####################################### TRAINING ##########################################
Open Train directory. Extract TrainDataPSSM.zip here. Show in the Train directory we have directory TrainDataPSSM and file adaptor_train.py

+ Step 1: Choose seed for random number using to devide training dataset (include train and dev test) into 5 folds. For example, if random number = 53: 
Open adaptor_train.py
split_seed = 53

+ Step 2: Using one fold for validation set, other folds for train set. Repeat 5 times. Report results for each time and for ensemble methods (mean probabilities of 5 models on test sample).  
$ cd Train/
$ nohup python3 adaptor_train.py > log_seed_53_train.txt & 
It takes about two hours for training. 

Results:
+ File log: log_seed_53.txt which contains 
	- list of train_losses (by epoch), train_lst_acc, train_lst_sensitivity, train_lst_specificity, train_lst_mcc, train_lst_auc, 
 	- Val confusion matrix, val_losses, val_lst_acc, val_lst_sensitivity, val_lst_specificity, val_lst_mcc, val_lst_auc
	- Model architecture summary
	 
+ The training process generate automatically "model_seed53" directory which contains:
	- Model files of 5 folds, for example: model_seed53/adaptor_3_7.pkl, 3 is the fold and 7 is the best epoch.
	- logfile_loss_model_3_7.csv contains epoch,train loss,train acc,train_sens,train_spec,train_mcc,train_auc,val loss,val acc,val_sens,val_spec,val_mcc,val_auc
	

####################################### TESTING ##########################################
Open Test directory. Extract TestDataPSSM.zip here. Show in the Test directory we have directory TestDataPSSM and file adaptor_test.py
$ cd Test
$ nohup python3 adaptor_test.py > log_seed_53_test.txt & 
Results: 
+ File log: log_seed_53.txt which contains testing result of each fold and ensemble method: Test loss, Test acc, Test confusion matrix, Test sensitivity, Test specificity, Test mcc, Test auc
+ "model_seed61" directory which contains "test_result.csv" which contains results on test set: model_fn,AUC,MCC,Accuracy,Sensitivity,Specificity
 
