EEG-DTI

EEG-DTI predicts drug-target interactions using end-to-end heterogeneous graph representation learning to generate low-dimensional feature representations for drugs and targets.

Key Features:

• End-to-End Learning Framework: Integrates end-to-end heterogeneous graph representation learning with graph convolutional networks to predict drug-target interactions.
• Heterogeneous Graph Convolutional Networks (GCNs): Employs GCNs to model complex networks comprising drugs, proteins, diseases, and side effects.
• Low-Dimensional Feature Representation: Learns compact embeddings for drugs and targets to enable efficient large-scale DTI prediction.
• Comprehensive Heterogeneous Network: Utilizes a network of multiple biological entity types and their interactions to capture cross-entity dependencies.

Scientific Applications:

• Drug Discovery Prioritization: Predicts potential DTIs to prioritize candidate compounds for experimental validation in early-stage drug development.

Methodology:

Constructs a heterogeneous network with nodes representing entities (e.g., drugs, proteins) and edges representing interactions, applies graph convolutional networks to learn feature representations in an end-to-end manner, and uses the learned features to predict drug-target interactions.