def test_validate_fasta(self):
""" Overall module runs properly """
validate_fasta(self.sample_fasta_fp, self.sample_mapping_fp,
self.output_dir)
expected_log_fp = join(self.output_dir,
split(self.sample_fasta_fp)[1] + "_report.log")
log_f = open(expected_log_fp, "U")
actual_log_lines = [line.strip() for line in log_f][1:]
expected_log_lines = """Percent duplicate labels: 0.000
Percent QIIME-incompatible fasta labels: 0.000
Percent of labels that fail to map to SampleIDs: 0.000
Percent of sequences with invalid characters: 0.000
Percent of sequences with barcodes detected: 0.000
Percent of sequences with barcodes detected at the beginning of the sequence: 0.000
Percent of sequences with primers detected: 0.000""".split('\n')
self.assertEqual(actual_log_lines, expected_log_lines)
# Check with all optional values included
validate_fasta(self.sample_fasta_fp, self.sample_mapping_fp,
self.output_dir, tree_fp=self.sample_tree_5tips_fp, tree_subset=True,
tree_exact_match=True, same_seq_lens=True, all_ids_found=True)
expected_log_fp = join(self.output_dir,
split(self.sample_fasta_fp)[1] + "_report.log")
log_f = open(expected_log_fp, "U")
actual_log_lines = [line.strip() for line in log_f][1:]
expected_log_lines = """Percent duplicate labels: 0.000
Percent QIIME-incompatible fasta labels: 0.000
Percent of labels that fail to map to SampleIDs: 0.000
Percent of sequences with invalid characters: 0.000
Percent of sequences with barcodes detected: 0.000
Percent of sequences with barcodes detected at the beginning of the sequence: 0.000
Percent of sequences with primers detected: 0.000
Sequence lengths report
Counts of sequences, followed by their sequence lengths:
1\t35
1\t32
1\t27
Sample ID in fasta sequences report
The following SampleIDs were not found:
seq2
Fasta label subset in tree tips report
All fasta labels were a subset of tree tips.
Fasta label/tree tip exact match report
All fasta labels found in tree tips.
The following tips were not in fasta labels:
seq2
seq5
seq4""".split('\n')
self.assertEqual(actual_log_lines, expected_log_lines)
def test_validate_fasta_suppress_primers_barcodes(self):
""" Overall module test with primer/barcode check suppressed """
# Should raise errors when both primer/barcode check not suppressed
self.assertRaises(ValueError, validate_fasta, self.sample_fasta_fp,
self.sample_mapping_file_errors_fp, self.output_dir)
self.assertRaises(ValueError, validate_fasta, self.sample_fasta_fp,
self.sample_mapping_file_errors_fp, self.output_dir,
suppress_primer_checks=True)
self.assertRaises(ValueError, validate_fasta, self.sample_fasta_fp,
self.sample_mapping_file_errors_fp, self.output_dir,
suppress_barcode_checks=True)
# No errors when both suppressed
validate_fasta(self.sample_fasta_fp, self.sample_mapping_file_errors_fp,
self.output_dir, suppress_primer_checks = True,
suppress_barcode_checks = True)
expected_log_fp = join(self.output_dir,
split(self.sample_fasta_fp)[1] + "_report.log")
log_f = open(expected_log_fp, "U")
actual_log_lines = [line.strip() for line in log_f][1:]
expected_log_lines = """Percent duplicate labels: 0.000
Percent QIIME-incompatible fasta labels: 0.000
Percent of labels that fail to map to SampleIDs: 0.000
Percent of sequences with invalid characters: 0.000
Percent of sequences with barcodes detected: 0.000
Percent of sequences with barcodes detected at the beginning of the sequence: 0.000
Percent of sequences with primers detected: 0.000""".split('\n')
self.assertEqual(actual_log_lines, expected_log_lines)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
verbose = opts.verbose
input_fasta_fp = opts.input_fasta_fp
mapping_fp = opts.mapping_fp
output_dir = opts.output_dir
tree_fp = opts.tree_fp
tree_subset = opts.tree_subset
tree_exact_match = opts.tree_exact_match
same_seq_lens = opts.same_seq_lens
all_ids_found = opts.all_ids_found
create_dir(output_dir)
# Test optional filepaths and requirements
try:
test_mapping_fp = open(mapping_fp, "U")
test_mapping_fp.close()
except IOError:
raise IOError,("Unable to open mapping file, please check "
"filepath and read permissions.")
