def map_to_up_primary(ids):
primary_ids = []
for up_id in ids:
if not uniprot_client.is_secondary(up_id):
primary_ids.append(up_id)
continue
primary_id = uniprot_client.get_primary_id(up_id)
primary_ids.append(primary_id)
# If there are no primary IDs, we return None
if not primary_ids:
return None
# Try to get primary IDs if there are
# If there is more than one primary ID then we return the first one
elif len(primary_ids) > 1:
human_upids = [
id for id in primary_ids if uniprot_client.is_human(id)
]
if not human_upids:
logger.info('More than one primary id but none human, '
'choosing the first: %s' %
','.join(primary_ids))
primary_id = primary_ids[0]
elif len(human_upids) > 1:
logger.info('More than one human primary id, choosing '
'the first: %s' % ','.join(human_upids))
primary_id = human_upids[0]
# Only one, so use it
else:
primary_id = human_upids[0]
# One primary ID, so use it
else:
primary_id = primary_ids[0]
# Make sure it's unicode
return str(primary_id)
def map_to_up_primary(ids):
primary_ids = []
for up_id in ids:
if not uniprot_client.is_secondary(up_id):
primary_ids.append(up_id)
continue
primary_id = uniprot_client.get_primary_id(up_id)
primary_ids.append(primary_id)
# If there are no primary IDs, we return None
if not primary_ids:
return None
# Try to get primary IDs if there are
# If there is more than one primary ID then we return the first one
if len(primary_ids) > 1:
logger.info('More than one primary id: %s' %
','.join(primary_ids))
# Make sure it's unicode
primary_id = str(primary_ids[0])
return primary_id
def sanitize_up_ids(up_ids):
# First, we map any secondary IDs to primary IDs
up_ids = {uniprot_client.get_primary_id(up_id) for up_id in up_ids}
# We filter out IDs that are actually mnemonics, these are just mixed
# in without any differentiation from other IDs
up_ids = {up_id for up_id in up_ids if '_' not in up_id}
# TODO: should we do anything about isoforms?
# We separate out specific sets of IDs
human_ids = [up_id for up_id in up_ids if uniprot_client.is_human(up_id)]
reviewed_non_human_ids = [
up_id for up_id in up_ids if not uniprot_client.is_human(up_id)
# get_mnemonic is just a quick way to see if we have this entry
and uniprot_client.get_mnemonic(up_id, web_fallback=False)
]
if human_ids:
return human_ids
elif reviewed_non_human_ids:
return reviewed_non_human_ids
else:
return []
def test_get_primary_id_secondary_nohuman():
assert uniprot_client.get_primary_id('P31848') in \
['P0A5M5', 'P9WIU6', 'P9WIU7']
def test_get_primary_id_secondary_hashuman():
assert uniprot_client.get_primary_id('Q96J62') == 'P61978'
def test_get_primary_id_primary():
assert uniprot_client.get_primary_id('Q02750') == 'Q02750'
def test_get_primary_id_secondary_nohuman():
assert uniprot_client.get_primary_id('P31848') in \
['P0A5M5', 'P9WIU6', 'P9WIU7']
def test_get_primary_id_secondary_hashuman():
assert uniprot_client.get_primary_id('Q96J62') == 'P61978'
def test_get_primary_id_primary():
assert uniprot_client.get_primary_id('Q02750') == 'Q02750'