def get_short_name(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
output = None
accession = item['uniprot']['entry']['accession'][0]
if 'shortName' in item['uniprot']['entry']['protein']['recommendedName']:
evidence = Evidence()
recommendedName = item['uniprot']['entry']['protein'][
'recommendedName']
if type(recommendedName['shortName']) == str:
evidence.value = recommendedName['shortName']
else:
evidence.value = recommendedName['shortName']['#text']
if '@evidence' in recommendedName['shortName']:
evidence_numbers_in_db = recommendedName['shortName'][
'@evidence'].split(' ')
for evidence_number_in_db in evidence_numbers_in_db:
evidence_in_db = item['uniprot']['entry']['evidence'][
int(evidence_number_in_db) - 1]
if evidence_in_db['@key'] != evidence_number_in_db:
raise ValueError(
'Evidence number does not match evidence @key')
_add_reference_to_evidence(evidence, evidence_in_db)
evidence.add_reference({'database': 'UniProtKB', 'id': accession})
output = evidence
return output
def get_name(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
evidence = Evidence()
fullName = item['uniprot']['entry']['protein']['recommendedName'][
'fullName']
if type(fullName) == str:
evidence.value = fullName
elif type(fullName) == OrderedDict:
if '#text' in fullName:
evidence.value = fullName['#text']
if '@evidence' in fullName:
evidence_numbers_in_db = fullName['@evidence'].split()
for evidence_number_in_db in evidence_numbers_in_db:
evidence_in_db = item['uniprot']['entry']['evidence'][
int(evidence_number_in_db) - 1]
if evidence_in_db['@key'] != evidence_number_in_db:
raise ValueError(
'Evidence number does not match evidence @key')
_add_reference_to_evidence(evidence, evidence_in_db)
accession = item['uniprot']['entry']['accession'][0]
evidence.add_reference({'database': 'UniProtKB', 'id': accession})
return evidence
def get_alternative_names(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
from .get_short_name import get_short_name
from .get_uniprot import get_uniprot
output = []
uniprot = get_uniprot(item, entity=entity)
ref_uniprot = uniprot.references[0]
short_name = get_short_name(item)
if short_name is not None:
output.append(short_name)
if 'alternativeName' in item['uniprot']['entry']['protein']:
alternativeName = item['uniprot']['entry']['protein']['alternativeName']
if type(alternativeName)==OrderedDict:
alternativeName = [alternativeName]
if type(alternativeName)!=list:
raise ValueError("alternativeName is not a list")
for aux in alternativeName:
if type(aux)==OrderedDict:
for key, value in aux.items():
if key not in ['fullName', 'shortName']:
raise ValueError("Uknown alternative name type")
evidence = Evidence()
if type(value)==str:
evidence.value = value
else:
evidence.value = value['#text']
if '@evidence' in value:
evidence_numbers_in_db = value['@evidence'].split(' ')
for evidence_number_in_db in evidence_numbers_in_db:
evidence_in_db = item['uniprot']['entry']['evidence'][int(evidence_number_in_db)-1]
if evidence_in_db['@key']!=evidence_number_in_db:
raise ValueError('Evidence number does not match evidence @key')
_add_reference_to_evidence(evidence, evidence_in_db)
evidence.add_reference(ref_uniprot)
output.append(evidence)
else:
raise ValueError("Uknown alternativeName")
return output
def get_dbreference(item, entity='all', dbname=None):
from ._add_reference_to_evidence import _add_reference_to_evidence
uniprot = item['uniprot']['entry']['accession'][0]
dbReference = item['uniprot']['entry']['dbReference']
output = []
for db in dbReference:
if db['@type']==dbname:
accession = db['@id']
evidence = Evidence()
evidence.value = accession
if dbname=='ChEMBL':
evidence.add_reference({'database':'ChEMBL', 'id':accession})
elif dbname=='EC':
evidence.add_reference({'database':'EC', 'id':accession})
elif dbname=='DIP':
evidence.add_reference({'database':'DIP', 'id':accession})
elif dbname=='ELM':
evidence.add_reference({'database':'ELM', 'id':accession})
elif dbname=='IntAct':
evidence.add_reference({'database':'IntAct', 'id':accession})
