A Novel Method to Predict Drug-Target Interactions Based on Large-Scale Graph Representation Learning.

Area of Science:

• Computational biology
• Bioinformatics
• Drug discovery

Background:

• Identifying drug-target interactions (DTIs) is crucial for drug discovery and repositioning.
• Traditional experimental methods for DTI identification are time-consuming and labor-intensive.
• Computational models offer a faster alternative for predicting high-quality DTI candidates.

Purpose of the Study:

• To propose a novel computational method, LGDTI, for predicting DTIs.
• To leverage large-scale graph representation learning for enhanced DTI prediction.
• To capture both local and global structural information within biological networks.

Main Methods:

• LGDTI integrates Graph Convolutional Networks (GCN) to aggregate first-order neighbor information.
• DeepWalk, a graph embedding method, is employed to learn high-order neighbor information.
• Features from GCN and DeepWalk are combined and fed into a Random Forest classifier for DTI prediction.

Main Results:

• LGDTI achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.9455.
• The method obtained an Area Under the Precision-Recall Curve (AUPR) of 0.9491.
• Performance was validated using 5-fold cross-validation and comparison with state-of-the-art methods.

Conclusions:

• LGDTI demonstrates efficient and robust prediction of undiscovered DTIs.
• The model offers a promising computational approach for accelerating drug discovery.
• The proposed method provides novel perspectives for researchers in the field.