def test_tabix_db(self):
#init
self.init_test('test_tabix_database')
ucsc_ctrller = self.__create_ucsc_ctrller_instance()
test_file = os.path.join(self.data_dir,
'test_tabix_database.txt')
working_file = os.path.join(self.working_dir,
'test_tabix_database.txt')
expected_out_file = os.path.join(self.working_dir,
'test_tabix_database.txt.gz')
self.copy_file(test_file, working_file)
#test if the 'tabix' files are produced
out_file = ucsc_ctrller.tabix_db(working_file)
self.assertEqual(expected_out_file, out_file,
"Tabix doesn't work correctly")
self.assertTrue(os.path.exists(out_file),
"Tabix doesn't work correctly")
self.assertTrue(os.path.exists(out_file+'.tbi'),
"Tabix doesn't work correctly")
#test if it is readable
ucsc_reader = UcscReader()
ucsc_reader.read(out_file)
readable = False
for rec in ucsc_reader.fetch_snps('chr3', 108572604, 108572605):
readable = True
self.assertEqual(rec.start_pos,
'108572604',
"Database tabixing doesn't work correctly")
break
self.assertTrue(readable,
"Tabixed ucsc database is not readable")
class Referer(Configure):
"""To connect to reference database"""
def __init__(self):
Configure.__init__(self)
def load_config(self):
Configure.load_config(self)
self.__ucsc_reader = UcscReader()
self.__ucsc_reader.read(self.config_values[combivep_settings.LATEST_UCSC_FILE_NAME])
self.__ljb_reader = LjbReader()
self.__ljb_reader.read(self.config_values[combivep_settings.LATEST_LJB_FILE_PREFIX] + '.txt.gz')
def validate_snp(self, chrom, pos, ref, alt):
"""
This function checks if a given snp is valid by referencing
with UCSC database
The inputs of this function are in string format except "pos", which is integer.
"chrom" can be either in format "chr1" or "1"
"pos" is 1-based index
return True if the snp is presented in UCSC reference database
and False otherwise.
"""
for rec in self.__ucsc_reader.fetch_array_snps(chrom, int(pos)-1, int(pos)):
if rec[combivep_settings.UCSC_0_INDEX_REF] != ref:
continue
if ref == alt:
continue
ucsc_alts = rec[combivep_settings.UCSC_0_INDEX_OBSERVED].split('/')
for ucsc_alt in ucsc_alts:
if ucsc_alt == alt:
return True
return False
def get_scores(self, chrom, pos, ref, alt):
"""
This function returns precomputed prediction scores from LJB database
The inputs of this function are in string format except "pos", which is integer.
"chrom" can be either in format "chr1" or "1"
"pos" is 1-based index
return hash scores if the snp is precomputed and None otherwise
"""
return self.__ljb_reader.get_scores(chrom, pos, ref, alt)
def test_tabix_database(self):
#init
self.init_test('test_tabix_database')
self.init_ucsc_controller_instance()
test_file = os.path.join(self.data_dir, 'test_tabix_database.txt')
working_file = os.path.join(self.working_dir, 'test_tabix_database.txt')
expected_out_file = os.path.join(self.working_dir, 'test_tabix_database.txt.gz')
self.copy_file(test_file, working_file)
#test if the 'tabix' files are produced
out_file = self.__ucsc_controller.tabix_database(working_file)
self.assertEqual(expected_out_file, out_file, "Tabix doesn't work correctly")
self.assertTrue(os.path.exists(out_file), "Tabix doesn't work correctly")
self.assertTrue(os.path.exists(out_file+'.tbi'), "Tabix doesn't work correctly")
#test if it is readable
ucsc_reader = UcscReader()
ucsc_reader.read(out_file)
readable = False
for rec in ucsc_reader.fetch_array_snps('chr3', 108572604, 108572605):
readable = True
self.assertEqual(rec[combivep_settings.UCSC_0_INDEX_START_POS], '108572604', "Database tabixing doesn't work correctly")
break
self.assertTrue(readable, "Tabixed ucsc database is not readable")