def test_taxonomy_add_lineage1():
t = Taxonomy()
lineage = {
'{}_'.format(rank): '{}{}'.format(rank, 1)
for rank in TAXON_RANKS
}
t.add_lineage(**lineage)
def test_taxonomy_add_lineage2():
t = Taxonomy()
lineage = {
'{}_'.format(rank): '{}{}'.format(rank, 1)
for rank in TAXON_RANKS
}
lineage['subspecies'] = 'subspecies1'
t.add_lineage(**lineage)
def taxonomy():
t = Taxonomy()
for x in range(10):
lineage = {
'{}_'.format(rank): '{}{}'.format(rank, x)
for rank in TAXON_RANKS
}
t.add_lineage(**lineage)
return t
def test_taxonomy_add_lineage3():
t = Taxonomy()
lineage = {
'{}_'.format(rank): '{}{}'.format(rank, 1)
for rank in TAXON_RANKS
}
lineage['subspecies'] = 'subspecies1'
lineage['cultivar'] = 'cultivar1'
with pytest.raises(ValueError):
t.add_lineage(**lineage)
def test_taxonomy_serialise_msgpack(taxonomy, tmpdir):
file_name = tmpdir.join('tx.msgpack').strpath
taxonomy.save_data(file_name)
tx2 = Taxonomy(file_name)
assert taxonomy._taxa == tx2._taxa
def test_taxonomy_find_by_name_rank4():
taxonomy = Taxonomy()
taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p1',
order='test_o')
taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p2',
order='test_o')
taxon_id = taxonomy.find_by_name('test_o', rank='order', strict=False)
assert len(taxon_id) == 2
def test_taxonomy_find_by_name_rank3():
taxonomy = Taxonomy()
taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p1',
order='test_o')
taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p2',
order='test_o')
with pytest.raises(KeyError):
taxon_id = taxonomy.find_by_name('test_o', rank='order')
def test_taxonomy_add_taxon2():
taxonomy = Taxonomy()
taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p1',
order='test_o')
taxon_id = taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p2',
order='test_o')
with pytest.raises(KeyError):
taxonomy.add_taxon('test_o', rank='order', parent_id=1)
def test_taxonomy_add_taxon1():
taxonomy = Taxonomy()
taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p1',
order='test_o')
taxon_id = taxonomy.add_lineage(superkingdom='test_k',
phylum='test_p2',
order='test_o')
parent_id = taxonomy.find_by_name('test_p2', rank='phylum')
assert taxon_id == taxonomy.add_taxon('test_o',
rank='order',
parent_id=parent_id)
def test_taxonomy_parse_gtdb_lineage():
lineage = 'd__kingdom1;p__phylum1;o__order1;c__class1;f__family1;g__genus1;s__species1'
result_dict = Taxonomy.parse_gtdb_lineage(lineage)
assert result_dict == {
'superkingdom': 'kingdom1',
'phylum': 'phylum1',
'order': 'order1',
'class': 'class1',
'family': 'family1',
'genus': 'genus1',
'species': 'species1'
}
def test_read_ncbi_taxonomy(taxonomy_files):
taxonomy = Taxonomy()
taxonomy.read_from_ncbi_dump(*taxonomy_files)
assert taxonomy.find_by_name('prevotella', rank='genus') == 838