def test_020_gff3_i0(self):
"""check functions for retreiving input filename, and reading and validating a GFF3 file and separate FASTA file"""
newmunger = GFFMunger(None)
newmunger.input_file_arg = test_gff_no_fasta
newmunger.fasta_file_arg = test_fasta_file
newmunger.output_file = self.output_file
self.assertEqual(test_gff_no_fasta, newmunger.get_gff3_source())
self.assertTrue(newmunger.validate_GFF3(test_gff_no_fasta))
self.assertTrue(newmunger.validate_FASTA(test_fasta_file))
self.assertEqual(newmunger.gffutils_db_filename,
newmunger.import_gff3())
self.assertEqual(newmunger.gffutils_db_filename,
newmunger.import_gff3(test_gff_no_fasta))
# when FASTA is in separate file, GFFMunger.extract_GFF3_components() should read only metadata
self.assertEqual(expected_num_metadata_lines,
newmunger.extract_GFF3_components())
self.assertEqual(expected_num_metadata_lines,
newmunger.extract_GFF3_components(test_gff_no_fasta))
with warnings.catch_warnings():
warnings.filterwarnings("ignore", "unclosed file <_io\.FileIO",
ResourceWarning, "six", 581)
warnings.filterwarnings("ignore", "", ResourceWarning, "gffutils",
0)
self.assertIsInstance(newmunger.import_fasta(), pyfaidx.Fasta)
self.assertIsInstance(newmunger.import_fasta(test_fasta_file),
pyfaidx.Fasta)
try:
newmunger.move_polypeptide_annotations()
except:
self.fail("Failed to process valid GFF3 file " +
test_gff_no_fasta)
warnings.resetwarnings()
self.assertTrue(newmunger.export_gff3())
os.remove(newmunger.output_file)
newmunger.clean_up()
def test_000_gff3_with_fasta_io(self):
"""check functions for reading and validating a GFF3 file including FASTA, and writing it as output"""
gffmunger = GFFMunger(None)
gffmunger.input_file_arg = test_gff_file
gffmunger.output_file = self.output_file
self.assertEqual(test_gff_file, gffmunger.get_gff3_source())
self.assertTrue(gffmunger.validate_GFF3(test_gff_file))
self.assertEqual(gffmunger.gffutils_db_filename,
gffmunger.import_gff3())
self.assertEqual(gffmunger.gffutils_db_filename,
gffmunger.import_gff3(test_gff_file))
self.assertEqual(expected_num_input_lines,
gffmunger.extract_GFF3_components())
self.assertEqual(expected_num_input_lines,
gffmunger.extract_GFF3_components(test_gff_file))
with warnings.catch_warnings():
warnings.filterwarnings("ignore",
"unclosed file <_io\.TextIOWrapper",
ResourceWarning, "gffutils", 668)
try:
gffmunger.move_polypeptide_annotations()
except:
self.fail("Failed to process valid GFF3 file " + test_gff_file)
warnings.resetwarnings()
self.assertTrue(gffmunger.export_gff3())
os.remove(gffmunger.output_file)
gffmunger.clean_up()
class GFF_Tests(unittest.TestCase):
test_gff_db = None
db_available = False
@classmethod
def setUpClass(self):
with warnings.catch_warnings():
warnings.filterwarnings("ignore",
"unclosed file <_io\.TextIOWrapper",
ResourceWarning, "gffutils", 133)
warnings.filterwarnings("ignore", "generator '_FileIterator\.",
PendingDeprecationWarning, "gffutils", 186)
warnings.filterwarnings("ignore",
"unclosed file <_io\.TextIOWrapper",
ResourceWarning, "gffutils", 668)
self.test_gff_db = gffutils.create_db(
test_gff_file,
dbfn=str(test_gff_db_file).replace('<uid>',
uuid.uuid4().hex),
force=True, # overwrite previous testing db file
merge_strategy='error',
keep_order=
False, # True doesn't appear to maintain attribute order :-/ (and turning this off may be faster)
sort_attribute_values=False)
if (self.test_gff_db is not None
and isinstance(self.test_gff_db, expected_db_class)):
self.db_available = True
self.gffmunger = GFFMunger(None)
def test_000_db_created(self):
"""test that gffutils.create_db was able to create a database from the sample GFF file"""
#self.assertIsNotNone(self.test_gff_db)
#self.assertIsInstance(self.test_gff_db, expected_db_class)
self.assertTrue(self.db_available)
#@unittest.skipUnless(db_available, "no db available")
def test_010_gffutils_find_gene(self):
"""test that gffutils can find a sample gene identifier from the GFF"""
if (not self.db_available):
self.skipTest('no db available')
this_gene = self.test_gff_db[sample_gff_gene_id]
self.assertIsNotNone(this_gene)
self.assertIsInstance(this_gene, expected_feature_class)
def test_020_gffutils_find_features(self):
"""test that gffutils can find features in GFF"""
if (not self.db_available):
self.skipTest('no db available')
for this_featuretype in sample_gff_featuretypes:
for this_feature in self.test_gff_db.children(
sample_gff_gene_id, featuretype=this_featuretype):
self.assertIsNotNone(this_feature)
self.assertIsInstance(this_feature, expected_feature_class)
# print(this_feature)
def test_030_gffutils_find_attributes(self):
"""test that gffutils can find attributes in GFF"""
if (not self.db_available):
self.skipTest('no db available')
for this_featuretype in sample_gff_featuretypes:
for this_feature in self.test_gff_db.children(
sample_gff_gene_id, featuretype=this_featuretype):
#print(this_feature)
self.assertIsNotNone(this_feature)
self.assertIsInstance(this_feature, expected_feature_class)
these_attributes = this_feature.attributes
self.assertIsNotNone(these_attributes)
self.assertIsInstance(these_attributes,
expected_attributes_class)
for this_item in sorted(these_attributes.items()):
self.assertIsNotNone(this_item)
#print('{this_item[0]}: {this_item[1]}'.format(this_item=this_item))
def test_040_gff_components_no_fasta(self):
"""test separation of GFF3 file without FASTA, into metadata and features """
self.gffmunger.extract_GFF3_components(test_gff_file)
self.assertIsNotNone(self.gffmunger.input_metadata)
if self.gffmunger.read_features_to_buffer:
self.assertIsNotNone(self.gffmunger.input_features)
else:
self.assertIsNone(self.gffmunger.input_features)
self.assertIsNone(self.gffmunger.input_fasta)
def test_045_gff_components_with_fasta(self):
"""test separation of GFF3 file with FASTA, into metadata, features and FASTA data"""
self.gffmunger.extract_GFF3_components(test_gff_and_fasta_file)
self.assertIsNotNone(self.gffmunger.input_fasta)