if tree_fp:
try:
test_tree_fp = open(tree_fp, "U")
test_tree_fp.close()
except IOError:
raise IOError,("Unable to open provided tree filepath, please "+\
"filepath and permissions.")
if tree_subset or tree_exact_match:
if not tree_fp:
raise ValueError,('Must provide tree filepath if -s or -e options '+\
'are enabled.')
validate_fasta(input_fasta_fp, mapping_fp, output_dir, tree_fp, tree_subset,
tree_exact_match, same_seq_lens, all_ids_found,
opts.suppress_barcode_checks, opts.suppress_primer_checks)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
verbose = opts.verbose
input_fasta_fp = opts.input_fasta_fp
mapping_fp = opts.mapping_fp
output_dir = opts.output_dir
tree_fp = opts.tree_fp
tree_subset = opts.tree_subset
tree_exact_match = opts.tree_exact_match
same_seq_lens = opts.same_seq_lens
all_ids_found = opts.all_ids_found
create_dir(output_dir)
# Test optional filepaths and requirements
try:
test_mapping_fp = open(mapping_fp, "U")
test_mapping_fp.close()
except IOError:
raise IOError("Unable to open mapping file, please check "
"filepath and read permissions.")
if tree_fp:
try:
test_tree_fp = open(tree_fp, "U")
test_tree_fp.close()
except IOError:
raise IOError("Unable to open provided tree filepath, please " +
"filepath and permissions.")
if tree_subset or tree_exact_match:
if not tree_fp:
raise ValueError('Must provide tree filepath if -s or -e options ' +
'are enabled.')
validate_fasta(
input_fasta_fp, mapping_fp, output_dir, tree_fp, tree_subset,
tree_exact_match, same_seq_lens, all_ids_found,
opts.suppress_barcode_checks, opts.suppress_primer_checks)
def test_validate_fasta_with_invalid(self):
""" Overall module runs properly """
validate_fasta(self.sample_fasta_invalid_fp, self.sample_mapping_fp,
self.output_dir)
expected_log_fp = join(self.output_dir,
split(self.sample_fasta_invalid_fp)[1] + "_report.log")
log_f = open(expected_log_fp, "U")
actual_log_lines = [line.strip() for line in log_f][1:]
expected_log_lines = """Percent duplicate labels: 0.250
Percent QIIME-incompatible fasta labels: 0.500
Percent of labels that fail to map to SampleIDs: 0.750
Percent of sequences with invalid characters: 0.500
Percent of sequences with barcodes detected: 0.250
Percent of sequences with barcodes detected at the beginning of the sequence: 0.000
Percent of sequences with primers detected: 0.250""".split('\n')
self.assertEqual(actual_log_lines, expected_log_lines)
def test_validate_fasta_with_invalid(self):
""" Overall module runs properly """
validate_fasta(self.sample_fasta_invalid_fp, self.sample_mapping_fp,
self.output_dir)
expected_log_fp = join(
self.output_dir,
split(self.sample_fasta_invalid_fp)[1] + "_report.log")
log_f = open(expected_log_fp, "U")
actual_log_lines = [line.strip() for line in log_f][1:]
expected_log_lines = """Percent duplicate labels: 0.250
Percent QIIME-incompatible fasta labels: 0.500
Percent of labels that fail to map to SampleIDs: 0.750
Percent of sequences with invalid characters: 0.500
Percent of sequences with barcodes detected: 0.250
Percent of sequences with barcodes detected at the beginning of the sequence: 0.000
Percent of sequences with primers detected: 0.250""".split('\n')
self.assertEqual(actual_log_lines, expected_log_lines)