Multitype Perception Method for Drug-Target Interaction Prediction.

Area of Science:

• Computational chemistry
• Bioinformatics
• Artificial intelligence in drug discovery

Background:

• Deep learning is increasingly used for predicting drug-target interactions (DTIs).
• Existing methods often fail to integrate diverse knowledge from various interaction types (e.g., drug-drug, drug-target, drug-enzyme).
• This limitation hinders the full exploitation of knowledge diversity for improved DTI prediction.

Purpose of the Study:

• To propose a novel multitype perception method (MPM) for DTI prediction.
• To address the challenge of exploiting knowledge diversity across different interaction types.
• To enhance the accuracy and robustness of DTI prediction models.

Main Methods:

• Developed a multitype perception method (MPM) comprising a type perceptor and a multitype predictor.
• The type perceptor learns distinguished edge representations, retaining specific features for each interaction type.
• The multitype predictor calculates type similarity and uses a domain gate module for adaptive weighting.

Main Results:

• MPM effectively exploits knowledge diversity across different link types for DTI prediction.
• The type perceptor component maximizes prediction performance for individual interaction types.
• Extensive experiments confirmed that MPM outperforms current state-of-the-art methods.

Conclusions:

• The proposed MPM offers a significant advancement in DTI prediction by integrating diverse interaction knowledge.
• This approach enhances the predictive power of deep learning models in drug discovery.
• MPM provides a more comprehensive strategy for understanding and predicting drug-target relationships.