def test_taxonomy_from_list_summarize_group_simple_cached_model(self):
# NOTE: do not delete this test when converting _alt methods to non
# _alt methods
with patch(
'microsetta_public_api.repo._taxonomy_repo.TaxonomyRepo.'
'tables',
new_callable=PropertyMock) as mock_tables:
mock_tables.return_value = {
'some-table': {
'table': self.table,
'feature-data-taxonomy': self.taxonomy_df,
'model': TaxonomyModel(self.table, self.taxonomy_df)
},
}
response, code = _summarize_group(
self.post_body['sample_ids'],
"some-table",
taxonomy_repo=TaxonomyRepo(),
)
self.assertEqual(code, 200)
exp_keys = [
'taxonomy', 'features', 'feature_values', 'feature_variances'
]
obs = json.loads(response)
self.assertCountEqual(exp_keys, obs.keys())
self.assertEqual(
'((((feature-1,((feature-2)e)d)c)b,'
'(((feature-3)h)g)f)a);', obs['taxonomy'])
self.assertListEqual(['feature-1', 'feature-2', 'feature-3'],
obs['features'])
assert_allclose([1. / 10, 6. / 10, 3. / 10], obs['feature_values'])
assert_allclose([0, 0, 0], obs['feature_variances'])
def model(self, table_name):
model = self._get_resource(table_name, component='model')
if model is None:
table = self.table(table_name)
features = self.feature_data_taxonomy(table_name)
variances = self.variances(table_name)
model = TaxonomyModel(table, features, variances)
return model
def setUpClass(cls):
cls.post_body = {
'sample_ids': [
'sample-1',
'sample-2',
]
}
cls.table = biom.Table(np.array([[0, 1, 2], [2, 4, 6], [3, 0, 1]]),
['feature-1', 'feature-2', 'feature-3'],
['sample-1', 'sample-2', 'sample-3'])
cls.taxonomy_df = pd.DataFrame(
[['feature-1', 'a; b; c', 0.123],
['feature-2', 'a; b; c; d; e', 0.345],
['feature-3', 'a; f; g; h', 0.678]],
columns=['Feature ID', 'Taxon', 'Confidence'])
cls.taxonomy_df.set_index('Feature ID', inplace=True)
# variances
cls.table_vars = biom.Table(
np.array([[0, 1, 2], [2, 4, 6],
[3, 0, 1]]), ['feature-1', 'feature-2', 'feature-3'],
['sample-1', 'sample-2', 'sample-3'])
cls.taxonomy_model = TaxonomyModel(cls.table, cls.taxonomy_df,
cls.table_vars)
def _transform_single_table(dict_, resource_name):
taxonomy = {
'feature-data-taxonomy': dict_.pop('feature-data-taxonomy', None)
}
supported_table_types = {'qza', 'biom'}
table_type = dict_.get('table-format', 'qza')
if table_type not in supported_table_types:
raise ValueError(f"'table-format'={table_type} not in supported table "
f"types: {supported_table_types}.")
_validate_dict_of_paths(
dict_,
resource_name,
allow_none=True,
required_fields=['table'],
non_ext_entries=['q2-type', 'table-format', 'cache-taxonomy'],
allow_extras=True,
extensions=['.' + table_type])
_validate_dict_of_paths(
taxonomy,
resource_name,
allow_none=True,
)
if taxonomy['feature-data-taxonomy'] is not None:
dict_.update(taxonomy)
semantic_types = {
'feature-data-taxonomy': FeatureData[Taxonomy],
}
biom_kws = set()
if table_type == 'qza':
semantic_types.update({
'table':
dict_.get('table-type', FeatureTable[Frequency]),
'variances':
FeatureTable[Frequency],
})
elif table_type == 'biom':
biom_kws.update({'table', 'variances'})
else:
# shouldn't happen because error check earlier but seems better than
# silently ignoring....
raise ValueError(f"'table-type'={table_type} not in supported table "
f"types: {supported_table_types}.")
views = {
'table': biom.Table,
'feature-data-taxonomy': pd.DataFrame,
'variances': biom.Table,
}
new_resource = deepcopy(dict_)
for key, value in dict_.items():
if key in semantic_types:
new_resource[key] = _parse_q2_data(
value,
semantic_types[key],
view_type=views.get(key, None),
)
elif key in biom_kws:
new_resource[key] = biom.load_table(value)
cache_taxonomy = new_resource.get('cache-taxonomy', True)
if 'feature-data-taxonomy' in new_resource and cache_taxonomy:
table = new_resource['table']
taxonomy = new_resource['feature-data-taxonomy']
variances = new_resource.get('variances', None)
# rank_level=5 -> genus
model = TaxonomyModel(table, taxonomy, variances, rank_level=5)
new_resource['model'] = model
return new_resource