def get_subunit_structure(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
output = []
accession = item['uniprot']['entry']['accession'][0]
### Info in comment
for comment in item['uniprot']['entry']['comment']:
if comment['@type']=='subunit':
if type(comment['text'])==list:
for comment_subunit in comment['text']:
subunit = evi.Evidence()
subunit.value = comment_subunit['#text']
if '@evidence' in comment_subunit:
evidence_numbers_in_db = comment_subunit['@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(subunit, evidence_in_db)
subunit.add_reference({'database':'UniProtKB', 'id':accession})
output.append(subunit)
else:
subunit = evi.Evidence()
subunit.value = comment['text']['#text']
if '@evidence' in comment['text']:
evidence_numbers_in_db = comment['text']['@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(subunit, evidence_in_db)
subunit.add_reference({'database':'UniProtKB', 'id':accession})
output.append(subunit)
if len(output)==0:
output = None
elif len(output)==1:
output = output[0]
return output
def get_activity_regulation(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
from .get_uniprot import get_uniprot
output = []
uniprot = get_uniprot(item, entity=entity)
ref_uniprot = uniprot.references[0]
### Info in comment
for comment in item['uniprot']['entry']['comment']:
if comment['@type'] == 'activity regulation':
if type(comment) != OrderedDict:
raise ValueError(
"Comment type not recognized for activity regulation")
comment_text = comment['text']
if type(comment_text) in [OrderedDict, str]:
comment_text = [comment_text]
if type(comment_text) != list:
raise ValueError(
"Text in comment not recognized for activity regulation")
for aux in comment_text:
evidence = evi.Evidence()
if type(aux) == OrderedDict:
evidence.value = aux['#text']
if '@evidence' in aux:
evidence_numbers_in_db = aux['@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)
elif type(aux) == str:
evidence.value = aux
evidence.add_reference(ref_uniprot)
output.append(evidence)
return output
def get_function(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
from ._get_reference_from_dbevidence import _get_reference_from_dbevidence
from .get_uniprot_id import get_uniprot_id
output = {'function':[], 'references':[], 'notes':[]}
uniprot = get_uniprot_id(item, entity=entity)
ref_uniprot = uniprot.references[0]
### Info in comment
for comment in item['uniprot']['entry']['comment']:
if comment['@type']=='function':
if type(comment)!=OrderedDict:
raise ValueError("Comment type not recognized for function")
comment_text = comment['text']
if type(comment_text) in [OrderedDict, str]:
comment_text = [comment_text]
if type(comment_text)!=list:
raise ValueError("Text in comment not recognized for function")
for aux in comment_text:
evidence = evi.Evidence()
if type(aux)==OrderedDict:
evidence.value = aux['#text']
if '@evidence' in aux:
evidence_numbers_in_db = aux['@evidence'].split(' ')
for evidence_number_in_db in evidence_numbers_in_db:
ref = _get_reference_from_dbevidence(int(evidence_number_in_db), item)
evidence.add_reference(ref)
output['references'].append(ref)
elif type(aux)==str:
evidence.value = aux
evidence.add_reference(ref_uniprot)
output['function'].append(evidence)
output['references'].append(ref_uniprot)
return output
def get_ec(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
from .get_dbreference import get_dbreference
in_name = False
in_db = False
if 'ecNumber' in item['uniprot']['entry']['protein']['recommendedName']:
ecNumber = item['uniprot']['entry']['protein']['recommendedName'][
'ecNumber']
evidence = evi.Evidence()
if '#text' in ecNumber:
evidence.value = ecNumber['#text']
if '@evidence' in ecNumber:
evidence_numbers_in_db = ecNumber['@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})
