def test_compute_sequence_stats_no_unknowns(self):
"""Test computing seq stats on taxonomy with no unknown kws."""
exp = {"1": [3, 4, "AGGT"], "2": [3, 4, "AGGA"], "3": [5, 4, "AGGC"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map2)
self.assertEqual(obs, exp)
obs = compute_sequence_stats(self.fasta1, self.tax_map2, [])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_no_unknowns(self):
"""Test computing seq stats on taxonomy with no unknown kws."""
exp = {'1': [3, 4, 'AGGT'], '2': [3, 4, 'AGGA'], '3': [5, 4, 'AGGC']}
obs = compute_sequence_stats(self.fasta1, self.tax_map2)
self.assertEqual(obs, exp)
obs = compute_sequence_stats(self.fasta1, self.tax_map2, [])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_single_depth(self):
"""Test computing seq stats on taxonomy with only one level."""
exp = {"1": [3, 4, "AGGT"], "3": [3, 4, "AGGC"], "2": [1, 4, "AGGA"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map3, ["Z"])
self.assertEqual(obs, exp)
exp = {"1": [2, 4, "AGGT"], "3": [4, 4, "AGGC"], "2": [0, 4, "AGGA"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map3, ["A"])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_single_depth(self):
"""Test computing seq stats on taxonomy with only one level."""
exp = {'1': [3, 4, 'AGGT'], '3': [3, 4, 'AGGC'], '2': [1, 4, 'AGGA']}
obs = compute_sequence_stats(self.fasta1, self.tax_map3, ['Z'])
self.assertEqual(obs, exp)
exp = {'1': [2, 4, 'AGGT'], '3': [4, 4, 'AGGC'], '2': [0, 4, 'AGGA']}
obs = compute_sequence_stats(self.fasta1, self.tax_map3, ['A'])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_missing_taxonomy(self):
"""Test computing seq stats on a sequence with no taxonomy mapping."""
# Save stdout and replace it with something that will capture the print
# statement. Note: this code was taken from here:
# http://stackoverflow.com/questions/4219717/how-to-assert-output-
# with-nosetest-unittest-in-python/4220278#4220278
saved_stdout = sys.stdout
try:
out = StringIO()
sys.stdout = out
self.fasta1.append(">4")
self.fasta1.append("AACCGGTT")
exp = {
'1': [3, 4, 'AGGT'],
'3': [2, 4, 'AGGC'],
'2': [3, 4, 'AGGA'],
'4': [0, 8, 'AACCGGTT']
}
obs = compute_sequence_stats(self.fasta1, self.tax_map1, ['Z'])
self.assertEqual(obs, exp)
output = out.getvalue().strip()
self.assertEqual(
output, "Found sequence id '4' in the FASTA file "
"that wasn't in the taxonomy mapping file")
finally:
sys.stdout = saved_stdout
def test_compute_sequence_stats_unequal_read_lengths(self):
"""Test computing seq stats on unequal read lengths."""
exp = {
'1': [3, 6, 'AGGTAC'],
'3': [2, 7, 'AGGCAAA'],
'2': [3, 2, 'AG']
}
obs = compute_sequence_stats(self.fasta2, self.tax_map1, ['Z'])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_empty_taxonomy_levels(self):
"""Test computing seq stats on taxonomies with empty levels."""
exp = {
'1': [3, 6, 'AGGTAC'],
'3': [3, 7, 'AGGCAAA'],
'2': [3, 2, 'AG']
}
obs = compute_sequence_stats(self.fasta2, self.tax_map4, ['Z'])
self.assertEqual(obs, exp)
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_compute_sequence_stats_missing_taxonomy(self):
"""Test computing seq stats on a sequence with no taxonomy mapping."""
# Save stdout and replace it with something that will capture the print
# statement. Note: this code was taken from here:
# http://stackoverflow.com/questions/4219717/how-to-assert-output-
# with-nosetest-unittest-in-python/4220278#4220278
saved_stdout = sys.stdout
try:
out = StringIO()
sys.stdout = out
self.fasta1.append(">4")
self.fasta1.append("AACCGGTT")
exp = {"1": [3, 4, "AGGT"], "3": [2, 4, "AGGC"], "2": [3, 4, "AGGA"], "4": [0, 8, "AACCGGTT"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map1, ["Z"])
self.assertEqual(obs, exp)
output = out.getvalue().strip()
self.assertEqual(
output, "Found sequence id '4' in the FASTA file " "that wasn't in the taxonomy mapping file"
)
finally:
sys.stdout = saved_stdout
def test_compute_sequence_stats_multiple_unknown_keywords(self):
"""Test computing seq stats on taxonomy with multiple unknown kws."""
exp = {"1": [3, 4, "AGGT"], "3": [2, 4, "AGGC"], "2": [3, 4, "AGGA"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map2, ["Z", "F"])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_multiple_unknown_keywords(self):
"""Test computing seq stats on taxonomy with multiple unknown kws."""
exp = {'1': [3, 4, 'AGGT'], '3': [2, 4, 'AGGC'], '2': [3, 4, 'AGGA']}
obs = compute_sequence_stats(self.fasta1, self.tax_map2, ['Z', 'F'])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_unequal_depths(self):
"""Test computing seq stats on unequal taxonomic depths."""
exp = {"1": [3, 4, "AGGT"], "3": [2, 4, "AGGC"], "2": [3, 4, "AGGA"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map1, ["Z"])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_empty_taxonomy_levels(self):
"""Test computing seq stats on taxonomies with empty levels."""
exp = {"1": [3, 6, "AGGTAC"], "3": [3, 7, "AGGCAAA"], "2": [3, 2, "AG"]}
obs = compute_sequence_stats(self.fasta2, self.tax_map4, ["Z"])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_unequal_read_lengths(self):
"""Test computing seq stats on unequal read lengths."""
exp = {"1": [3, 6, "AGGTAC"], "3": [2, 7, "AGGCAAA"], "2": [3, 2, "AG"]}
obs = compute_sequence_stats(self.fasta2, self.tax_map1, ["Z"])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_all_unknown(self):
"""Test computing seq stats on taxonomy with all unknown tax levels."""
exp = {"1": [0, 4, "AGGT"], "2": [0, 4, "AGGA"], "3": [0, 4, "AGGC"]}
obs = compute_sequence_stats(self.fasta1, self.tax_map2, ["A", "B", "C", "D", "F", "Z"])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_all_unknown(self):
"""Test computing seq stats on taxonomy with all unknown tax levels."""
exp = {'1': [0, 4, 'AGGT'], '2': [0, 4, 'AGGA'], '3': [0, 4, 'AGGC']}
obs = compute_sequence_stats(self.fasta1, self.tax_map2,
['A', 'B', 'C', 'D', 'F', 'Z'])
self.assertEqual(obs, exp)
def test_compute_sequence_stats_unequal_depths(self):
"""Test computing seq stats on unequal taxonomic depths."""
exp = {'1': [3, 4, 'AGGT'], '3': [2, 4, 'AGGC'], '2': [3, 4, 'AGGA']}
obs = compute_sequence_stats(self.fasta1, self.tax_map1, ['Z'])
self.assertEqual(obs, exp)
