# DDINet

**Repository Path**: Tomhappy/DDINet

## Basic Information

- **Project Name**: DDINet
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-04-11
- **Last Updated**: 2026-04-11

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

## Paper Overview and Graphical Representation
Drug-drug interactions (DDIs) can cause unexpected side effects, posing challenges in both research and healthcare. As more drugs are prescribed together, the risk of harmful DDIs increases. Accurate prediction is crucial for improving drug safety and protecting patients. Many DDIs have been identified, but their mechanisms remain unclear. In this study, we introduce DDINet, a model that predicts DDIs using Morgan fingerprints to analyze drug structures. These fingerprints help capture key features for accurate predictions. DDINet achieves high accuracy, AUROC, and AUPR scores on both seen-unseen and unseen-unseen datasets, outperforming baseline models in both binary and multi-class classification tasks. It also handles imbalanced datasets effectively and performs well with the scaffold-splitting method. These results suggest that DDINet can aid drug development and clinical decision-making by detecting risky drug combinations early. Future work will integrate clinical data, additional drug properties, and advanced models like transformers and BERT to improve accuracy on unseen drug combinations.

![Data_and_Model_diagram-min_optimized_10000](https://github.com/user-attachments/assets/28dcfe9a-3a26-4443-b175-9457162a0b25)


## Experimental Setup
The model was developed in Python 3.11.14 using TensorFlow 2.20.0 and Keras 3.11.3, within Spyder 6.1.0. The environment also included NumPy 2.3.3, Pandas 2.3.3, Scikit-learn 1.7.2, and RDKit 2025.03.6. All experiments were conducted with CUDA 11.2 on an NVIDIA TITAN Xp GPU with 12 GB of memory.