def test_convert_nt_output_format(self):
# tests that a length 3 tupple output, and each is the correct numpy
# array type
sequences = read_fasta('test_data/test_frames.fa')
ids, aa_frames, frame, strand, seq_length_nt, seq_length = translate_all_frames(
sequences, both_strands=False)
orf_sequence, start_sites, stop_sites, orf_length, last_aa_is_stop = find_longest_orfs(
aa_frames)
# filter data by minimum orf length
keep = orf_length >= 6
aa_frames, frame, strand, seq_length_nt, ids, seq_length, start_sites, stop_sites, orf_sequence, last_aa_is_stop, orf_length = filter_objects(
keep, aa_frames, frame, strand, seq_length_nt, ids, seq_length,
start_sites, stop_sites, orf_sequence, last_aa_is_stop, orf_length)
output = convert_start_stop_to_nt(start_sites, stop_sites,
seq_length_nt, orf_length, frame,
last_aa_is_stop)
t_len = len(output) == 3
# test numpy types of all outputs
t0 = output[0].dtype == 'int64'
t1 = output[1].dtype == 'int64'
t2 = output[2].dtype == 'int64'
all_right_types = t0 and t1 and t2 and t_len
self.assertTrue(all_right_types)
def test_translate_bothstrands(self):
sequences = read_fasta('test_data/test_trans_all_frames.fa')
ids, aa_frames, frame, strand, seq_length_nt, seq_length = translate_all_frames(
sequences, both_strands=True)
frame_correct = np.all(frame == np.array([1, 1, 2, 2, 3, 3]))
strand_correct = np.all(
strand == np.array(['+', '-', '+', '-', '+', '-']))
trans_correct = np.all(aa_frames == np.array([
'MANATEE*', 'LFFGRVRH', 'WRTRPKN', 'YSSVAFA', 'GERDRRI', 'ILRSRSP'
]))
self.assertTrue(frame_correct and strand_correct and trans_correct)
def test_convert_utr_nt(self):
sequences = read_fasta('test_data/test_frames.fa')
ids, aa_frames, frame, strand, seq_length_nt, seq_length = translate_all_frames(
sequences, both_strands=False)
orf_sequence, start_sites, stop_sites, orf_length, last_aa_is_stop = find_longest_orfs(
aa_frames)
# filter data by minimum orf length
keep = orf_length >= 6
aa_frames, frame, strand, seq_length_nt, ids, seq_length, start_sites, stop_sites, orf_sequence, last_aa_is_stop, orf_length = filter_objects(
keep, aa_frames, frame, strand, seq_length_nt, ids, seq_length,
start_sites, stop_sites, orf_sequence, last_aa_is_stop, orf_length)
start_site_nt, stop_site_nt, utr3_length = convert_start_stop_to_nt(
start_sites, stop_sites, seq_length_nt, orf_length, frame,
last_aa_is_stop)
self.assertTrue(np.all(utr3_length == np.array([5, 4, 3])))
def test_translate_output_format(self):
# tests that a length 3 tupple output, and each is the correct numpy
# array type
sequences = read_fasta('test_data/test_trans_all_frames.fa')
output = translate_all_frames(sequences, both_strands=False)
t_len = len(output) == 6
# test numpy types of all outputs
t0 = output[0].dtype.type == np.str_
t1 = output[1].dtype.type == np.str_
t2 = output[2].dtype == 'int64'
t3 = output[3].dtype.type == np.str_
t4 = output[4].dtype == 'int64'
t5 = output[5].dtype == 'int64'
all_right_types = t0 and t1 and t2 and t3 and t4 and t5 and t_len
self.assertTrue(all_right_types)
def test_translate_seq_length(self):
sequences = read_fasta('test_data/test_trans_all_frames.fa')
ids, aa_frames, frame, strand, seq_length_nt, seq_length = translate_all_frames(
sequences, both_strands=False)
self.assertTrue(np.all(seq_length == np.array([8, 7, 7])))
def test_translate_alltransframes(self):
sequences = read_fasta('test_data/test_trans_all_frames.fa')
ids, aa_frames, frame, strand, seq_length_nt, seq_length = translate_all_frames(
sequences, both_strands=False)
self.assertTrue(
np.all(aa_frames == np.array(['MANATEE*', 'WRTRPKN', 'GERDRRI'])))