def _parse_hit(cls, hit):
# Sometimes (e.g. rs203319) a SNP will have two alleles, one with
# a SnpEff annotation and other without one, so this key will be
# missing:
if 'snpeff' not in hit:
return None
annotations = hit['snpeff']['ann']
# Make sure the annotations are always a *list* of annotations
# Right now, myvariant client sometimes returns a list of annotaions
# dictionaries and sometimes a single annotation dictionary.
annotations = listify(annotations)
for annotation in annotations:
annotation['genomic_allele'] = hit['allele']
annotation['coding_allele'] = \
infer_annotated_allele(annotation.get('hgvs_c'))
annotation['effects'] = annotation.get('effect', '').split('&')
if 'effect' in annotation:
del (annotation['effect'])
if 'hgvs_p' in annotation:
annotation['hgvs_p'] = parse_prot_change(annotation['hgvs_p'])
return annotations
def _parse_hit(cls, hit):
annotation = {'myvariant_hgvs_g': hit['_id']}
annotation['genomic_allele'] = infer_annotated_allele(hit['_id'])
annotation.update(cls._parse_hgvs_from_clinvar(hit))
annotation.update(cls._parse_hgvs_from_snpeff(hit))
annotation.update(cls._parse_hgvs_from_evs(hit))
annotation.update(cls._parse_hgvs_from_emv(hit))
return annotation
def _parse_hit(cls, hit):
if 'clinvar' not in hit:
return []
rcv_annotations = listify(hit['clinvar']['rcv'])
for annotation in rcv_annotations:
if 'clinical_significance' in annotation:
clinsig_string = annotation['clinical_significance']
annotation['clinical_significances'] = \
cls._split_significance_string(clinsig_string)
del (annotation['clinical_significance'])
else:
annotation['clinical_significances'] = []
conditions = listify(annotation['conditions'])
annotation['conditions'] = conditions
annotation['condition_names'] = cls._extract_condition_names(
conditions)
for condition in annotation['conditions']:
condition.update(cls._parse_condition_identifiers(condition))
if 'identifiers' in condition:
del (condition['identifiers'])
annotation['url'] = cls._url(annotation['accession'])
annotation['variant_url'] = cls._variant_url(
hit['clinvar']['variant_id'])
name_info = cls._parse_preferred_name(annotation['preferred_name'])
annotation.update(name_info)
if 'prot_change' in annotation:
annotation['prot_change'] = \
parse_prot_change(annotation['prot_change'])
annotation['genomic_allele'] = hit['allele']
annotation['coding_allele'] = \
infer_annotated_allele(annotation.get('cds_change', ''))
# Copy the variant info to each particular RCV entry:
for key, value in hit['clinvar'].items():
if key == 'rcv':
continue
# Flatten the dicts one level
if isinstance(value, dict):
for key2, val2 in value.items():
compound_key = '{}_{}'.format(key, key2)
annotation[compound_key] = val2
continue
annotation[key] = value
return rcv_annotations
def _parse_annotation(cls, hits):
# Gather the hits (i.e. different alleles) in the same order always:
hits = sorted(hits, key=itemgetter('_id'))
for hit in hits:
hit['allele'] = infer_annotated_allele(hit['_id'])
annotations = [cls._parse_hit(hit) for hit in hits]
# Parsing hits might result in empty values, since some hits do not
# contain any relevant annotation. We need to remove those explicitely:
annotations = [ann for ann in annotations if ann]
# If the annotations are already lists, merge them into a flat list:
if all(isinstance(ann, list) for ann in annotations):
annotations = list(chain.from_iterable(annotations))
# Let subclasses do some optional extra parsing here
annotations = cls._parse_annotations_hook(annotations)
if annotations:
return annotations
def test_infer_annotated_allele(cds_change, expected_allele):
assert infer_annotated_allele(cds_change) == expected_allele