def get_gse_gsm_dict(fn, gsm_raw_dict, gsm_dict):
gsms = gsm_raw_dict['gsm']
# gse_gsms_dict
fn_pkl = f'{fn}.pkl'
if os.path.isfile(fn_pkl):
gse_gsms_dict = load_table_dict_pkl(fn_pkl)
else:
gses = set()
for gses_raw in gsm_raw_dict["series_id"]:
gses_curr = gses_raw.split(',')
for gse in gses_curr:
gses.add(gse)
gses = list(gses)
gse_gsms_dict = {}
for gse in gses:
gse_gsms_dict[gse] = []
for gsm in gsms:
gses_raw = gsm_dict["series_id"][gsm]
gses_curr = gses_raw.split(',')
for gse in gses_curr:
gse_gsms_dict[gse].append(gsm)
save_table_dict_pkl(fn_pkl, gse_gsms_dict)
return gse_gsms_dict
def get_raw_dict(fn):
fn_xlsx = f'{fn}.xlsx'
fn_pkl = f'{fn}.pkl'
if os.path.isfile(fn_pkl):
gsm_raw_dict = load_table_dict_pkl(fn_pkl)
else:
gsm_raw_dict = load_table_dict_xlsx(fn_xlsx)
save_table_dict_pkl(fn_pkl, gsm_raw_dict)
return gsm_raw_dict
def get_gsm_dict(fn, gsm_raw_dict):
gsms = gsm_raw_dict['gsm']
fn_pkl = f'{fn}.pkl'
if os.path.isfile(fn_pkl):
gsm_dict = load_table_dict_pkl(fn_pkl)
else:
gsm_dict = {}
for key in tqdm(gsm_raw_dict, desc='gsm_dict processing'):
gsm_dict[key] = {}
for index, gsm in enumerate(gsms):
gsm_dict[key][gsm] = gsm_raw_dict[key][index]
save_table_dict_pkl(fn_pkl, gsm_dict)
return gsm_dict
GPL = '21145'
suffix = '22_09_20'
gsm_key = 'gsm'
gse_key = 'series_id'
source_key = 'source_name_ch1'
characteristics_key = 'characteristics_ch1'
fn_xlsx = f'{get_data_path()}/GPL{GPL}/GPL{GPL}_gsm_table_{suffix}.xlsx'
fn_pkl = f'{get_data_path()}/GPL{GPL}/GPL{GPL}_gsm_table_{suffix}.pkl'
if os.path.isfile(fn_pkl):
gsm_raw_dict = load_table_dict_pkl(fn_pkl)
else:
gsm_raw_dict = load_table_dict_xlsx(fn_xlsx)
save_table_dict_pkl(fn_pkl, gsm_raw_dict)
gsms = gsm_raw_dict[gsm_key]
fn = f'{get_data_path()}/GPL{GPL}/bad_words.txt'
f = open(fn)
bad_words = set(f.read().splitlines())
f.close()
fn = f'{get_data_path()}/GPL{GPL}/target_chs.txt'
f = open(fn)
target_chs = set(f.read().splitlines())
f.close()
# data_dict_passed
fn_pkl = f'{get_data_path()}/GPL{GPL}/GPL{GPL}_gsm_dict.pkl'