def obtain_variant_set_for_set(Set):
variant_set = variants.VariantSet()
variant_set.id = '{}-{}'.format(DATASET_ID, Set)
variant_set.name = '{}-{}'.format(SETNAME, Set)
variant_set.dataset_id = DATASET_ID
variant_set.reference_set_id = '{}-{}'.format(REFERENCE_SET_BASE, Set)
brca_meta(variant_set.metadata, DATASET_ID)
return variant_set
def get_variant_set(request, variant_set_id):
"""/variantsets/<set id> method"""
if not variant_set_id:
return HttpResponseBadRequest(json.dumps(
ErrorMessages['variantSetId']),
content_type='application/json')
dataset, id_ = variant_set_id.split('-')
if id_ in SET_IDS and dataset == 'brca':
variant_set = variants.VariantSet()
variant_set.id = '{}-{}'.format(dataset, id_)
variant_set.name = '{}-{}'.format(SETNAME, id_)
variant_set.dataset_id = DATASET_ID
variant_set.reference_set_id = '{}-{}'.format(REFERENCE_SET_BASE, id_)
brca_meta(variant_set.metadata, id_)
resp = json_format.MessageToDict(variant_set, True)
return JsonResponse(resp)
else:
return JsonResponse({'Invalid Set Id': variant_set_id}, status=404)
def icgc2pb(input_stream):
"""
Convert a single icgc somatic variant tsv file into ga4gh variant protobuf
ICGC SSM Format: http://docs.icgc.org/submission/guide/icgc-simple-somatic-mutation-format/
"""
output_stream = io.BytesIO()
for index, icgc in enumerate(csv.DictReader(input_stream, delimiter="\t")):
output_stream.write(
ga4gh.Variant(id=index,
variant_set_id=icgc["icgc_sample_id"],
reference_name=icgc["chromosome"],
start=icgc["chromosome_start"],
end=icgc["chromosome_end"],
reference_bases=icgc["mutated_from_allele"],
alternate_bases=icgc["mutated_to_allele"]))
output_stream.write(
ga4gh.VariantSet(id=icgc["icgc_sample_id"],
dataset_id=icgc["project_code"]))
return output_stream
