This is the implementation of EviDTI: Evidential Deep Learning for Guided Drug-Target Interaction Prediction
EviDTI is built on Python3, we recommend using a virtual conda environment as enviroment management for the installation of EviDTI and its dependencies. The virtual environment can be created as follows:
conda create -n evidti python==3.8
conda activate evidti
Download the source code of EviDTI from GitHub:
git clone https://github.com/username/my-project.git
Install EviDTI dependencies as following,this will take about one to two hours.:
conda env create -f environment.yaml
The data feature file used in this study can be download from the link https://zenodo.org/records/14056305
Firstly, seqence features of proteins,
2D as well as 3D input features of drugs need to be generated
Seqence features of proteins can be generated by extract_p_emb.py
python extract_p_emb.py
2D input features of drugs can be generated by extract_drug_2d_emb.py
python extract_drug_2d_emb.py
3D input features of drugs can be generated by extract_drug_3d_emb.py
python extract_drug_3d_emb.py
Multiple hyperparameters can be selected in bashconfigs/XX.yaml and XX_configs.py.
XX Replace with drugbank, davis or kiba depending on the dataset.
Run the main-drugbank.py to train the drugbank dataset
python main_drugbank.py --cfg configs/drugbank.yaml --data drugbank --split random
It's similar for kiba and davis.
python main_davis.py --cfg configs/davis.yaml --data davis --split random
python main_KIBA.py --cfg configs/kiba.yaml --data kiba --split random
After model training starts, the progress bar will be automatically shown on your command line， and the trained model parameters will be saved in "runs" dictory for the best epoch.
Model parameters can be found under the runs folder. It takes about 5 minutes to run a test on the DrugBank test set (~3000 samples) and output the results
python infernce.py
After predicting with your well trained model,
the predicting output will be saved in "runs" dictory.
Sample test results and saved model parameters are placed under https://zenodo.org/records/14056305 due to file size.
This project is covered under the Apache 2.0 License.
This is the implementation of EviDTI: Evidential Deep Learning for Guided Drug-Target Interaction Prediction
EviDTI is built on Python3, we recommend using a virtual conda environment as enviroment management for the installation of EviDTI and its dependencies. The virtual environment can be created as follows:
conda create -n evidti python==3.8
conda activate evidti
Download the source code of EviDTI from GitHub:
git clone https://github.com/username/my-project.git
Install EviDTI dependencies as following,this will take about one to two hours.:
conda env create -f environment.yaml
The data feature file used in this study can be download from the link https://zenodo.org/records/14056305
Firstly, seqence features of proteins,
2D as well as 3D input features of drugs need to be generated
Seqence features of proteins can be generated by extract_p_emb.py
python extract_p_emb.py
2D input features of drugs can be generated by extract_drug_2d_emb.py
python extract_drug_2d_emb.py
3D input features of drugs can be generated by extract_drug_3d_emb.py
python extract_drug_3d_emb.py
Multiple hyperparameters can be selected in bashconfigs/XX.yaml and XX_configs.py.
XX Replace with drugbank, davis or kiba depending on the dataset.
Run the main-drugbank.py to train the drugbank dataset
python main_drugbank.py --cfg configs/drugbank.yaml --data drugbank --split random
It's similar for kiba and davis.
python main_davis.py --cfg configs/davis.yaml --data davis --split random
python main_KIBA.py --cfg configs/kiba.yaml --data kiba --split random
After model training starts, the progress bar will be automatically shown on your command line， and the trained model parameters will be saved in "runs" dictory for the best epoch.
Model parameters can be found under the runs folder. It takes about 5 minutes to run a test on the DrugBank test set (~3000 samples) and output the results
python infernce.py
After predicting with your well trained model,
the predicting output will be saved in "runs" dictory.
Sample test results and saved model parameters are placed under https://zenodo.org/records/14056305 due to file size.
This project is covered under the Apache 2.0 License.