def test_seq_subset_load_from_all_genes(self):
seqfile_ops.write_genbank(self.gbk_filename, self.record)
subset_mode = 'features'
subset_args = {'types': ('genes'), 'tags': {}}
subset, subset_file = dataset_load.seq_subset_load(self.gbk_filename,
subset_mode,
subset_args)
self.assertIs(len(subset), 2)
def test_seq_subset_load_from_chop_by_size(self):
seqfile_ops.write_fasta(self.single_q_file, self.single_record)
subset_mode = 'size'
subset_args = {'size': 5, 'chop_mode': 'exact_size'}
subset, subset_file = dataset_load.seq_subset_load(self.single_q_file,
subset_mode,
subset_args)
self.assertIs(len(subset), 10)
self.assertEqual(subset[0].id, 'temp_1_0-5')
def test_seq_subset_load_from_cds_by_locus_tag(self):
seqfile_ops.write_genbank(self.gbk_filename, self.record)
subset_mode = 'features'
subset_args = {'types': ('CDS'), 'tags': {'locus_tag': ['locustag 1',
'locustag 4']}}
subset, subset_file = dataset_load.seq_subset_load(self.gbk_filename,
subset_mode,
subset_args)
self.assertIs(len(subset), 1)
self.assertEqual(subset[0].id, 'temp_5-10')
def test_seq_subset_load_from_mixed_features(self):
seqfile_ops.write_genbank(self.gbk_filename, self.record)
subset_mode = 'features'
subset_args = {'types': ('CDS', 'gene'), 'tags': {'locus_tag':
('locustag 3'),
'product':
(('product 2'))}}
subset, subset_file = dataset_load.seq_subset_load(self.gbk_filename,
subset_mode,
subset_args)
self.assertIs(len(subset), 2)
def test_seq_subset_load_from_multifasta(self):
seqfile_ops.write_fasta(self.multi_q_file, self.multi_records)
subset_mode = 'flatfile'
subset_args = None
subset, subset_file = dataset_load.seq_subset_load(self.multi_q_file,
subset_mode,
subset_args)
self.assertIs(len(subset), 5)
index = 0
for record in subset:
self.assertEqual(subset[index].id, self.multi_records[index].id)
index += 1
self.assertIs(subset_file, self.multi_q_file)
def test_seq_subset_load_from_coords(self):
seqfile_ops.write_fasta(self.single_q_file, self.single_record)
temp_file = open(self.coords_file, 'w')
temp_file.write(self.str_contents)
temp_file.close()
subset_mode = 'coordinates'
subset_args = {'file': self.coords_file, 'header': 1,
'columns': (1, 2)}
subset, subset_file = dataset_load.seq_subset_load(self.single_q_file,
subset_mode,
subset_args)
self.assertIs(len(subset), 3)
self.assertEqual(subset[0].id, 'temp_1_0-10')
self.assertEqual(str(subset[2].seq), 'TTTGGCGCTCGCGGCGGG')
