def test_build_hpo_term_missing_key(adapter, test_hpo_info, key):
## GIVEN a dictionary with hpo information
## WHEN deleteing key
test_hpo_info.pop(key)
## THEN calling build_hpo_term() will raise KeyError
with pytest.raises(KeyError):
build_hpo_term(test_hpo_info)
def test_build_hpo_term_with_genes(adapter):
## GIVEN a hpo term and a adapter with genes
hpo_info = {
'hpo_id':"HP:0000878",
'description': "11 pairs of ribs",
'hgnc_symbols': ['B3GALT6', 'RBBP8']
}
alias_genes = {}
alias_genes['B3GALT6'] = {
'true': 17978,
'ids': [17978],
}
alias_genes['RBBP8'] = {
'true': 9891,
'ids': [9891],
}
## WHEN building the hpo term
hpo_obj = build_hpo_term(hpo_info, alias_genes)
## THEN assert that the term has the correct information
assert hpo_obj['_id'] == hpo_obj['hpo_id'] == hpo_info['hpo_id']
## The adapter has no genes loaded so we expect this to be 0
assert len(hpo_obj['genes']) == 2
assert set(hpo_obj['genes']) == set([17978, 9891])
def test_build_hpo_term(adapter, test_hpo_info):
## GIVEN a hpo term
## WHEN building the hpo term
hpo_obj = build_hpo_term(test_hpo_info)
## THEN assert that the term has the correct information
assert hpo_obj["_id"] == hpo_obj["hpo_id"] == test_hpo_info["hpo_id"]
assert hpo_obj["description"] == test_hpo_info["description"]
assert len(hpo_obj["genes"]) == 2
def test_build_hpo_term(adapter):
## GIVEN a hpo term
hpo_info = {
"hpo_id": "HP:0000878",
"description": "11 pairs of ribs",
"genes": [1, 2],
}
## WHEN building the hpo term
hpo_obj = build_hpo_term(hpo_info)
## THEN assert that the term has the correct information
assert hpo_obj["_id"] == hpo_obj["hpo_id"] == hpo_info["hpo_id"]
assert hpo_obj["description"] == hpo_info["description"]
assert len(hpo_obj["genes"]) == 2
def test_build_hpo_term_non_existing_genes(adapter):
## GIVEN a hpo term
hpo_info = {
'hpo_id': "HP:0000878",
'description': "11 pairs of ribs",
'genes': [1, 2]
}
## WHEN building the hpo term
hpo_obj = build_hpo_term(hpo_info)
## THEN assert that the term has the correct information
assert hpo_obj['_id'] == hpo_obj['hpo_id'] == hpo_info['hpo_id']
assert hpo_obj['description'] == hpo_info['description']
assert len(hpo_obj['genes']) == 2
def test_build_hpo_term(adapter):
## GIVEN a hpo term
hpo_info = {
'hpo_id':"HP:0000878",
'description': "11 pairs of ribs",
'genes': [1, 2]
}
## WHEN building the hpo term
hpo_obj = build_hpo_term(hpo_info)
## THEN assert that the term has the correct information
assert hpo_obj['_id'] == hpo_obj['hpo_id'] == hpo_info['hpo_id']
assert hpo_obj['description'] == hpo_info['description']
assert len(hpo_obj['genes']) == 2
def test_build_hpo_term_non_existing_genes(adapter):
## GIVEN a hpo term
hpo_info = {
'hpo_id':"HP:0000878",
'description': "11 pairs of ribs",
'hgnc_symbols': ['B3GALT6', 'RBBP8']
}
## WHEN building the hpo term
hpo_obj = build_hpo_term(hpo_info, {})
## THEN assert that the term has the correct information
assert hpo_obj['_id'] == hpo_obj['hpo_id'] == hpo_info['hpo_id']
assert hpo_obj['description'] == hpo_info['description']
## The adapter has no genes loaded so we expect this to be 0
assert len(hpo_obj['genes']) == 0
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 load_hpo_terms(adapter,
hpo_lines=None,
hpo_gene_lines=None,
alias_genes=None):
"""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_gene_lines(iterable(str))
"""
# Store the hpo terms
hpo_terms = {}
# Fetch the hpo terms if no file
if not hpo_lines:
hpo_lines = fetch_hpo_terms()
# Fetch the hpo gene information if no file
if not hpo_gene_lines:
hpo_gene_lines = fetch_hpo_to_genes()
# Parse the terms
# This will yield dictionaries with information about the terms
LOG.info("Parsing hpo terms")
for term in parse_hpo_obo(hpo_lines):
hpo_terms[term['hpo_id']] = term
# Get a map with hgnc symbols to hgnc ids from scout
if not alias_genes:
alias_genes = adapter.genes_by_alias()
LOG.info("Adding gene information to hpo terms ...")
for hpo_to_symbol in parse_hpo_to_genes(hpo_gene_lines):
hgnc_symbol = hpo_to_symbol['hgnc_symbol']
hpo_id = hpo_to_symbol['hpo_id']
# Fetch gene info to get correct hgnc id
gene_info = alias_genes.get(hgnc_symbol)
if not gene_info:
continue
hgnc_id = gene_info['true']
if hpo_id not in hpo_terms:
continue
hpo_term = hpo_terms[hpo_id]
if not 'genes' in hpo_term:
hpo_term['genes'] = set()
hpo_term['genes'].add(hgnc_id)
start_time = datetime.now()
LOG.info("Loading the hpo terms...")
