class TestLabels(unittest.TestCase):
def setUp(self):
self.go = OBO('files/test_data/go.obo')
self.go.populate_annotations(GO_URL, remote_location=True)
self.dsrepair_term = self.go.get_term(DSREPAIR_ID)
lines = open('files/go_neg_slim.txt').readlines()
self.slim_terms = set([l.strip() for l in lines])
def testOntoLabelsNeg(self):
ol = OntoLabels(obo=self.go, slim_terms=self.slim_terms)
(pos, neg) = ol.get_labels(DSREPAIR_ID)
self.assertTrue(len(pos) > 0)
self.assertEqual(pos,set(self.dsrepair_term.get_annotated_genes()))
similar_term = self.go.get_term(DNA_REP_ID).get_annotated_genes()
slim_overlap = self.go.get_ancestors(DNA_REP_ID) & \
self.go.get_ancestors(DSREPAIR_ID) & \
self.slim_terms
self.assertTrue(DNA_METAB in slim_overlap)
self.assertTrue(len(neg & set(similar_term)) == 0)
for dterm in self.go.get_descendents(DNA_METAB):
dgenes = set(self.go.get_term(dterm).get_annotated_genes())
self.assertTrue(len(neg & dgenes) == 0)
sys.exit()
if args.pub_filter and args.nspace is None:
sys.stderr.write(
"--When filtering by publication, must provide GO namespace.\n")
sys.exit()
id_name = None
if args.idfile is not None:
id_name = IDMap(args.idfile)
gene_ontology = OBO(args.obo)
logger.info('Populating gene associations')
if args.ass:
gene_ontology.populate_annotations(args.ass,
gene_col=args.gcol,
term_col=args.term_col)
elif args.gmt:
gmt = GMT(args.gmt)
gene_ontology.populate_annotations_from_gmt(gmt)
else:
sys.stderr.write(
"--Provide gene annotations from an association file or a GMT file")
exit()
if args.pub_filter:
pub_counts = defaultdict(set)
for (term_id, term) in gene_ontology.go_terms.iteritems():
if term.namespace != args.nspace:
continue
for a in term.annotations:
args = parser.parse_args()
if args.obo is None:
sys.stderr.write("--obo file is required.\n")
sys.exit()
if args.pub_filter and args.nspace is None:
sys.stderr.write(
"--When filtering by publication, must provide GO namespace.\n")
sys.exit()
gene_ontology = OBO(args.obo)
logger.info('Populating gene associations')
if args.ass:
gene_ontology.populate_annotations(args.ass,
remote_location=args.remote_file,
gene_col=args.gcol,
term_col=args.term_col)
elif args.gmt:
gmt = GMT(args.gmt)
gene_ontology.populate_annotations_from_gmt(gmt)
else:
sys.stderr.write(
"--Provide gene annotations from an association file or a GMT file")
exit()
if args.pub_filter:
pub_counts = defaultdict(set)
for (term_id, term) in gene_ontology.go_terms.iteritems():
if term.namespace != args.nspace:
continue
for a in term.annotations: