def add_uniprot_data(self, enzyme_ids, source):
'''Gets Uniprot data.'''
fields = ['entry name', 'protein names', 'organism-id', 'ec']
enzyme_ids = [enzyme_id for enzyme_id in enzyme_ids
if enzyme_id not in self.__nodes]
uniprot_values = seq_utils.get_uniprot_values(enzyme_ids, fields,
batch_size=512,
verbose=True)
for uniprot_id, uniprot_value in uniprot_values.iteritems():
enzyme_node = {':LABEL': 'Enzyme',
'uniprot:ID(Enzyme)': uniprot_id}
self.__nodes[uniprot_id] = enzyme_node
organism_id = uniprot_value.pop('Organism ID')
if 'Entry name' in uniprot_value:
enzyme_node['entry'] = uniprot_value['Entry name']
if 'Protein names' in uniprot_value:
enzyme_node['names'] = uniprot_value['Protein names']
if len(enzyme_node['names']) > 0:
enzyme_node['name'] = enzyme_node['names'][0]
if 'EC number' in uniprot_value:
enzyme_node['ec-code'] = uniprot_value['EC number']
if organism_id is not None and len(organism_id) > 0:
self.__org_enz_rels.append([organism_id, 'expresses',
uniprot_id, {'source': source}])
def test_get_uniprot_values(self):
"""Tests get_uniprot_values method."""
result = seq_utils.get_uniprot_values(["P19367", "P42212"], ["organism-id", "protein names"], 1)
expected = {
"P19367": {
"Organism ID": "9606",
"Protein names": ["Hexokinase-1", "EC 2.7.1.1", "Brain form hexokinase", "Hexokinase type I", "HK I"],
},
"P42212": {"Organism ID": "6100", "Protein names": ["Green fluorescent protein"]},
}
self.assertEquals(result, expected)
def test_get_uniprot_values(self):
'''Tests get_uniprot_values method.'''
result = seq_utils.get_uniprot_values(['P19367', 'P42212'],
['organism-id',
'protein names'], 1)
expected = {'P19367': {'Entry': 'P19367',
'Organism ID': '9606',
'Protein names': ['Hexokinase-1',
'EC 2.7.1.1',
'Brain form hexokinase',
'Hexokinase type I',
'HK I',
'Hexokinase-A']},
'P42212': {'Entry': 'P42212',
'Organism ID': '6100',
'Protein names': ['Green fluorescent protein']}}
self.assertEqual(result, expected)
def _get_metadata(prot_id, tir, cai, target_org=None, uniprot_id=None):
'''Gets metadata.'''
name = prot_id
description = prot_id
links = []
parameters = []
if uniprot_id is not None:
uniprot_vals = seq_utils.get_uniprot_values([uniprot_id],
['entry name',
'protein names',
'organism-id',
'organism',
'ec'])
# Add metadata:
if len(uniprot_vals.keys()):
prot_id = uniprot_vals.keys()[0]
name = uniprot_vals[prot_id]['Entry name']
organism = uniprot_vals[prot_id]['Organism']
prot_names = uniprot_vals[prot_id]['Protein names']
description = ', '.join(prot_names) + ' (' + organism + ')'
ec_number = uniprot_vals[prot_id].get('EC number', None)
parameters.append({'name': 'Organism', 'value': organism})
links.append('http://identifiers.org/uniprot/' + uniprot_id)
if ec_number:
links.append('http://identifiers.org/ec-code/' + ec_number)
parameters.append({'name': 'Type', 'value': 'PART'})
parameters.append({'name': 'TIR', 'value': float("{0:.2f}".format(tir))})
parameters.append({'name': 'CAI', 'value': float("{0:.2f}".format(cai))})
if target_org:
links.append('http://identifiers.org/taxonomy/' + target_org)
metadata = {'name': name,
'shortDescription': description,
'links': links,
'parameters': parameters}
return metadata
def get_fasta(uniprot_id, filename, variants, digest=None):
'''Gets a FASTA file representing variants of uniprot id.'''
uniprot = seq_utils.get_uniprot_values([uniprot_id], ['sequence'])
get_fasta_from_seq(uniprot[uniprot_id]['Sequence'], filename, variants,
digest, uniprot_id)
def _get_feature(comp_def):
'''Get feature.'''
if dna_utils.SO_ASS_COMP in comp_def.roles:
# Assembly component:
return {
'typ': dna_utils.SO_ASS_COMP,
'name': comp_def.identity,
'seq': '',
'parameters': {
'Tm target': 70
},
'temp_params': {
'fixed': True,
'required': ['name', 'tm'],
'valid': True,
'id': comp_def.displayId
}
}
if SO_RBS in comp_def.roles:
# RBS:
return {
'typ': dna_utils.SO_RBS,
'name': comp_def.identity,
'end': 60,
'parameters': {
'TIR target': float(comp_def.displayId.split('_')[1])
},
'temp_params': {
'fixed': False,
'required': ['name', 'tir'],
'min_end': 35,
'max_end': 10000,
'valid': True,
'id': comp_def.displayId
}
}
if SO_CDS in comp_def.roles:
# CDS:
uniprot_id = comp_def.displayId.split('_')[0]
uniprot_vals = get_uniprot_values([uniprot_id], ['sequence'])
if uniprot_id not in uniprot_vals:
raise ValueError('Uniprot id not found: %s' % uniprot_id)
return {
'typ': dna_utils.SO_CDS,
'name': comp_def.identity,
'temp_params': {
'fixed': False,
'required': ['name', 'prot'],
'valid': True,
'id': comp_def.displayId,
'aa_seq': uniprot_vals[uniprot_id]['Sequence'],
'orig_seq': uniprot_vals[uniprot_id]['Sequence']
},
'desc': '',
'links': ['http://identifiers.org/uniprot/%s' % uniprot_id]
}
raise ValueError('Invalid roles in component definition: %s' % comp_def)
