def update_synonyms(self, sa): # SimpleAllele
if 'OtherNameList' in sa:
unifylistattribute(sa,
"OtherNameList",
"Name",
renamelistto='otherNames')
for i, name in enumerate(sa['otherNames']):
if isinstance(name, string_types):
sa['otherNames'][i] = {'#text': name}
def update_entry(self, entry):
del entry['id']
num(entry, "protein_count", int)
if 'abstract' in entry:
import json
entry['abstract'] = json.dumps(entry['abstract']) # , indent=4)
if 'taxonomy_distribution' in entry:
unifylistattribute(entry, 'taxonomy_distribution', 'taxon_data')
for i in entry['taxonomy_distribution']:
num(i, "proteins_count", int)
def update_simpleallele(self, sa):
self.update_synonyms(sa)
self.update_comment(sa)
unifylistattribute(sa,
"MolecularConsequenceList",
"MolecularConsequence",
renamelistto='molecularConsequence')
if 'molecularConsequence' in sa:
self.update_comment(sa['molecularConsequence'])
if 'FunctionalConsequence' in sa:
self.update_comment(sa['FunctionalConsequence'])
num(sa, 'AlleleID')
num(sa, 'VariationID')
def update_entry(self, entry): # ClinVar Variation Archive entry
if 'InterpretedRecord' in entry:
ir = entry['InterpretedRecord']
if 'SimpleAllele' in ir:
self.update_simpleallele(ir['SimpleAllele'])
if 'RCVList' in ir:
unifylistattribute(ir,
"RCVList",
"RCVAccession",
renamelistto='rcv')
for i, rcvacc in enumerate(ir['rcv']):
if isinstance(rcvacc, string_types):
ir['rcv'][i] = {'#text': rcvacc}
else:
unifylistattribute(rcvacc,
"InterpretedConditionList",
"InterpretedCondition",
renamelistto='interpretedCondition')
for j, ic in enumerate(rcvacc['interpretedCondition']):
if isinstance(ic, string_types):
rcvacc['interpretedCondition'][j] = {
'#text': ic
}
unifylistattribute(ir,
"ClinicalAssertionList",
"ClinicalAssertion",
renamelistto='clinicalAssertion')
for ca in ir['clinicalAssertion']:
num(ca, 'ID')
if "Interpretation" in ca:
self.update_date(ca)
self.update_comment(ca['Interpretation'])
unifylistattribute(ca,
"ObservedInList",
"ObservedIn",
renamelistto='observedIn')
if 'SimpleAllele' in ca:
sa = ca['SimpleAllele']
self.update_simpleallele(sa)
if 'Genotype' in ca:
self.update_genotype_haplotype(ca['Genotype'])
if 'Haplotype' in ca:
self.update_genotype_haplotype(ca['Haplotype'])
if 'TraitSet' in ca:
self.update_comment(ca['TraitSet'])
self.update_comment(ca)
for o in ca['observedIn']:
self.update_comment(o)
if not isinstance(o['Sample']['Species'], string_types):
o['Sample']['Species'] = o['Sample']['Species'][
'#text']
if 'TraitSet' in o:
self.update_comment(o['TraitSet'])
if 'ObservedData' in o:
self.update_comment(o['ObservedData'])
if 'Method' in o and 'ObsMethodAttribute' in o['Method']:
self.update_comment(o['Method']['ObsMethodAttribute'])