GraphDTA

GraphDTA predicts drug-target binding affinity using graph neural networks to estimate interaction strengths between small-molecule drugs and protein targets for applications such as drug repurposing.

Key Features:

• Graph-based drug representation: Represents small-molecule drugs as molecular graphs with atoms as nodes and chemical bonds as edges.
• Graph Neural Networks (GNNs): Applies graph neural network architectures to learn representations from molecular graphs and predict binding affinity, demonstrating superior performance relative to non-deep-learning and other deep-learning methods.
• Implementation: Implemented in Python.

Scientific Applications:

• Drug repurposing: Predicts binding affinities to prioritize existing drugs for novel protein targets.
• Experimental prioritization: Ranks drug-target pairs to guide selection of candidates for experimental validation.

Methodology:

Represents molecules as graphs (nodes = atoms, edges = bonds) and processes these graphs with a graph neural network architecture to learn patterns that correlate with binding affinities.