def test_parse_accession_taxa_table5(ncbi_taxa_file):
acc_ids = {'XM_006695435'}
iterator = parse_accession_taxa_table(ncbi_taxa_file,
key=0,
value=1,
acc_ids=acc_ids)
iterator = list(acc_id for acc_id, taxon_id in iterator)
assert list(acc_ids) == iterator
def filter_taxa_command(verbose, table, taxonomy, include_taxon_id,
include_taxon_name, exclude_taxon_id,
exclude_taxon_name, progress, input_file, output_file):
mgkit.logger.config_log(level=logging.DEBUG if verbose else logging.INFO)
LOG.info('Writing to file (%s)',
getattr(output_file, 'name', repr(output_file)))
taxonomy = taxon.Taxonomy(taxonomy)
exclude_ids = validate_taxon_ids(exclude_taxon_id, taxonomy) | \
validate_taxon_names(exclude_taxon_name, taxonomy)
include_ids = validate_taxon_ids(include_taxon_id, taxonomy) | \
validate_taxon_names(include_taxon_name, taxonomy)
if exclude_ids:
LOG.info("Excluding Taxa: %s", exclude_ids)
exclude_func = functools.partial(filter_taxon_by_id_list,
filter_list=exclude_ids,
exclude=True,
func=functools.partial(
taxon.is_ancestor, taxonomy))
exclude_func = memoize(exclude_func)
else:
exclude_func = None
if include_ids:
LOG.info("Only include Taxa: %s", include_ids)
include_func = functools.partial(filter_taxon_by_id_list,
filter_list=include_ids,
exclude=False,
func=functools.partial(
taxon.is_ancestor, taxonomy))
include_func = memoize(include_func)
else:
include_func = None
if table:
iterator = blast.parse_accession_taxa_table(input_file,
key=0,
value=1,
num_lines=None)
if progress:
iterator = tqdm(iterator)
for acc_id, taxon_id in iterator:
if include_func is not None:
if not include_func(taxon_id):
continue
if exclude_func is not None:
if not exclude_func(taxon_id):
continue
output_file.write("{}\t{}\n".format(acc_id,
taxon_id).encode('ascii'))
else:
iterator = gff.parse_gff(input_file)
if progress:
iterator = tqdm(iterator)
for annotation in iterator:
if annotation.taxon_id is None:
continue
if include_func is not None:
if not include_func(annotation.taxon_id):
continue
if exclude_func is not None:
if not exclude_func(annotation.taxon_id):
continue
annotation.to_file(output_file)
def test_parse_accession_taxa_table1(ncbi_taxa_file):
assert next(
parse_accession_taxa_table(ncbi_taxa_file,
key=0,
value=1,
no_zero=False))[0] == 'KH113978'
def test_parse_accession_taxa_table3(ncbi_taxa_file):
assert next(
parse_accession_taxa_table(ncbi_taxa_file,
key=0,
value=1,
no_zero=True))[0] == 'XM_006695435'
def test_parse_accession_taxa_table2(ncbi_taxa_file):
assert next(
parse_accession_taxa_table(ncbi_taxa_file,
key=0,
value=1,
no_zero=False))[1] == 0