def test_get_distinct(self):
"""should return list of strings"""
db = Database(account=account,
release=Release,
species='human',
db_type='variation')
tn, tc = 'variation_feature', 'consequence_types'
expected = set(('3_prime_UTR_variant', 'splice_acceptor_variant',
'5_prime_UTR_variant'))
got = db.getDistinct(tn, tc)
self.assertNotEqual(set(got) & expected, set())
db = Database(account=account,
release=Release,
species='human',
db_type='core')
tn, tc = 'gene', 'biotype'
expected = set([
'protein_coding', 'pseudogene', 'processed_transcript', 'Mt_tRNA',
'Mt_rRNA', 'IG_V_gene', 'IG_J_gene', 'IG_C_gene', 'IG_D_gene',
'miRNA', 'misc_RNA', 'snoRNA', 'snRNA', 'rRNA'
])
got = set(db.getDistinct(tn, tc))
self.assertNotEqual(set(got) & expected, set())
db = Database(account=account, release=Release, db_type='compara')
got = set(db.getDistinct('homology', 'description'))
expected = set([
'gene_split', 'alt_allele', 'other_paralog', 'ortholog_one2many',
'ortholog_one2one', 'within_species_paralog', 'ortholog_many2many'
])
self.assertEqual(len(got & expected), len(expected))
def test_get_distinct(self):
"""should return list of strings"""
db = Database(account=account,
release=Release,
species='human',
db_type='variation')
tn, tc = 'variation_feature', 'consequence_type'
expected = set((('3PRIME_UTR', 'ESSENTIAL_SPLICE_SITE'),
('3PRIME_UTR', 'SPLICE_SITE'),
('5PRIME_UTR', 'ESSENTIAL_SPLICE_SITE')))
self.assertNotEquals(set(db.getDistinct(tn, tc)) & expected, set())
db = Database(account=account,
release=Release,
species='human',
db_type='core')
tn, tc = 'gene', 'biotype'
expected = set([
'protein_coding', 'pseudogene', 'processed_transcript', 'Mt_tRNA',
'Mt_rRNA', 'IG_V_gene', 'IG_J_gene', 'IG_C_gene', 'IG_D_gene',
'miRNA', 'misc_RNA', 'snoRNA', 'snRNA', 'rRNA'
])
got = set(db.getDistinct(tn, tc))
self.assertEquals(len(got & expected), len(expected))
db = Database(account=account, release=Release, db_type='compara')
got = set(db.getDistinct('homology', 'description'))
expected = set([
'apparent_ortholog_one2one', 'between_species_paralog',
'ortholog_many2many', 'ortholog_one2many', 'ortholog_one2one',
'within_species_paralog'
])
self.assertEquals(len(got & expected), len(expected))
def _connect_db(self, db_type):
connection = dict(account=self._account,
release=self.Release,
species=self.Species,
pool_recycle=self._pool_recycle)
if self._core_db is None and db_type == 'core':
self._core_db = Database(db_type='core', **connection)
gen_rel = self.CoreDb.db_name.GeneralRelease
gen_rel = int(re.findall(r'^\d+', str(gen_rel))[0])
self._gen_release = gen_rel
elif self._var_db is None and db_type == 'variation':
self._var_db = Database(db_type='variation', **connection)
elif self._other_db is None and db_type == 'otherfeatures':
self._other_db = Database(db_type='otherfeatures', **connection)
def _connect_db(self):
# TODO can the connection be all done in init?
connection = dict(account=self._account,
release=self.Release,
pool_recycle=self._pool_recycle)
if self._compara_db is None:
self._compara_db = Database(db_type='compara',
division=self.division,
**connection)
def test_get_table_row_counts(self):
"""should return correct row counts for some tables"""
expect = {
'homo_sapiens_core_56_37a.analysis': 57L,
'homo_sapiens_core_56_37a.seq_region': 55604L,
'homo_sapiens_core_56_37a.assembly': 102068L,
'homo_sapiens_core_56_37a.qtl': 0L
}
human = Database(account=account,
release=Release,
species='human',
db_type='core')
table_names = [n.split('.')[1] for n in expect]
got = dict(human.getTablesRowCount(table_names).getRawData())
self.assertEquals(got, expect)
def test_table_has_column(self):
"""return correct values for whether a Table has a column"""
account = get_ensembl_account(release=Release)
var61 = Database(account=account,
release=61,
species='human',
db_type='variation')
var62 = Database(account=account,
release=62,
species='human',
db_type='variation')
self.assertTrue(
var61.tableHasColumn('transcript_variation',
'peptide_allele_string'))
self.assertFalse(
var61.tableHasColumn('transcript_variation', 'pep_allele_string'))
self.assertTrue(
var62.tableHasColumn('transcript_variation', 'pep_allele_string'))
self.assertFalse(
var62.tableHasColumn('transcript_variation',
'peptide_allele_string'))
def test_get_table_row_counts(self):
"""should return correct row counts for some tables"""
expect = {
'homo_sapiens_core_76_38.analysis': 61,
'homo_sapiens_core_76_38.seq_region': 55616,
'homo_sapiens_core_76_38.assembly': 102090
}
human = Database(account=account,
release=Release,
species='human',
db_type='core')
table_names = [n.split('.')[1] for n in expect]
got = dict(human.getTablesRowCount(table_names).getRawData())
for dbname in expect:
self.assertTrue(got[dbname] >= expect[dbname])
def test_connect(self):
human = Database(account=account,
release=Release,
species='human',
db_type='core')
gene = human.getTable('gene')