evidence_in_name = evidence
in_name = True
evidence = get_dbreference(item, dbname='EC')
if evidence is not None:
evidence_in_db = evidence
in_db = True
if (in_db == False) and (in_name == False):
return None
elif (in_db == True) and (in_name == True):
return evi.join([evidence_in_name, evidence_in_db])
else:
raise ValueError("EC is not in name and neither in dbreference")
def get_metabolic_pathways(item, entity='all'):
from ._add_reference_to_evidence import _add_reference_to_evidence
from .get_uniprot import get_uniprot
output = []
uniprot = get_uniprot(item, entity=entity)
ref_uniprot = uniprot.references[0]
### Info in comment
for comment in item['uniprot']['entry']['comment']:
if comment['@type'] == 'pathway':
if type(comment) != OrderedDict:
raise ValueError(
"Comment type not recognized for methabolic pathway")
evidence = evi.Evidence()
evidence.value = comment['text']
if '@evidence' in comment:
evidence_numbers_in_db = comment['@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)
return output
def get_organism(item, entity='all', as_card=False):
from sabueso.cards import OrganismCard, organism_dict
from ._add_reference_to_evidence import _add_reference_to_evidence
output = deepcopy(organism_dict)
accession = item['uniprot']['entry']['accession'][0]
if 'organism' in item['uniprot']['entry']:
common_name = evi.Evidence()
scientific_name = evi.Evidence()
ncbi_taxonomy = evi.Evidence()
organism = item['uniprot']['entry']['organism']
dbref_type = organism['dbReference']['@type']
dbref_id = organism['dbReference']['@id']
if dbref_type != 'NCBI Taxonomy':
raise ValueError('Unknown reference in database')
ncbi_taxonomy.value = dbref_id
ncbi_taxonomy.add_reference({
'database': 'NCBI_Taxonomy',
'id': dbref_id
})
ncbi_taxonomy.add_reference({'database': 'UniProtKB', 'id': accession})
output['ncbi_taxonomy'] = ncbi_taxonomy
if type(organism['name']) == list:
for name in organism['name']:
if name['@type'] == 'common':
common_name.value = name['#text']
common_name.add_reference({
'database': 'NCBI_Taxonomy',
'id': dbref_id
})
common_name.add_reference({
'database': 'UniProtKB',
'id': accession
})
output['common_name'] = common_name
if name['@type'] == 'scientific':
scientific_name.value = name['#text']
scientific_name.add_reference({
'database': 'NCBI_Taxonomy',
'id': dbref_id
})
scientific_name.add_reference({
'database': 'UniProtKB',
'id': accession
})
output['scientific_name'] = scientific_name
else:
common_name.value = organism['name']['#text']
common_name.add_reference({
'database': 'NCBI_Taxonomy',
'id': dbref_id
})
common_name.add_reference({
'database': 'UniProtKB',
'id': accession
})
output['common_name'] = common_name
output['references'].append(
evi.reference({
'database': 'NCBI_Taxonomy',
'id': dbref_id
}))
output['references'].append(
evi.reference({
'database': 'UniProtKB',
'id': accession
}))
if as_card:
output = OrganismCard(output)
return output
def get_catalytic_activity(item, entity='all', as_cards=False):
from sabueso.cards import CatalyticActivityCard, catalytic_activity_dict
from ._add_reference_to_evidence import _add_reference_to_evidence
from ._get_reference_from_dbevidence import _get_reference_from_dbevidence
from ._get_reference_from_dbreference import _get_reference_from_dbreference
from .get_uniprot import get_uniprot
output = []
uniprot = get_uniprot(item, entity=entity)
ref_uniprot = uniprot.references[0]
### Info in comment
for comment in item['uniprot']['entry']['comment']:
if comment['@type'] == 'catalytic activity':
if type(comment) != OrderedDict:
raise ValueError(
"Comment type not recognized for catalytic activity")
aux_dict = deepcopy(catalytic_activity_dict)
aux_dict['reaction'] = evi.Evidence()
aux_dict['physiological_direction'] = evi.Evidence()
