def test_parse_metadata_state_descriptions(self):
"""parse_metadata_state_descriptions should return correct states from string."""
s = ''
self.assertEqual(parse_metadata_state_descriptions(s), {})
s = 'Study:Twin,Hand,Dog;BodySite:Palm,Stool'
self.assertEqual(
parse_metadata_state_descriptions(s), {
'Study': set(['Twin', 'Hand', 'Dog']),
'BodySite': set(['Palm', 'Stool'])
})
# category names with colons i. e. ontology-derived
s = 'Study:Twin,Hand,Dog;site:UBERON:feces,UBERON:ear canal;'+\
'env_feature:ENVO:farm soil,ENVO:national park'
self.assertEqual(
parse_metadata_state_descriptions(s), {
'Study': set(['Twin', 'Hand', 'Dog']),
'site': set(['UBERON:feces', 'UBERON:ear canal']),
'env_feature': set(['ENVO:farm soil', 'ENVO:national park'])
})
s = "Treatment:A,B,C;env_matter:ENVO:nitsol,ENVO:farm soil;env_biom:"+\
"ENVO:Tropical dry (including Monsoon forests) and woodlands,"+\
"ENVO:Forest: including woodlands;country:GAZ:Persnickety Islands"+\
",St. Kitt's and Nevis"
self.assertEqual(parse_metadata_state_descriptions(s), {"country":
set(["GAZ:Persnickety Islands", "St. Kitt's and Nevis"]),
"env_biom":set(["ENVO:Tropical dry (including Monsoon forests) "+\
"and woodlands", "ENVO:Forest: including woodlands"]), "env_matter":
set(["ENVO:nitsol","ENVO:farm soil"]), 'Treatment':set(["A", "B",
"C"])})
def test_get_sample_ids(self):
"""get_sample_ids should return sample ids matching criteria."""
self.assertEqual(get_sample_ids(self.map_data, self.map_headers,\
parse_metadata_state_descriptions('Study:Twin')), [])
self.assertEqual(get_sample_ids(self.map_data, self.map_headers,\
parse_metadata_state_descriptions('Study:Dog')), ['a','b'])
self.assertEqual(get_sample_ids(self.map_data, self.map_headers,\
parse_metadata_state_descriptions('Study:*,!Dog')), ['c','d','e'])
self.assertEqual(get_sample_ids(self.map_data, self.map_headers,\
parse_metadata_state_descriptions('Study:*,!Dog;BodySite:Stool')), ['e'])
self.assertEqual(get_sample_ids(self.map_data, self.map_headers,\
parse_metadata_state_descriptions('BodySite:Stool')), ['a','b','e'])
def test_parse_metadata_state_descriptions(self):
"""parse_metadata_state_descriptions should return correct states from string."""
s = ""
self.assertEqual(parse_metadata_state_descriptions(s), {})
s = "Study:Twin,Hand,Dog;BodySite:Palm,Stool"
self.assertEqual(
parse_metadata_state_descriptions(s),
{"Study": set(["Twin", "Hand", "Dog"]), "BodySite": set(["Palm", "Stool"])},
)
# category names with colons i. e. ontology-derived
s = "Study:Twin,Hand,Dog;site:UBERON:feces,UBERON:ear canal;" + "env_feature:ENVO:farm soil,ENVO:national park"
self.assertEqual(
parse_metadata_state_descriptions(s),
{
"Study": set(["Twin", "Hand", "Dog"]),
"site": set(["UBERON:feces", "UBERON:ear canal"]),
"env_feature": set(["ENVO:farm soil", "ENVO:national park"]),
},
)
s = (
"Treatment:A,B,C;env_matter:ENVO:nitsol,ENVO:farm soil;env_biom:"
+ "ENVO:Tropical dry (including Monsoon forests) and woodlands,"
+ "ENVO:Forest: including woodlands;country:GAZ:Persnickety Islands"
+ ",St. Kitt's and Nevis"
)
self.assertEqual(
parse_metadata_state_descriptions(s),
{
"country": set(["GAZ:Persnickety Islands", "St. Kitt's and Nevis"]),
"env_biom": set(
[
"ENVO:Tropical dry (including Monsoon forests) " + "and woodlands",
"ENVO:Forest: including woodlands",
]
),
"env_matter": set(["ENVO:nitsol", "ENVO:farm soil"]),
"Treatment": set(["A", "B", "C"]),
},
)
def sample_ids_from_metadata_description(mapping_f,valid_states_str):
""" Given a description of metadata, return the corresponding sample ids
"""
map_data, map_header, map_comments = parse_mapping_file(mapping_f)
valid_states = parse_metadata_state_descriptions(valid_states_str)
sample_ids = get_sample_ids(map_data, map_header, valid_states)
if len(sample_ids)<1:
raise ValueError,"All samples have been filtered out for the criteria"+\
" described in the valid states"
return sample_ids
def sample_ids_from_metadata_description(mapping_f, valid_states_str):
""" Given a description of metadata, return the corresponding sample ids
"""
map_data, map_header, map_comments = parse_mapping_file(mapping_f)
valid_states = parse_metadata_state_descriptions(valid_states_str)
sample_ids = get_sample_ids(map_data, map_header, valid_states)
if len(sample_ids) < 1:
raise ValueError,"All samples have been filtered out for the criteria"+\
" described in the valid states"
return sample_ids
