Usage

Environment Configuration

conda create -n bioact python=3.9
conda activate bioact
pip install -r requirements.txt

Model Training

Model training consists of two parts: data preprocessing and model training.

1. Data Preprocessing

The ADR prediction model (HetSia-SafeNet) is trained using 9202 drugs with 18 ADR labels. The raw ADR dataset is stored in ./data_process/ADR_tree_label. You can download the ADR_tree_label folder from this link. Run ./data_process/drug_ac50_cmax_get.ipynb to combine off-target data predicted by PreMOTA and Cmax,free values predicted by the MotifAttnNet model. This will generate the required drug features to construct the training dataset.You can download ADR_multitask_dataset, ADR_multitask_dataset_random_split, and ADR_multitask_dataset_scaffold_split from this link and place them in the ./Data folder.

2. Data Splitting

Run ./Data/data_split.ipynb to split the dataset either randomly or using a scaffold-based approach.

3. Model Training

Run ./train_adr_property_adremb.py to train the model and save the trained model in the ./result_split directory.

property_feature_adr_emb_random: Model trained with randomly split data. property_feature_adr_emb_scaffold: Model trained with scaffold-based split data.

Model Prediction and Visualization

1. Data Preparation for Prediction (Merging Off-Target and Cmax Data)

Run ./data_process/data_pro_dose_study_test_indrugs.ipynb to combine off-target data from PreMOTA and Cmax/free values from MotifAttnNet, creating a format that is compatible with the model for prediction.

2. Model Prediction and Visualization

Run ./dose_study_drugs_visual.ipynb to perform predictions and visualize the results.