nr_terms = len(hpo_terms)
hpo_bulk = []
with progressbar(hpo_terms.values(),
label="Loading hpo terms",
length=nr_terms) as bar:
for hpo_info in bar:
hpo_bulk.append(build_hpo_term(hpo_info))
if len(hpo_bulk) > 10000:
adapter.load_hpo_bulk(hpo_bulk)
hpo_bulk = []
if hpo_bulk:
adapter.load_hpo_bulk(hpo_bulk)
LOG.info("Loading done. Nr of terms loaded {0}".format(nr_terms))
LOG.info("Time to load terms: {0}".format(datetime.now() - start_time))
def load_hpo_terms(adapter,
hpo_lines=None,
hpo_gene_lines=None,
alias_genes=None):
"""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)): lines from file http://purl.obolibrary.org/obo/hp.obo
hpo_gene_lines(iterable(str)): lines from file
https://ci.monarchinitiative.org/view/hpo/job/hpo.annotations/lastSuccessfulBuild/artifact/rare-diseases/util/annotation/phenotype_to_genes.txt
alias_genes
"""
# Fetch the hpo terms if no file
if not hpo_lines:
hpo_lines = fetch_hpo_terms()
# Parse the terms
LOG.info("Parsing hpo terms")
hpo_terms = build_hpo_tree(hpo_lines)
# Fetch the hpo gene information if no file
if not hpo_gene_lines:
hpo_gene_lines = fetch_hpo_to_genes_to_disease()
# Get a map with hgnc symbols to hgnc ids from scout
if not alias_genes:
alias_genes = adapter.genes_by_alias()
LOG.info("Adding gene information to hpo terms ...")
for hpo_to_symbol in parse_hpo_to_genes(hpo_gene_lines):
hgnc_symbol = hpo_to_symbol["hgnc_symbol"]
hpo_id = hpo_to_symbol["hpo_id"]
# Fetch gene info to get correct hgnc id
gene_info = alias_genes.get(hgnc_symbol)
if not gene_info:
continue
hgnc_id = gene_info["true"]
if hpo_id not in hpo_terms:
continue
hpo_term = hpo_terms[hpo_id]
if not "genes" in hpo_term:
hpo_term["genes"] = set()
hpo_term["genes"].add(hgnc_id)
start_time = datetime.now()
LOG.info("Loading the hpo terms...")
nr_terms = len(hpo_terms)
hpo_bulk = []
with progressbar(hpo_terms.values(),
label="Loading hpo terms",
length=nr_terms) as bar:
for hpo_info in bar:
hpo_bulk.append(build_hpo_term(hpo_info))
if len(hpo_bulk) > 10000:
adapter.load_hpo_bulk(hpo_bulk)
hpo_bulk = []
if hpo_bulk:
adapter.load_hpo_bulk(hpo_bulk)
LOG.info("Loading done. Nr of terms loaded {0}".format(nr_terms))
LOG.info("Time to load terms: {0}".format(datetime.now() - start_time))
def load_hpo_terms(adapter, hpo_lines=None, hpo_gene_lines=None, alias_genes=None):
"""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_gene_lines(iterable(str))
"""
# Store the hpo terms
hpo_terms = {}
# Fetch the hpo terms if no file
if not hpo_lines:
hpo_lines = fetch_hpo_terms()
# Fetch the hpo gene information if no file
if not hpo_gene_lines:
hpo_gene_lines = fetch_hpo_to_genes()
# Parse the terms
# This will yield dictionaries with information about the terms
LOG.info("Parsing hpo terms")
for term in parse_hpo_obo(hpo_lines):
hpo_terms[term['hpo_id']] = term
# Get a map with hgnc symbols to hgnc ids from scout
if not alias_genes:
alias_genes = adapter.genes_by_alias()
LOG.info("Adding gene information to hpo terms ...")
for hpo_to_symbol in parse_hpo_to_genes(hpo_gene_lines):
hgnc_symbol = hpo_to_symbol['hgnc_symbol']
hpo_id = hpo_to_symbol['hpo_id']
# Fetch gene info to get correct hgnc id
gene_info = alias_genes.get(hgnc_symbol)
if not gene_info:
continue
hgnc_id = gene_info['true']
if hpo_id not in hpo_terms:
continue
hpo_term = hpo_terms[hpo_id]
if not 'genes' in hpo_term:
hpo_term['genes'] = set()
hpo_term['genes'].add(hgnc_id)
start_time = datetime.now()
LOG.info("Loading the hpo terms...")
nr_terms = len(hpo_terms)
hpo_bulk = []
with progressbar(hpo_terms.values(), label="Loading hpo terms", length=nr_terms) as bar:
for hpo_info in bar:
hpo_bulk.append(build_hpo_term(hpo_info))
if len(hpo_bulk) > 10000:
adapter.load_hpo_bulk(hpo_bulk)
hpo_bulk = []
if hpo_bulk:
adapter.load_hpo_bulk(hpo_bulk)
LOG.info("Loading done. Nr of terms loaded {0}".format(nr_terms))
LOG.info("Time to load terms: {0}".format(datetime.now() - start_time))
