def test1(self):
outfiles = taxonomy.get_ncbi_tax_data(dest_dir=outputdir)
names_data, nodes_data = taxonomy.read_bacterial_taxonomy(primary_only=False, **outfiles)
dbname = complete_test_db
if not os.access(dbname, os.F_OK):
log.warning('creating %s' % dbname)
dbname = taxonomy.create_bacterial_taxonomy_db(
names_data=names_data,
nodes_data=nodes_data,
dbname=dbname,
dest_dir=outputdir
)
self.assertTrue(os.access(dbname, os.F_OK))
else:
log.info('%s exists, skipping this test' % dbname)
def testCreateBacterialTaxonomyDb02(self):
outfiles = taxonomy.get_ncbi_tax_data(dest_dir=outputdir)
names_data, nodes_data = taxonomy.read_bacterial_taxonomy(primary_only=False, **outfiles)
dbname = taxonomy.create_bacterial_taxonomy_db(
names_data=itertools.islice(names_data, 0, 1000),
nodes_data=itertools.islice(nodes_data, 0, 1000),
dbname=self.funcname+'.db',
dest_dir=outputdir
)
self.assertTrue(os.access(dbname, os.F_OK))
con = sqlite.connect(dbname)
synonyms = con.cursor().execute('select count(*) from names where is_primary = 0').fetchone()[0]
# nodes.source should be zero by default
sources = con.cursor().execute('select source_id from nodes').fetchall()
self.assertEqual( set(x[0] for x in sources), set([1]) )
def testCreateBacterialTaxonomyDb01(self):
outfiles = taxonomy.get_ncbi_tax_data(dest_dir=outputdir)
names_data, nodes_data = taxonomy.read_bacterial_taxonomy(**outfiles)
dbname = taxonomy.create_bacterial_taxonomy_db(
names_data=itertools.islice(names_data, 0, 1000),
nodes_data=itertools.islice(nodes_data, 0, 1000),
dbname=self.funcname+'.db',
dest_dir=outputdir
)
self.assertTrue(os.access(dbname, os.F_OK))
con = sqlite.connect(dbname)
synonyms = con.cursor().execute('select count(*) from names where is_primary = 0').fetchone()[0]
self.assertEqual(synonyms, 0)
ranks = con.cursor().execute('select rank from nodes group by rank').fetchall()
## rank names should not contain whitespace
self.assertEqual(ranks, [('_'.join(x[0].split()),) for x in ranks])