Toxicity prediction

This repository contains the source code for predicting the toxicity of chemical compounds from based on SMILES strings.

Files

• _Assessment Report

• data.zip includes the pre-processed smiles features. Please uncompress this file.

• toxicity_modul.py contains the necessary functions for toxicity prediction.

• smiles_feature_pre-processing.py generates the smiles features from given dataset.

• _model_wt_training.py runs the traning process of the CNN model with class weighted + toxicity of the other targets as features. python model_wt_training.py

• model_w_training.py runs the traning process of the CNN model with class weighted.
  python model_w_training.py

• model_res_training.py runs the traning process of the CNN model with over-sampled(minority class) dataset . python model_res_training.py

• parameter_tuning.py is script for hyperparameter tuning.

• predict_unknowns.py runs the prediction using the pre-built models.

• folder figures contains pre-generated figures.

• folder pre-built_model contains pre-built CNN models.

• folder pre-computed_results contains the predictive result using pre-built models for the chemical compounds which the toxicity was previously unknown for 12 taegets.