def test_query_protein_deprecated():
g = uniprot_client.query_protein('Q8NHX1')
assert g is not None
gene_name = uniprot_client.get_gene_name('Q8NHX1')
assert gene_name == 'MAPK3'
assert unicode_strs(gene_name)
gene_name = uniprot_client.get_gene_name('Q8NHX1', web_fallback=False)
assert gene_name == 'MAPK3'
assert unicode_strs(gene_name)
def test_query_protein_deprecated():
g = uniprot_client.query_protein('Q8NHX1')
assert g is not None
gene_name = uniprot_client.get_gene_name('Q8NHX1')
assert gene_name == 'MAPK3'
assert unicode_strs(gene_name)
gene_name = uniprot_client.get_gene_name('Q8NHX1', web_fallback=False)
assert gene_name == 'MAPK3'
assert unicode_strs(gene_name)
def _get_name_by_id(self, entity_id):
entity_term = self.tree.find("TERM/[@id='%s']" % entity_id)
name = entity_term.find("name")
if name is None:
warnings.warn('Entity without a name')
return ''
try:
dbid = entity_term.attrib["dbid"]
except:
warnings.warn('No grounding information for %s' % name.text)
return self._get_valid_component_name(name.text)
dbids = dbid.split('|')
hgnc_ids = [i for i in dbids if i.startswith('HGNC')]
up_ids = [i for i in dbids if i.startswith('UP')]
#TODO: handle protein families like 14-3-3 with IDs like
# XFAM:PF00244.15, FA:00007
if hgnc_ids:
if len(hgnc_ids) > 1:
warnings.warn('%d HGNC IDs reported.' % len(hgnc_ids))
hgnc_id = re.match(r'HGNC\:([0-9]*)', hgnc_ids[0]).groups()[0]
hgnc_name = self._get_hgnc_name(hgnc_id)
return self._get_valid_component_name(hgnc_name)
elif up_ids:
if len(hgnc_ids) > 1:
warnings.warn('%d UniProt IDs reported.' % len(up_ids))
up_id = re.match(r'UP\:([A-Z0-9]*)', up_ids[0]).groups()[0]
up_rdf = up_client.query_protein(up_id)
# First try to get HGNC name
hgnc_name = up_client.get_hgnc_name(up_rdf)
if hgnc_name is not None:
return self._get_valid_component_name(hgnc_name)
# Next, try to get the gene name
gene_name = up_client.get_gene_name(up_rdf)
if gene_name is not None:
return self._get_valid_component_name(gene_name)
# By default, return the text of the name tag
name_txt = name.text.strip('|')
return self._get_valid_component_name(name_txt)
def _get_name_by_id(self, entity_id):
entity_term = self.tree.find("TERM/[@id='%s']" % entity_id)
name = entity_term.find("name")
if name is None:
warnings.warn('Entity without a name')
return ''
try:
dbid = entity_term.attrib["dbid"]
except:
warnings.warn('No grounding information for %s' % name.text)
return self._get_valid_component_name(name.text)
dbids = dbid.split('|')
hgnc_ids = [i for i in dbids if i.startswith('HGNC')]
up_ids = [i for i in dbids if i.startswith('UP')]
#TODO: handle protein families like 14-3-3 with IDs like
# XFAM:PF00244.15, FA:00007
if hgnc_ids:
if len(hgnc_ids) > 1:
warnings.warn('%d HGNC IDs reported.' % len(hgnc_ids))
hgnc_id = re.match(r'HGNC\:([0-9]*)', hgnc_ids[0]).groups()[0]
hgnc_name = self._get_hgnc_name(hgnc_id)
return self._get_valid_component_name(hgnc_name)
elif up_ids:
if len(hgnc_ids) > 1:
warnings.warn('%d UniProt IDs reported.' % len(up_ids))
up_id = re.match(r'UP\:([A-Z0-9]*)', up_ids[0]).groups()[0]
up_rdf = up_client.query_protein(up_id)
# First try to get HGNC name
hgnc_name = up_client.get_hgnc_name(up_rdf)
if hgnc_name is not None:
return self._get_valid_component_name(hgnc_name)
# Next, try to get the gene name
gene_name = up_client.get_gene_name(up_rdf)
if gene_name is not None:
return self._get_valid_component_name(gene_name)
# By default, return the text of the name tag
name_txt = name.text.strip('|')
return self._get_valid_component_name(name_txt)
def test_query_protein_exists():
g = uniprot_client.query_protein('P00533')
assert g is not None
def test_query_protein_nonexist():
g = uniprot_client.query_protein('XXXX')
assert g is None
def test_query_protein_exists():
g = uniprot_client.query_protein('P00533')
assert g is not None
def test_query_protein_nonexist():
g = uniprot_client.query_protein('XXXX')
assert g is None
def test_query_protein_deprecated():
g = uniprot_client.query_protein('Q8NHX1')
assert(g is not None)
assert(uniprot_client.get_hgnc_name('Q8NHX1') == 'MAPK3')
