def test_blastparse(self):
queries, clusters = core.blastparse(self.sample_data, Record(), 0, 0)
# check we process the right number of queries
self.assertEqual(len(queries), len(set([i[0] for i in self.sample_data_as_lists])))
# check we have entries for every gene_cluster we found
subjects = [self.parse_subject_wrapper(i) for i in self.sample_data_as_lists]
self.verify_subjects_and_clusters_represented(subjects, clusters)
# test perc_coverage threshold (value arbitrary due to mocking)
coverage_threshold = 650
queries, clusters = core.blastparse(self.sample_data, Record(), coverage_threshold, 0)
new_subjects = [s for s in subjects if s.perc_coverage > coverage_threshold]
assert new_subjects and len(new_subjects) < len(subjects), "coverage test has become meaningless"
self.verify_subjects_and_clusters_represented(new_subjects, clusters)
# test perc_identity threshold
ident_threshold = 35
queries, clusters = core.blastparse(self.sample_data, Record(), 0, ident_threshold)
new_subjects = [s for s in subjects if s.perc_ident > ident_threshold]
assert new_subjects and len(new_subjects) < len(subjects), "identity% test has become meaningless"
self.verify_subjects_and_clusters_represented(new_subjects, clusters)
# test combo threshold
queries, clusters = core.blastparse(self.sample_data, Record(), coverage_threshold, ident_threshold)
new_subjects = [s for s in subjects if s.perc_ident > ident_threshold and s.perc_coverage > coverage_threshold]
assert new_subjects and len(new_subjects) < len(subjects), "combo test has become meaningless"
self.verify_subjects_and_clusters_represented(new_subjects, clusters)
def test_blastparse_on_empty(self):
for blast in ["", "\n", "\r\n", "\n\n"]:
queries, clusters = core.blastparse(blast, Record(), 0, 0)
self.assertEqual(len(queries), 0)
self.assertEqual(len(clusters), 0)