elif dbname=='BindingDB':
evidence.add_reference({'database':'BindingDB', 'id':accession})
elif dbname=='BioGRID':
evidence.add_reference({'database':'BioGRID', 'id':accession})
elif dbname=='iPTMnet':
evidence.add_reference({'database':'iPTMnet', 'id':accession})
elif dbname=='MINT':
evidence.add_reference({'database':'MINT', 'id':accession})
elif dbname=='PhosphoSitePlus':
evidence.add_reference({'database':'PhosphoSitePlus', 'id':accession})
elif dbname=='ProDom':
evidence.add_reference({'database':'ProDom', 'id':accession})
elif dbname=='ProteinModelPortal':
evidence.add_reference({'database':'ProteinModelPortal', 'id':accession})
elif dbname=='STRING':
evidence.add_reference({'database':'STRING', 'id':accession})
elif dbname=='SMR':
evidence.add_reference({'database':'SwissModel', 'id':accession})
else:
raise ValueError('Database name not recognized')
if '@evidence' in db:
evidence_numbers_in_db = db['@evidence'].split()
for evidence_number_in_db in evidence_numbers_in_db:
evidence_in_db = item['uniprot']['entry']['evidence'][int(evidence_number_in_db)-1]
if evidence_in_db['@key']!=evidence_number_in_db:
raise ValueError('Evidence number does not match evidence @key')
_add_reference_to_evidence(evidence, evidence_in_db)
evidence.add_reference({'database':'UniProtKB', 'id':uniprot})
output.append(evidence)
if len(output)>1:
return output
elif len(output)==1:
return output[0]
else:
return None
def get_organism_scientific_name(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
output = None
accession = item['uniprot']['entry']['accession'][0]
if 'organism' in item['uniprot']['entry']:
organism = item['uniprot']['entry']['organism']
if type(organism['name'])==list:
for name in organism['name']:
if name['@type']=='scientific':
evidence = Evidence()
evidence.value = name['#text']
dbref_type = organism['dbReference']['@type']
dbref_id = organism['dbReference']['@id']
if dbref_type=='NCBI Taxonomy':
evidence.add_reference({'database':'NCBI_Taxonomy', 'id':dbref_id})
else:
raise ValueError('Unknown reference in database')
evidence.add_reference({'database':'UniProtKB', 'id':accession})
output = evidence
return output
def get_uniprot_id(item, entity='all'):
evidence = Evidence()
if type(item['uniprot']['entry']['accession']) is list:
accession = item['uniprot']['entry']['accession'][0]
else:
accession = item['uniprot']['entry']['accession']
evidence.value = accession
evidence.add_reference({'database':'UniProtKB', 'id':accession})
return evidence
def get_canonical_sequence(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
seq = item['uniprot']['entry']['sequence']['#text']
accession = item['uniprot']['entry']['accession'][0]
evidence = Evidence()
evidence.value = seq
evidence.add_reference({'database': 'UniProtKB', 'id': accession})
return evidence
def get_uniprot_entry_name(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
evidence = Evidence()
evidence.value = item['uniprot']['entry']['name']
accession = item['uniprot']['entry']['accession'][0]
evidence.add_reference({'database': 'UniProtKB', 'id': accession})
return evidence
def get_host(item, entity='all'):
if 'organismHost' in item['uniprot']['entry']:
evidence = Evidence()
host = item['uniprot']['entry']['organismHost']['name']['#text']
accession = item['uniprot']['entry']['accession'][0]
ncbi_taxonomy = item['uniprot']['entry']['organismHost'][
'dbReference']['@id']
evidence.value = host
evidence.add_NCBI_Taxonomy(id=ncbi_taxonomy)
evidence.add_UniProtKB(id=accession)
return evidence
else:
return None
def get_pdbs(item, entity='all'):
output = []
uniprot = item['uniprot']['entry']['accession'][0]
dbReference = item['uniprot']['entry']['dbReference']
for db in dbReference:
if db['@type'] == 'PDB':
accession = db['@id']
evidence = Evidence()
evidence.value = accession
evidence.add_reference({'database': 'PDB', 'id': accession})
evidence.add_reference({'database': 'UniProtKB', 'id': uniprot})
output.append(evidence)
return output