def test_validate_taxrank_taxmap_taxtree_fail2(self):
# test for case which taxmap file has more taxids
# than is present in the taxonomy tree.
input_taxrank = _prep_taxranks(self.taxranks)
input_taxmap = _prep_taxmap(self.taxmap)
self.assertRaises(ValueError, _validate_taxrank_taxmap_taxtree,
input_taxrank, input_taxmap, self.taxtree)
def test_validate_taxrank_taxmap_taxtree_pass(self):
# all files should meet the criteria and return None
input_taxrank = _prep_taxranks(self.taxranks)
input_taxmap = _prep_taxmap(self.taxmap2)
exp = _validate_taxrank_taxmap_taxtree(input_taxrank, input_taxmap,
self.taxtree)
self.assertEqual(None, exp)
def test_prep_taxranks(self):
obs_taxranks = _prep_taxranks(self.taxranks)
obs_taxranks.sort_index(inplace=True)
dd = {'taxid': ['2', '11084', '42913', '42914', '42915',
'11089', '24228', '24229', '42916', '42917'],
'taxid_taxonomy': ['Archaea', 'Aenigmarchaeota',
'Aenigmarchaeia', 'Aenigmarchaeales',
'Candidatus_Aenigmarchaeum',
'Deep_Sea_Euryarchaeotic_Group(DSEG)',
'Altiarchaeota', 'Altiarchaeia',
'Altiarchaeales', 'Altiarchaeaceae'],
'taxrank': ['domain', 'phylum', 'class', 'order', 'genus',
'class', 'phylum', 'class', 'order', 'family']}
exp_taxranks = pd.DataFrame(dd)
exp_taxranks.set_index('taxid', inplace=True)
exp_taxranks.sort_index(inplace=True)
assert_frame_equal(obs_taxranks, exp_taxranks)
def test_compile_taxonomy_output_default(self):
input_taxrank = _prep_taxranks(self.taxranks)
silva_tax = _build_base_silva_taxonomy(self.taxtree, input_taxrank,
ALLOWED_RANKS,
rank_propagation=True)
input_taxmap = _prep_taxmap(self.taxmap2)
updated_taxmap = pd.merge(input_taxmap, silva_tax, left_on='taxid',
right_index=True)
obs_6r_tax = _compile_taxonomy_output(updated_taxmap,
ranks=DEFAULT_RANKS,
include_species_labels=False)
obs_6r_tax.sort_index(inplace=True)
# expected 6-rank taxonomy
t1 = ("d__Archaea; p__Aenigmarchaeota; c__Aenigmarchaeia; "
"o__Aenigmarchaeales; f__Aenigmarchaeales; "
"g__Candidatus_Aenigmarchaeum")
exp_6r_tax = pd.Series(t1, index=['AB600437.1.1389'])
exp_6r_tax.rename('Taxon', inplace=True)
exp_6r_tax.index.name = 'Feature ID'
exp_6r_tax.sort_index(inplace=True)
assert_series_equal(obs_6r_tax, exp_6r_tax)
def test_validate_taxrank_taxmap_taxtree_fail(self):
# test for missing taxid in tree file
input_taxrank = _prep_taxranks(self.taxranks)
input_taxmap = _prep_taxmap(self.taxmap2)
self.assertRaises(ValueError, _validate_taxrank_taxmap_taxtree,
input_taxrank, input_taxmap, self.taxtree2)
def test_build_base_silva_taxonomy(self):
input_taxranks = _prep_taxranks(self.taxranks)
obs_taxonomy = _build_base_silva_taxonomy(self.taxtree, input_taxranks,
ALLOWED_RANKS)
obs_taxonomy.sort_index(inplace=True)
tid = {
'taxid': [
'2', '11084', '42913', '42914', '42915', '11089', '24228',
'24229', '42916', '42917'
],
'd__': [
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea',
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea'
],
'sk__': [
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea',
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea'
],
'k__': [
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea',
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea'
],
'ks__': [
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea',
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea'
],
'sp__': [
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea',
'Archaea', 'Archaea', 'Archaea', 'Archaea', 'Archaea'
],
'p__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Aenigmarchaeota', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Altiarchaeota', 'Altiarchaeota', 'Altiarchaeota',
'Altiarchaeota'
],
'ps__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Aenigmarchaeota', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Altiarchaeota', 'Altiarchaeota', 'Altiarchaeota',
'Altiarchaeota'
],
'pi__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Aenigmarchaeota', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Altiarchaeota', 'Altiarchaeota', 'Altiarchaeota',
'Altiarchaeota'
],
'sc__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Aenigmarchaeota', 'Aenigmarchaeota', 'Aenigmarchaeota',
'Altiarchaeota', 'Altiarchaeota', 'Altiarchaeota',
'Altiarchaeota'
],
'c__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeia', 'Aenigmarchaeia',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeia', 'Altiarchaeia'
],
'cs__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeia', 'Aenigmarchaeia',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeia', 'Altiarchaeia'
],
'ci__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeia', 'Aenigmarchaeia',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeia', 'Altiarchaeia'
],
'so__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeia', 'Aenigmarchaeia',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeia', 'Altiarchaeia'
],
'o__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeales', 'Aenigmarchaeales',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeales', 'Altiarchaeales'
],
'os__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeales', 'Aenigmarchaeales',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeales', 'Altiarchaeales'
],
'sf__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeales', 'Aenigmarchaeales',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeales', 'Altiarchaeales'
],
'f__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeales', 'Aenigmarchaeales',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeales', 'Altiarchaeaceae'
],
'fs__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeales', 'Aenigmarchaeales',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeales', 'Altiarchaeaceae'
],
'g__': [
'Archaea', 'Aenigmarchaeota', 'Aenigmarchaeia',
'Aenigmarchaeales', 'Candidatus_Aenigmarchaeum',
'Deep_Sea_Euryarchaeotic_Group(DSEG)', 'Altiarchaeota',
'Altiarchaeia', 'Altiarchaeales', 'Altiarchaeaceae'
]
}
exp_taxonomy = pd.DataFrame(tid)
exp_taxonomy.set_index('taxid', inplace=True)
exp_taxonomy.sort_index(inplace=True)
assert_frame_equal(obs_taxonomy, exp_taxonomy)