DeepNC: a framework for drug-target interaction prediction with graph neural networks.

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning

Background:

• Drug-target interaction (DTI) prediction is crucial for drug development.
• Deep learning models accelerate DTI prediction but often ignore molecular graph structures.
• Graph neural networks (GNNs) can effectively learn molecular features from graph representations.

Purpose of the Study:

• To propose Deep Neural Computation (DeepNC), an advanced deep learning framework for DTI prediction.
• To leverage GNNs to capture the structural and chemical properties of molecules for improved DTI prediction.
• To evaluate DeepNC's performance on benchmark and independent datasets for binding affinity prediction.

Main Methods:

• DeepNC framework integrates three GNN algorithms: Generalized Aggregation Networks (GENConv), Graph Convolutional Networks (GCNConv), and Hypergraph Convolution-Hypergraph Attention (HypergraphConv).
• GNN layers are used to learn drug features, while a 1-D convolution network learns target features.
• Predicted drug and target representations are fed into fully-connected layers to estimate binding affinity.

Main Results:

• DeepNC models were evaluated on the Davis, Kiba, and Allergy datasets.
• The framework demonstrated improved performance in predicting drug-target binding affinity.
• DeepNC outperformed baseline methods in terms of mean square error and concordance index.

Conclusions:

• DeepNC provides an effective deep learning approach for DTI prediction.
• The integration of multiple GNNs enhances the model's ability to learn complex molecular representations.
• DeepNC shows significant potential for accelerating the drug discovery pipeline through accurate binding affinity prediction.