def __call__(self, seq_path, result_path=None, log_path=None):
"""Returns dict mapping {seq_id:(taxonomy, confidence)} for
each seq.
Parameters:
seq_path: path to file of sequences
result_path: path to file of results. If specified, dumps the
result to the desired path instead of returning it.
log_path: path to log, which should include dump of params.
"""
if log_path:
self.writeLog(log_path)
reference_sequences_fp = self.Params['reference_sequences_fp']
assert reference_sequences_fp, \
"Must provide reference_sequences_fp when calling an RtaxTaxonAssigner."
id_to_taxonomy_fp = self.Params['id_to_taxonomy_fp']
assert id_to_taxonomy_fp, \
"Must provide id_to_taxonomy_fp when calling an RtaxTaxonAssigner."
# delimiter = self.Params['delimiter']
read_1_seqs_fp = self.Params['read_1_seqs_fp']
assert read_1_seqs_fp, \
"Must provide read_1_seqs_fp when calling an RtaxTaxonAssigner."
# following params may all be null
read_2_seqs_fp = self.Params['read_2_seqs_fp']
single_ok = self.Params['single_ok']
no_single_ok_generic = self.Params['no_single_ok_generic']
header_id_regex = self.Params['header_id_regex']
assert header_id_regex, \
"Must not provide empty header_id_regex when calling an RtaxTaxonAssigner; leave unset"\
"to use default if in doubt."
read_id_regex = self.Params['read_id_regex']
amplicon_id_regex = self.Params['amplicon_id_regex']
# seq_file = open(seq_path, 'r')
results = rtax.assign_taxonomy(
seq_path,
reference_sequences_fp,
id_to_taxonomy_fp,
read_1_seqs_fp,
read_2_seqs_fp,
single_ok=single_ok,
no_single_ok_generic=no_single_ok_generic,
header_id_regex=header_id_regex,
read_id_regex=read_id_regex,
amplicon_id_regex=amplicon_id_regex,
output_fp=result_path,
log_path=log_path,
base_tmp_dir=get_qiime_temp_dir())
return results
def __call__(self, seq_path, result_path=None, log_path=None):
"""Returns dict mapping {seq_id:(taxonomy, confidence)} for
each seq.
Parameters:
seq_path: path to file of sequences
result_path: path to file of results. If specified, dumps the
result to the desired path instead of returning it.
log_path: path to log, which should include dump of params.
"""
if log_path:
self.writeLog(log_path)
reference_sequences_fp = self.Params["reference_sequences_fp"]
assert reference_sequences_fp, "Must provide reference_sequences_fp when calling an RtaxTaxonAssigner."
id_to_taxonomy_fp = self.Params["id_to_taxonomy_fp"]
assert id_to_taxonomy_fp, "Must provide id_to_taxonomy_fp when calling an RtaxTaxonAssigner."
# delimiter = self.Params['delimiter']
read_1_seqs_fp = self.Params["read_1_seqs_fp"]
assert read_1_seqs_fp, "Must provide read_1_seqs_fp when calling an RtaxTaxonAssigner."
# following params may all be null
read_2_seqs_fp = self.Params["read_2_seqs_fp"]
single_ok = self.Params["single_ok"]
no_single_ok_generic = self.Params["no_single_ok_generic"]
header_id_regex = self.Params["header_id_regex"]
assert header_id_regex, (
"Must not provide empty header_id_regex when calling an RtaxTaxonAssigner; leave unset"
"to use default if in doubt."
)
read_id_regex = self.Params["read_id_regex"]
amplicon_id_regex = self.Params["amplicon_id_regex"]
# seq_file = open(seq_path, 'r')
results = rtax.assign_taxonomy(
seq_path,
reference_sequences_fp,
id_to_taxonomy_fp,
read_1_seqs_fp,
read_2_seqs_fp,
single_ok=single_ok,
no_single_ok_generic=no_single_ok_generic,
header_id_regex=header_id_regex,
read_id_regex=read_id_regex,
amplicon_id_regex=amplicon_id_regex,
output_fp=result_path,
log_path=log_path,
base_tmp_dir=get_qiime_temp_dir(),
)
return results
def test_single_end_classification(self):
self._paths_to_clean_up += cleanAll(self.read_1_seqs_fp)
result = assign_taxonomy(self.input_seqs_fp,
self.reference_seqs_fp,
self.id_to_taxonomy_fp,
self.read_1_seqs_fp,
None,
header_id_regex="\\S+\\s+(\\S+?)\/")
self.assertEqual(result, rtax_expected_result_single)
def test_paired_end_classification_with_fallback(self):
self._paths_to_clean_up += cleanAll(self.read_1_seqs_fp)
self._paths_to_clean_up += cleanAll(self.read_2_seqs_fp)
result = assign_taxonomy(self.input_seqs_fp,
self.reference_seqs_fp,
self.id_to_taxonomy_fp,
self.read_1_seqs_fp,
self.read_2_seqs_fp,
single_ok=True,
header_id_regex="\\S+\\s+(\\S+?)\/")
self.assertEqual(result, rtax_expected_result_paired_with_fallback)
def test_single_end_classification(self):
self._paths_to_clean_up += cleanAll(self.read_1_seqs_fp)
result = assign_taxonomy(self.input_seqs_fp, self.reference_seqs_fp, self.id_to_taxonomy_fp, self.read_1_seqs_fp, None ,header_id_regex="\\S+\\s+(\\S+?)\/")
self.assertEqual(result, rtax_expected_result_single)
def test_paired_end_classification_with_fallback(self):
self._paths_to_clean_up += cleanAll(self.read_1_seqs_fp)
self._paths_to_clean_up += cleanAll(self.read_2_seqs_fp)
result = assign_taxonomy(self.input_seqs_fp, self.reference_seqs_fp, self.id_to_taxonomy_fp, self.read_1_seqs_fp, self.read_2_seqs_fp,single_ok=True,header_id_regex="\\S+\\s+(\\S+?)\/")
self.assertEqual(result, rtax_expected_result_paired_with_fallback)
