def main():
# example: python bart_plot.py user_key
# print (sys.argv)
# get argv
script_name = sys.argv[0]
user_key = sys.argv[1] # user_key is needed
import do_process
user_data = do_process.get_user_data(user_key)
user_path = user_data['user_path']
bart_result_file = ''
bart_auc_file = ''
auc_result_dict = {}
bart_title = [
'tf_name', 'tf_score', 'p_value', 'z_score', 'max_auc', 'r_rank'
]
bart_output_dir = os.path.join(user_path, 'download/bart_output')
bart_auc_ext = '_auc.txt'
for root, dirs, files in os.walk(bart_output_dir):
for bart_file in files:
if bart_auc_ext in bart_file:
bart_auc_file = os.path.join(root, bart_file)
user_file_name = bart_file.strip(bart_auc_ext)
auc_result_dict[user_file_name] = {}
auc_result_dict[user_file_name]['auc'] = bart_auc_file
bart_res_ext = '_bart_results.txt'
for root, dirs, files in os.walk(bart_output_dir):
for bart_file in files:
if bart_res_ext in bart_file:
bart_result_file = os.path.join(root, bart_file)
user_file_name = bart_file.strip(bart_res_ext)
if user_file_name not in auc_result_dict:
auc_result_dict[user_file_name] = {}
# something definitely went wrong if no _auc.txt files!
# the plot can not be plotted!!
# only the statistics will be shown!
auc_result_dict[user_file_name]['res'] = bart_result_file
for user_file_name, bart_files in auc_result_dict.items():
bart_auc_file = bart_files['auc']
bart_result_file = bart_files['res']
AUCs = get_AUCs(bart_auc_file)
bart_df = pd.read_csv(bart_result_file,
sep='\t',
names=bart_title[1:],
index_col=0,
skiprows=1)
plot_top_tf(bart_df, bart_output_dir, AUCs)
def is_bart_done(user_path):
user_key = os.path.basename(user_path)
import do_process
user_data = do_process.get_user_data(user_key)
for user_file in user_data['files']:
uploaded_file = os.path.basename(user_file).split('.')[
0] # path/to/user/upload/filename.bam(txt)
res_file_path = os.path.join(
user_path,
'download/bart_output/' + uploaded_file + '_bart_results.txt'
) # path/to/user/download/filename_bart_results.txt
if not os.path.exists(res_file_path):
return False
return True
def send_mail():
usercase_dir = os.path.join(PROJECT_DIR, 'usercase')
for userkey in os.listdir(usercase_dir):
user_path = os.path.join(usercase_dir, userkey)
# iterate the user path
if not os.path.isdir(user_path):
continue
user_data = do_process.get_user_data(userkey)
if not user_data:
continue
if 'status' not in user_data: # not yet processing
continue
if 'user_email' in user_data:
if user_data['user_email'] == '':
continue
if user_data['status'] == 'Done':
logger.info('Send Email: send success e-mail to user {}'.format(userkey))
send_flag, send_msg = utils.send_user_key(user_data['user_email'], userkey, user_data['status'])
if send_flag:
logger.info('Send Email: ' + send_msg)
else:
logger.error('Send Email: ' + send_msg)
user_data['status'] = 'Sent'
do_process.init_user_config(user_path, user_data)
if user_data['status'] == 'Error':
logger.info('Job Finish: send error e-mail to user {}'.format(userkey))
send_flag, send_msg = utils.send_user_key(user_data['user_email'], userkey, user_data['status'])
if send_flag:
logger.info('Send Email: ' + send_msg)
else:
logger.error('Send Email: ' + send_msg)
user_data['status'] = 'Sent'
do_process.init_user_config(user_path, user_data)
def main():
# example: python exe_mb_pipline.py 3 user_key True/False
# print (sys.argv)
# get argv
script_name = sys.argv[0]
repeat_times = int(sys.argv[1])
user_key = sys.argv[2]
bart_flag = bool(sys.argv[3])
import do_process
user_data = do_process.get_user_data(user_key)
user_path = user_data['user_path']
files = user_data['files']
err_msg = ""
for i in range(repeat_times):
marge_output_dir = os.path.join(user_path, 'marge_{}'.format(i))
if init_marge(marge_output_dir):
config_marge(user_data, marge_output_dir)
subprocess.call(["snakemake", "-n"], stdout=subprocess.PIPE, cwd=marge_output_dir)
else:
err_msg += "Error in init marge NO.%d \n" % (i+1)
import multiprocessing
pool = multiprocessing.Pool(processes=MARGE_CORE)
for i in range(repeat_times):
marge_output_dir = os.path.join(user_path, 'marge_{}'.format(i))
pool.apply_async(exe_marge, args=(marge_output_dir, ))
pool.close()
pool.join()
# get marge output
auc_scores = []
auc_files = []
import re
pattern = r"\d+\.?\d*" # integer or float
# find AUC score
for i in range(repeat_times):
marge_output_dir = os.path.join(user_path, 'marge_{}'.format(i))
for upload_file in files:
filename = os.path.basename(upload_file)
filename, file_ext = os.path.splitext(filename)
regression_score_file = os.path.join(marge_output_dir, 'margeoutput/regression/{}_target_regressionInfo.txt'.format(filename))
if not os.path.exists(regression_score_file):
err_msg += "File not exists: %s" % (regression_score_file)
continue
with open(regression_score_file, 'r') as fopen:
for line in fopen:
if 'AUC = ' in line:
score = re.findall(pattern, line)[0]
auc_scores.append(float(score))
auc_files.append(marge_output_dir)
# find max AUC score
max_auc = max(auc_scores)
max_index = -1
for i in range(len(auc_scores)):
if auc_scores[i] == max_auc:
max_index = i
break
# find max AUC folder & change it to folder /marge_data
if max_index == -1:
err_msg += "Severe error in marge process!!\n"
else:
auc_file = auc_files[i]
os.rename(auc_file, os.path.join(user_path, 'marge_data'))
# if bart
print (user_data)
if bart_flag:
exe_bart_geneset(user_data)
print (err_msg)
def get_result():
user_key = request.args['user_key']
user_data = do_process.get_user_data(user_key)
results = do_process.generate_results(user_data)
return render_template('result_demonstration.html', results=results)