liuze-nwafu/m6abrp

A machine learning-based model for predicting YTHDF2 binding regions in mRNAs via sequence-based properties

m6ABRP

A machine learning-based model for predicting YTHDF2 binding regions in mRNAs via sequence-based properties.

Dependencies:

 1--sys  

 2--numpy  

 3--sklearn 

 4--joblib  

m6ABRP is implemented using Python2.7.

feature_selection.py

This script can be used for feature encoding automatically.

model_training.py

Usage: python model_training.py training_positive_dataset training_negative_dataset model_file scale_file pca_file
This script is used to train the m6ABRP tool.
Outputs:

 1--a model file, model.pkl, which can be directly used for prediction. 
 
 2--a normalized file, normalization.pkl, which can be used to normalized the input data. 
 
 3--a pca model file, pca.pkl, which can be used to generate the principal components.  

model_indepedent_testing.py

Usage: python model_indepedent_testing.py test_positive_dataset test_negative_dataset model_file scale_file pca_file
This script is used to evaluate the performance of m6ABRP on the indepedent testing dataset.

The model_file, scale_file and pca_file generated in the training process must be involved.
Outputs:

 1--m6ABRP_score.txt, which consists of the prediction score of each sample.  
 
 2--test_label.txt, which includes the label of each sample.