def test_sort_seqs_by_taxonomic_depth_equal_keys(self):
"""Test sorting sequences when keys are all equal."""
exp = [['1', 3, 4, 'AGGT'], ['3', 3, 4, 'AGGC'], ['2', 3, 4, 'AGGA']]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats1)
# Since we were given a dictionary of sequence stats, the order will
# stay the same as was in the dictionary since we are doing a stable
# sort. Thus, we can't simply test for equality because the key order
# in the dictionary is not guaranteed.
self.assertEqual(obs[0][1:3], exp[0][1:3])
self.assertEqual(obs[1][1:3], exp[1][1:3])
self.assertEqual(obs[2][1:3], exp[2][1:3])
def test_sort_seqs_by_taxonomic_depth_equal_keys(self):
"""Test sorting sequences when keys are all equal."""
exp = [["1", 3, 4, "AGGT"], ["3", 3, 4, "AGGC"], ["2", 3, 4, "AGGA"]]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats1)
# Since we were given a dictionary of sequence stats, the order will
# stay the same as was in the dictionary since we are doing a stable
# sort. Thus, we can't simply test for equality because the key order
# in the dictionary is not guaranteed.
self.assertEqual(obs[0][1:3], exp[0][1:3])
self.assertEqual(obs[1][1:3], exp[1][1:3])
self.assertEqual(obs[2][1:3], exp[2][1:3])
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
seq_stats = compute_sequence_stats(
open(opts.input_fasta_fp, 'U').readlines(),
open(opts.input_taxonomy_map, 'U').readlines(),
['Incertae_sedis', 'unidentified'])
seq_stats_sorted = sort_seqs_by_taxonomic_depth(seq_stats)
# Write out our sorted sequences.
out_fasta_f = open(opts.output_fp, 'w')
for seq in seq_stats_sorted:
out_fasta_f.write('>' + seq[0] + '\n' + seq[3] + '\n')
out_fasta_f.close()
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
seq_stats = compute_sequence_stats(
open(opts.input_fasta_fp, 'U').readlines(),
open(opts.input_taxonomy_map, 'U').readlines(),
['Incertae_sedis', 'unidentified'])
seq_stats_sorted = sort_seqs_by_taxonomic_depth(seq_stats)
# Write out our sorted sequences.
out_fasta_f = open(opts.output_fp, 'w')
for seq in seq_stats_sorted:
out_fasta_f.write('>' + seq[0] + '\n' + seq[3] + '\n')
out_fasta_f.close()
def test_sort_seqs_by_taxonomic_depth_mising_seq_in_fasta(self):
"""Test sorting with a sequence that was missing from FASTA file."""
saved_stdout = sys.stdout
try:
out = StringIO()
sys.stdout = out
exp = [["2", 3, 7, "AGGAGTC"], ["3", 3, 5, "AGGTA"]]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats6)
self.assertEqual(obs, exp)
output = out.getvalue().strip()
self.assertEqual(
output, "Found sequence id '42' in the taxonomy " "mapping file that wasn't in the FASTA file"
)
finally:
sys.stdout = saved_stdout
def test_sort_seqs_by_taxonomic_depth_mising_seq_in_fasta(self):
"""Test sorting with a sequence that was missing from FASTA file."""
saved_stdout = sys.stdout
try:
out = StringIO()
sys.stdout = out
exp = [['2', 3, 7, 'AGGAGTC'], ['3', 3, 5, 'AGGTA']]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats6)
self.assertEqual(obs, exp)
output = out.getvalue().strip()
self.assertEqual(
output, "Found sequence id '42' in the taxonomy "
"mapping file that wasn't in the FASTA file")
finally:
sys.stdout = saved_stdout
def test_sort_seqs_by_taxonomic_depth_all_same_tax_depth(self):
"""Test sorting with all taxonomic depths being the same."""
exp = [["2", 3, 7, "AGGAGTC"], ["3", 3, 5, "AGGTA"], ["1", 3, 4, "AGGT"]]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats5)
self.assertEqual(obs, exp)
def test_sort_seqs_by_taxonomic_depth_single_seq(self):
"""Test sorting a single sequence."""
exp = [["4", 3, 4, "AGGT"]]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats3)
self.assertEqual(obs, exp)
def test_sort_seqs_by_taxonomic_depth_unequal_keys(self):
"""Test sorting sequences when keys are all unequal."""
exp = [["2", 6, 7, "AGGAGTC"], ["3", 5, 1, "A"], ["1", 3, 4, "AGGT"]]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats2)
self.assertEqual(obs, exp)
def test_sort_seqs_by_taxonomic_depth_all_same_tax_depth(self):
"""Test sorting with all taxonomic depths being the same."""
exp = [['2', 3, 7, 'AGGAGTC'], ['3', 3, 5, 'AGGTA'],
['1', 3, 4, 'AGGT']]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats5)
self.assertEqual(obs, exp)
def test_sort_seqs_by_taxonomic_depth_single_seq(self):
"""Test sorting a single sequence."""
exp = [['4', 3, 4, 'AGGT']]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats3)
self.assertEqual(obs, exp)
def test_sort_seqs_by_taxonomic_depth_unequal_keys(self):
"""Test sorting sequences when keys are all unequal."""
exp = [['2', 6, 7, 'AGGAGTC'], ['3', 5, 1, 'A'], ['1', 3, 4, 'AGGT']]
obs = sort_seqs_by_taxonomic_depth(self.seq_stats2)
self.assertEqual(obs, exp)
