def test_parse_hpo_lines():
## GIVEN some lines that correspends to a hpo file
hpo_lines = [
"#Format: HPO-ID<tab>HPO-Name<tab>Gene-ID<tab>Gene-Name",
"HP:0000878\t11 pairs of ribs\t126792\tB3GALT6",
"HP:0000878\t11 pairs of ribs\t5932\tRBBP8"
]
## WHEN parsing the lines
hpo_dict = parse_hpo_phenotypes(hpo_lines)
## THEN assert that the correct information was parsed
assert len(hpo_dict) == 1
assert "HP:0000878" in hpo_dict
assert set(hpo_dict['HP:0000878']['hgnc_symbols']) == set(['B3GALT6', 'RBBP8'])
def test_parse_hpo_lines():
## GIVEN some lines that correspends to a hpo file
hpo_lines = [
"#Format: HPO-ID<tab>HPO-Name<tab>Gene-ID<tab>Gene-Name",
"HP:0000878\t11 pairs of ribs\t126792\tB3GALT6",
"HP:0000878\t11 pairs of ribs\t5932\tRBBP8"
]
## WHEN parsing the lines
hpo_dict = parse_hpo_phenotypes(hpo_lines)
## THEN assert that the correct information was parsed
assert len(hpo_dict) == 1
assert "HP:0000878" in hpo_dict
assert set(hpo_dict['HP:0000878']['hgnc_symbols']) == set(
['B3GALT6', 'RBBP8'])
def load_hpo_terms(adapter, hpo_lines, genes):
"""Load the hpo terms into the database
Parse the hpo lines, build the objects and add them to the database
Args:
adapter(MongoAdapter)
hpo_lines(iterable(str))
"""
hpo_terms = parse_hpo_phenotypes(hpo_lines)
start_time = datetime.now()
logger.info("Loading the hpo terms...")
for nr_terms, hpo_id in enumerate(hpo_terms):
hpo_info = hpo_terms[hpo_id]
hpo_obj = build_hpo_term(hpo_info, genes)
adapter.load_hpo_term(hpo_obj)
logger.info("Loading done. Nr of terms loaded {0}".format(nr_terms))
logger.info("Time to load terms: {0}".format(datetime.now() - start_time))
def hpo_terms(request, hpo_terms_file):
"""Get a dictionary with the hpo terms"""
print('')
hpo_terms_handle = get_file_handle(hpo_terms_file)
return parse_hpo_phenotypes(hpo_terms_handle)
def test_parse_hpo_terms(hpo_terms_handle):
hpo_terms = parse_hpo_phenotypes(hpo_terms_handle)
for hpo_id in hpo_terms:
assert hpo_terms[hpo_id]['hpo_id'] == hpo_id
