def read_GWAS(file, chsize):
logger.info('initiating read_file at {}'.format(get_time()))
for chnk_num, chunk in enumerate(
pd.read_csv(file, chunksize=chsize, header=None, sep='\t')):
print("entering file {}: chunk: {}".format(file, chnk_num))
if chnk_num < 1:
check_headers(chunk)
iterate_file(chnk_num, GWAS_set=chunk.itertuples(index=True))
def iterate_file(chnk_num, GWAS_set, SNP='SNP', i=0):
logger.info('initiating iterate_file at {}'.format(get_time()))
try:
for i, _, _, _, _, _, _, FRQ, effect, SE, P, *V in GWAS_set:
stat_test.init_compute(cur_study, FRQ, effect, SE, P, V, chnk_num)
except:
logger.error(sys.exc_info())
logger.error(
'during parseRsmerge at chunk {}, at SNP {}, at row {}'.format(
chnk_num, SNP, i))
def init_file_reader():
global cur_study
logging.basicConfig(filename='../SNP.log',
filemode='a',
level=logging.DEBUG)
logger.info('initiating SNP conformer at {}'.format(get_time()))
# list containing all the 'study' object with meta data as attributes
files = read_meta()
for n, study in enumerate(files):
path = study.get('path')
read_GWAS(path, 5000)
def iterate_file(chnk_num, GWAS_chnk, GWAS_set, SNP='SNP', i=0):
logger.info('initiating iterate_file at {}'.format(get_time()))
try:
for i, _, _, SNP, BP, _, _, A1, A2, FRQ, effect, SE, P in GWAS_chnk:
liftover.liftover_check(SNP, GWAS_set, chnk_num)
reference_check.reference_check(SNP, BP, A1, A2, FRQ, effect,
GWAS_set, chnk_num)
except:
logger.error(sys.exc_info())
logger.error(
'during parseRsmerge at chunk {}, at SNP {}, at row {}'.format(
chnk_num, SNP, i))
def init_file_reader():
logging.basicConfig(filename='../SNP.log',
filemode='a',
level=logging.DEBUG)
logger.info('initiating SNP conformer at {}'.format(get_time()))
# initiate collection object for querying the database
liftover.init_collection()
reference_check.init_collection()
# call read file, to initiate read, SNP liftover and reference check procedure
read_file(GWAS, 5000)
# get stats at end of procedure for future reference
get_stats()
def liftover_check(SNP, gwas_set, chk_num):
global collection
global liftover_sum, nohit_sum
logger.info('initiating SNP conformer at {}'.format(get_time()))
try:
match = collection.fetch('rs_low', SNP)
if match:
entry = match[0]
rs_cur = entry.get('rs_cur')
if rs_cur != SNP:
gwas_set.replace(SNP, rs_cur, inplace=True)
liftover_sum += 1
else:
# log or not?
pass
else:
nohit_sum += 1
except:
logger.error(sys.exc_info())
logger.error('during liftover at chunk {}, at row {}'.format(
chk_num, SNP))
def init_collection():
global collection
logger.info('initiating SNP1000 collection at {}'.format(get_time()))
collection = DbQuerier('../DB/GWAS').db_collection('SNP1000')