def test_parse_diseases_lines():
## GIVEN a iterable of disease lines
disease_lines = [
"#Format: diseaseId<tab>gene-id<tab>gene-symbol",
"OMIM:300818 5277 PIGA",
"OMIM:300868 5277 PIGA",
"ORPHANET:447 5277 PIGA",
"OMIM:101400 2263 FGFR2",
"OMIM:101400 7291 TWIST1",
"OMIM:139500",
]
## WHEN parsing the diseases
diseases = parse_hpo_diseases(disease_lines)
## THEN assert that the diseases are parsed correct
assert diseases["OMIM:300818"]["source"] == "OMIM"
assert diseases["OMIM:300818"]["hgnc_symbols"] == set(["PIGA"])
assert diseases["ORPHANET:447"]["source"] == "ORPHANET"
assert diseases["ORPHANET:447"]["hgnc_symbols"] == set(["PIGA"])
assert diseases["OMIM:101400"]["source"] == "OMIM"
assert diseases["OMIM:101400"]["hgnc_symbols"] == set(["FGFR2", "TWIST1"])
assert diseases["OMIM:139500"]["source"] == "OMIM"
assert diseases["OMIM:139500"]["hgnc_symbols"] == set([])
def test_parse_diseases_lines():
## GIVEN a iterable of disease lines
disease_lines = [
"#Format: diseaseId<tab>gene-id<tab>gene-symbol",
"OMIM:300818 5277 PIGA",
"OMIM:300868 5277 PIGA",
"ORPHANET:447 5277 PIGA",
"OMIM:101400 2263 FGFR2",
"OMIM:101400 7291 TWIST1",
"OMIM:139500",
]
## WHEN parsing the diseases
diseases = parse_hpo_diseases(disease_lines)
## THEN assert that the diseases are parsed correct
assert diseases["OMIM:300818"]['source'] == 'OMIM'
assert diseases["OMIM:300818"]['hgnc_symbols'] == set(['PIGA'])
assert diseases["ORPHANET:447"]['source'] == "ORPHANET"
assert diseases["ORPHANET:447"]['hgnc_symbols'] == set(['PIGA'])
assert diseases["OMIM:101400"]['source'] == 'OMIM'
assert diseases["OMIM:101400"]['hgnc_symbols'] == set(['FGFR2','TWIST1'])
assert diseases["OMIM:139500"]['source'] == 'OMIM'
assert diseases["OMIM:139500"]['hgnc_symbols'] == set([])
def load_disease_terms(adapter, genemap_lines, genes, hpo_disease_lines):
"""Load the omim phenotypes into the database
Parse the phenotypes from genemap2.txt and find the associated hpo terms
from ALL_SOURCES_ALL_FREQUENCIES_diseases_to_genes_to_phenotypes.txt.
Args:
adapter(MongoAdapter)
genemap_lines(iterable(str))
genes(dict): Dictionary with all genes found in database
hpo_disease_lines(iterable(str))
"""
disease_terms = get_mim_phenotypes(genemap_lines=genemap_lines)
hpo_diseases = parse_hpo_diseases(hpo_disease_lines)
start_time = datetime.now()
logger.info("Loading the hpo disease...")
for nr_diseases, disease_number in enumerate(disease_terms):
disease_info = disease_terms[disease_number]
disease_id = "OMIM:{0}".format(disease_number)
if disease_id in hpo_diseases:
hpo_terms = hpo_diseases[disease_id]['hpo_terms']
if hpo_terms:
disease_info['hpo_terms'] = hpo_terms
disease_obj = build_disease_term(disease_info, genes)
adapter.load_disease_term(disease_obj)
logger.info("Loading done. Nr of diseases loaded {0}".format(nr_diseases))
logger.info("Time to load diseases: {0}".format(datetime.now() -
start_time))
def test_parse_diseases(hpo_disease_handle):
## GIVEN a iterable of disease lines
## WHEN parsing the diseases
diseases = parse_hpo_diseases(hpo_disease_handle)
## THEN assert that the diseases are parsed correct
for disease_id in diseases:
source = disease_id.split(":")[0]
disease_nr = int(disease_id.split(":")[1])
disease_term = diseases[disease_id]
assert disease_term["source"] == source
assert disease_term["disease_nr"] == disease_nr
def test_parse_diseases(hpo_disease_handle):
## GIVEN a iterable of disease lines
## WHEN parsing the diseases
diseases = parse_hpo_diseases(hpo_disease_handle)
## THEN assert that the diseases are parsed correct
for disease_id in diseases:
source = disease_id.split(':')[0]
disease_nr = int(disease_id.split(':')[1])
disease_term = diseases[disease_id]
assert disease_term['source'] == source
assert disease_term['disease_nr'] == disease_nr
def load_disease_terms(adapter, genemap_lines, genes=None, hpo_disease_lines=None):
"""Load the omim phenotypes into the database
Parse the phenotypes from genemap2.txt and find the associated hpo terms
from ALL_SOURCES_ALL_FREQUENCIES_diseases_to_genes_to_phenotypes.txt.
Args:
adapter(MongoAdapter)
genemap_lines(iterable(str))
genes(dict): Dictionary with all genes found in database
hpo_disease_lines(iterable(str))
"""
# Get a map with hgnc symbols to hgnc ids from scout
if not genes:
genes = adapter.genes_by_alias()
# Fetch the disease terms from omim
disease_terms = get_mim_phenotypes(genemap_lines=genemap_lines)
if not hpo_disease_lines:
hpo_disease_lines = fetch_hpo_phenotype_to_terms()
hpo_diseases = parse_hpo_diseases(hpo_disease_lines)
start_time = datetime.now()
nr_diseases = None
LOG.info("Loading the hpo disease...")
for nr_diseases, disease_number in enumerate(disease_terms):
disease_info = disease_terms[disease_number]
disease_id = "OMIM:{0}".format(disease_number)
if disease_id in hpo_diseases:
hpo_terms = hpo_diseases[disease_id]['hpo_terms']
if hpo_terms:
disease_info['hpo_terms'] = hpo_terms
disease_obj = build_disease_term(disease_info, genes)
adapter.load_disease_term(disease_obj)
LOG.info("Loading done. Nr of diseases loaded {0}".format(nr_diseases))
LOG.info("Time to load diseases: {0}".format(datetime.now() - start_time))
def test_pheno_terms(request, hpo_disease_file):
"""Get a file handle to a hpo disease file"""
print('')
hpo_disease_handle = get_file_handle(hpo_disease_file)
diseases = parse_hpo_diseases(hpo_disease_handle)
return diseases