
)
if vcf_sv_path:
case_object.nr_sv_variants = get_vcf_variant_count(vcf_path=vcf_sv_path)
case_object.sv_individuals = individuals_list
case_object.sv_inds = individuals
adapter.add_case(case_object.dict(by_alias=True, exclude={"id"}))
return Case(**adapter.case({"case_id": case_id}))
def get_profiles(adapter: MongoAdapter, vcf_file: str) -> Dict[str, str]:
"""
Reads VCF file containing one or more samples.
Creates a dictionary where each sample ID from VCF file is a key.
Retrieves coordinates for each variant from loqusdb.profile_variants
Adds each variant of each sample as value of the dictionary.
Returns a dictionary :
{SAMPLE_ID : [var1, var2, ..., var50]}
"""
vcf = VCF(vcf_file, threads=settings.cyvcf_threads)
individuals = vcf.samples
profiles = {individual: [] for individual in individuals}
for profile_variant in adapter.profile_variants():
ref = profile_variant["ref"]
alt = profile_variant["alt"]
pos = profile_variant["pos"]
end = pos + 1
chrom = profile_variant["chrom"]
region = f"{chrom}:{pos}-{end}"
# Find variants in region
found_variant = False
for variant in vcf(region):
variant_id = get_variant_id(variant)
# If variant id i.e. chrom_pos_ref_alt matches
if variant_id == profile_variant["_id"]:
found_variant = True
# find genotype for each individual in vcf
for i, individual in enumerate(individuals):
genotype = GENOTYPE_MAP[variant.gt_types[i]]
if genotype == "hom_alt":
gt_str = f"{alt}{alt}"
elif genotype == "het":
gt_str = f"{ref}{alt}"
else:
gt_str = f"{ref}{ref}"
# Append genotype to profile string of individual
profiles[individual].append(gt_str)
# Break loop if variant is found in region
break
# If no call was found for variant, give all samples a hom ref genotype
if not found_variant:
for individual in individuals:
profiles[individual].append(f"{ref}{ref}")
return profiles
def database(uri: str = None, db_name: str = None) -> MongoAdapter:
uri = uri or settings.uri
db_name = db_name or settings.db_name
try:
client = get_client(uri=uri, )
except DB_Error:
raise HTTPException(status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
detail="Could not connect to database")
return MongoAdapter(client, db_name=db_name)
def cli(ctx, database, username, password, authdb, port, host, uri, verbose,
config, test, genome_build):
"""loqusdb: manage a local variant count database."""
loglevel = "INFO"
if verbose:
loglevel = "DEBUG"
coloredlogs.install(level=loglevel)
LOG.info("Running loqusdb version %s", __version__)
configs = {}
if config:
try:
configs = yaml.safe_load(config)
except yaml.YAMLError as err:
LOG.warning(err)
ctx.abort()
uri = configs.get("uri") or uri
if test:
uri = "mongomock://"
try:
client = get_client(
host=configs.get("host") or host,
port=configs.get("port") or port,
username=configs.get("username") or username,
password=configs.get("password") or password,
authdb=authdb or database or "loqusdb",
uri=uri,
)
except DB_Error as err:
LOG.warning(err)
ctx.abort()
database = configs.get("db_name") or database
if not database:
database = "loqusdb"
if uri:
uri_info = uri_parser.parse_uri(uri)
database = uri_info.get("database")
adapter = MongoAdapter(client, db_name=database)
genome_build = genome_build or configs.get("genome_build") or GRCH37
ctx.obj = {}
ctx.obj["db"] = database
if uri:
ctx.obj["uri"] = uri
else:
ctx.obj["port"] = port
ctx.obj["host"] = host
ctx.obj["adapter"] = adapter
ctx.obj["version"] = __version__
ctx.obj["genome_build"] = genome_build
def read_sv(
chrom: str,
pos: int,
end: int,
sv_type: str,
db: MongoAdapter = Depends(database),
end_chrom: str = None,
):
structural_variant = db.get_structural_variant({
"chrom": chrom,
"end_chrom": end_chrom or chrom,
"sv_type": sv_type,
"pos": pos,
"end": end,
})
if not structural_variant:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
detail="Variant not found")
structural_variant["total"] = db.nr_cases(snv_cases=False, sv_cases=True)
return structural_variant
def delete_case(case_id: str, db: MongoAdapter = Depends(database)):
"""Delete a specific case given petname ID"""
existing_case = db.case({"case_id": case_id})
if not existing_case:
return JSONResponse(f"Case {case_id} does not exist",
status_code=status.HTTP_404_NOT_FOUND)
try:
delete(adapter=db,
case_obj=existing_case,
genome_build=settings.genome_build)
return JSONResponse(f"Case {case_id} had been deleted",
status_code=status.HTTP_200_OK)
except Exception as e:
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=
f"Error {e.__class__.__name__}: {e}; Case may be partially deleted",
)
def check_profile_duplicates(adapter: MongoAdapter, profiles: dict) -> None:
"""Compare profile variants from upload with all profiles of all cases in database.
Raises error if profile matches any of the existing profiles"""
for existing_case in adapter.cases():
if existing_case.get("individuals") is None:
continue
for individual in existing_case["individuals"]:
if not individual.get("profile"):
continue
for sample, profile in profiles.items():
similarity = compare_profiles(profile, individual["profile"])
if similarity >= settings.load_hard_threshold:
raise ProfileDuplicationError(
f"Profile of sample {sample} "
f"matches existing profile {individual.get('ind_id')}"
)
def load_case(
case_id: str,
snv_file: str,
profile_file: str,
sv_file: Optional[str] = None,
db: MongoAdapter = Depends(database),
):
"""Upload a case to loqusdb"""
if db.case({"case_id": case_id}):
return JSONResponse(f"Case {case_id} already exists",
status_code=status.HTTP_409_CONFLICT)
if ((sv_file and not Path(sv_file).exists())
or not Path(snv_file).exists() or not Path(profile_file).exists()):
raise HTTPException(
detail="Input file path does not exist",
status_code=status.HTTP_406_NOT_ACCEPTABLE,
)
try:
case_object: Case = build_case_object(
case_id=case_id,
vcf_path=snv_file,
vcf_sv_path=sv_file,
profile_path=profile_file,
adapter=db,
)
insert_case_variants(adapter=db, case_obj=case_object)
return JSONResponse(jsonable_encoder(case_object),
status_code=status.HTTP_200_OK)
except LoqusdbAPIError as e:
LOG.error(e)
raise HTTPException(
detail=f"Exception {e.__class__.__name__}: {e.message}",
status_code=status.HTTP_400_BAD_REQUEST,
)
except Exception as e:
LOG.error(e)
raise HTTPException(
detail=f"Exception {e.__class__.__name__} {e}",
status_code=status.HTTP_400_BAD_REQUEST,
)
def real_mongo_adapter(request, real_mongo_client):
"""Return a mongo adapter"""
db_name = REAL_DATABASE
adapter = MongoAdapter(real_mongo_client, db_name)
return adapter
def mongo_adapter(request, mongo_client):
"""Return a mongo adapter"""
db_name = TEST_DATABASE
adapter = MongoAdapter(mongo_client, db_name)
return adapter
def test_connect(mongo_client):
db_name = "test"
adapter = MongoAdapter(mongo_client, db_name)
assert adapter.db_name == db_name
def test_init_app(mongo_client):
app = MockFlaskApp()
adapter = MongoAdapter()
adapter.init_app(app)
assert adapter.db_name == app.config["MONGO_DBNAME"]
