def read_config(config_path, verbose=False):
"""returns ensembl release, local path, and db specifics from the provided
config path"""
parser = configparser.ConfigParser()
parser.read_file(config_path)
release = parser.get("release", "release")
remote_path = parser.get("remote path", "path")
local_path = parser.get("local path", "path")
local_path = abspath(local_path)
species_dbs = {}
for section in parser.sections():
if section in ("release", "remote path", "local path"):
continue
dbs = [db.strip() for db in parser.get(section, "db").split(",")]
if section == "compara":
species_dbs["compara"] = dbs
continue
# handle synonymns
species = Species.get_species_name(section, level="raise")
for synonym in Species.get_synonymns(species):
species_dbs[synonym] = dbs
return release, remote_path, local_path, species_dbs
def test_get_name_type(self):
"""should return the (latin|common) name given a latin, common or ensembl
db prefix names"""
self.assertEqual(Species.get_species_name("human"), "H**o sapiens")
self.assertEqual(Species.get_species_name("homo_sapiens"),
"H**o sapiens")
self.assertEqual(Species.get_common_name("Mus musculus"), "Mouse")
self.assertEqual(Species.get_common_name("mus_musculus"), "Mouse")
def test_get_ensembl_format(self):
"""should take common or latin names and return the corresponding
ensembl db prefix"""
self.assertEqual(Species.get_ensembl_db_prefix("human"),
"homo_sapiens")
self.assertEqual(Species.get_ensembl_db_prefix("mouse"),
"mus_musculus")
self.assertEqual(Species.get_ensembl_db_prefix("Mus musculus"),
"mus_musculus")
def test_get_synonyms(self):
"""correctly returns synonyms given either a common or latin name"""
expect = ["Canis familiaris", "Canis lupus familiaris"]
got = Species.get_synonymns("Canis familiaris")
self.assertEqual(set(got), set(expect))
got = Species.get_synonymns("Canis lupus familiaris")
self.assertEqual(set(got), set(expect))
got = Species.get_synonymns("dog")
self.assertEqual(set(got), set(expect))
def test_synonyms_work(self):
"""species with synonyms should allow correct lookups"""
self.assertEqual(Species.get_species_name("Canis lupus familiaris"),
"Canis familiaris")
self.assertEqual(Species.get_common_name("Canis lupus familiaris"),
"Dog")
self.assertEqual(
Species.get_ensembl_db_prefix("Canis lupus familiaris"),
"canis_familiaris")
self.assertEqual(Species.get_compara_name("Canis lupus familiaris"),
"Dog")
def read_config(config_path, verbose=False):
"""returns ensembl release, local path, and db specifics from the provided
config path"""
parser = configparser.ConfigParser()
parser.read_file(config_path)
release = parser.get("release", "release")
remote_path = parser.get("remote path", "path")
local_path = parser.get("local path", "path")
local_path = abspath(local_path)
species_dbs = {}
for section in parser.sections():
if section in ("release", "remote path", "local path"):
continue
if section != "compara":
species = Species.get_species_name(section, level="raise")
else:
species = "compara"
dbs = [db.strip() for db in parser.get(section, "db").split(",")]
species_dbs[species] = dbs
if verbose:
click.secho("DOWNLOADING\n ensembl release=%s" % release, fg="green")
click.secho("\n".join([" %s" % d for d in species_dbs]), fg="green")
click.secho("\nWRITING to output path=%s\n" % local_path, fg="green")
return release, remote_path, local_path, species_dbs
def test_add_new_species(self):
"""should correctly add a new species/common combination and infer the
correct ensembl prefix"""
species_name, common_name = "Otolemur garnettii", "Bushbaby"
Species.amend_species(species_name, common_name)
self.assertEqual(Species.get_species_name(species_name), species_name)
self.assertEqual(Species.get_species_name("Bushbaby"), species_name)
self.assertEqual(Species.get_species_name(common_name), species_name)
self.assertEqual(Species.get_common_name(species_name), common_name)
self.assertEqual(Species.get_common_name("Bushbaby"), common_name)
self.assertEqual(Species.get_ensembl_db_prefix("Bushbaby"),
"otolemur_garnettii")
self.assertEqual(Species.get_ensembl_db_prefix(species_name),
"otolemur_garnettii")
self.assertEqual(Species.get_ensembl_db_prefix(common_name),
"otolemur_garnettii")
def test_get_compara_name(self):
"""should correctly form valid names for assignment onto objects"""
self.assertEqual(Species.get_compara_name("pika"), "Pika")
self.assertEqual(Species.get_compara_name("C.elegans"), "Celegans")
self.assertEqual(Species.get_compara_name("Caenorhabditis elegans"),
"Celegans")
def test_amend_existing(self):
"""should correctly amend an existing species"""
species_name = "Ochotona princeps"
common_name1 = "american pika"
common_name2 = "pika"
ensembl_pref = "ochotona_princeps"
Species.amend_species(species_name, common_name1)
self.assertEqual(Species.get_common_name(species_name), common_name1)
Species.amend_species(species_name, common_name2)
self.assertEqual(Species.get_species_name(common_name2), species_name)
self.assertEqual(Species.get_species_name(ensembl_pref), species_name)
self.assertEqual(Species.get_common_name(species_name), common_name2)
self.assertEqual(Species.get_common_name(ensembl_pref), common_name2)
self.assertEqual(Species.get_ensembl_db_prefix(species_name),
ensembl_pref)
self.assertEqual(Species.get_ensembl_db_prefix(common_name2),
ensembl_pref)
def test_species_attr(self):
"""gene has species attribute"""
self.assertEqual(self.brca2.species, "H**o sapiens")
mac = self.macaq.get_gene_by_stableid("ENSMMUG00000008639")
sp_name = Species.get_species_name("macaque")
self.assertEqual(mac.species, sp_name)
def test_to_table(self):
"""returns a table object"""
table = Species.to_table()
self.assertIsInstance(table, Table)
self.assertTrue(table.shape[0] > 20)
self.assertEqual(table.shape[1], 5)