aux_dict['reaction'].value = comment['reaction']['text']
if '@evidence' in comment['reaction']:
evidence_numbers = comment['reaction']['@evidence'].split(' ')
for ii in evidence_numbers:
ref = _get_reference_from_dbevidence(int(ii), item)
if ref is not None:
aux_dict['references'].append(ref)
aux_dict['reaction'].add_reference(ref)
if 'dbReference' in comment['reaction']:
dbreference = comment['physiologicalReaction']['dbReference']
if type(dbreference) == OrderedDict:
dbreference = [dbreference]
for aux in dbreference:
ref = _get_reference_from_dbreference(
aux['@type'], aux['@id'])
aux_dict['references'].append(ref)
aux_dict['reaction'].add_reference(ref)
aux_dict['physiological_direction'].value = comment[
'physiologicalReaction']['@direction']
if '@evidence' in comment['physiologicalReaction']:
evidence_numbers = comment['physiologicalReaction'][
'@evidence'].split(' ')
for ii in evidence_numbers:
ref = _get_reference_from_dbevidence(int(ii), item)
if ref is not None:
aux_dict['references'].append(ref)
aux_dict['physiological_direction'].add_reference(ref)
if 'dbReference' in comment['physiologicalReaction']:
dbreference = comment['physiologicalReaction']['dbReference']
if type(dbreference) == OrderedDict:
dbreference = [dbreference]
for aux in dbreference:
ref = _get_reference_from_dbreference(
aux['@type'], aux['@id'])
aux_dict['references'].append(ref)
aux_dict['physiological_direction'].add_reference(ref)
aux_dict['references'].append(ref_uniprot)
output.append(aux_dict)
if as_cards:
output = [CatalyticActivityCard(ii) for ii in output]
return output
def get_location_in_cell(item, entity='all', as_cards=False):
from sabueso.cards import LocationInCellCard, location_in_cell_dict
from ._add_reference_to_evidence import _add_reference_to_evidence
from ._get_reference_from_dbevidence import _get_reference_from_dbevidence
from .get_uniprot import get_uniprot
uniprot = get_uniprot(item, entity=entity)
ref_uniprot = uniprot.references[0]
output = []
for comment in item['uniprot']['entry']['comment']:
if comment['@type'] == 'subcellular location':
aux_dict = deepcopy(location_in_cell_dict)
for subcellularLocation in comment['subcellularLocation']:
aux_output = []
if 'location' in subcellularLocation:
location = subcellularLocation['location']
if type(location) == OrderedDict:
location = [location]
for ll in location:
evidence = evi.Evidence()
evidence.value = ll['#text']
if '@evidence' in ll:
evidence_numbers = ll['@evidence'].split(' ')
for ii in evidence_numbers:
ref = _get_reference_from_dbevidence(
int(ii), item)
if ref is not None:
evidence.add_reference(ref)
aux_dict['references'].append(ref)
evidence.add_reference(ref_uniprot)
aux_output.append(evidence)
if 'topology' in subcellularLocation:
topology = subcellularLocation['topology']
if type(topology) == OrderedDict:
topology = [topology]
for ll in topology:
evidence = evi.Evidence()
evidence.value = ll['#text']
if '@evidence' in ll:
evidence_numbers = ll['@evidence'].split(' ')
for ii in evidence_numbers:
ref = _get_reference_from_dbevidence(
int(ii), item)
if ref is not None:
evidence.add_reference(ref)
aux_dict['references'].append(ref)
evidence.add_reference(ref_uniprot)
aux_output.append(evidence)
aux_dict['location'].append(aux_output)
evidence = evi.Evidence()
evidence.value = comment['text']['#text']
if '@evidence' in comment['text']:
evidence_numbers = comment['text']['@evidence'].split(' ')
for ii in evidence_numbers:
ref = _get_reference_from_dbevidence(int(ii), item)
if ref is not None:
evidence.add_reference(ref)
aux_dict['references'].append(ref)
evidence.add_reference(ref_uniprot)
aux_dict['note'] = evidence
output.append(aux_dict)
if as_cards:
output = [LocationInCellCard(ii) for ii in output]
return output