A Method of Optimizing Weight Allocation in Data Integration Based on Q-Learning for Drug-Target Interaction...

Area of Science:

• Computational chemistry
• Bioinformatics
• Drug discovery

Background:

• Predicting drug-target interactions (DTIs) is vital for discovering new drugs.
• Current methods fuse heterogeneous data (e.g., chemical structures, protein sequences) but struggle with optimal weight allocation.
• Reasonable weight allocation of heterogeneous information remains a significant challenge in DTI prediction.

Purpose of the Study:

• To propose a novel model, QLNRLMF, for predicting DTIs by effectively fusing heterogeneous information.
• To address the challenge of optimizing weight allocation for different data sources in DTI prediction.
• To enhance the accuracy and performance of DTI prediction models.

Main Methods:

• Utilized Q-learning algorithm to optimize weights for drug-drug and target-target similarity matrices derived from diverse data sources (chemical structure, protein sequence, DTIs).
• Generated new similarity matrices through linear combination based on optimized weights.
• Integrated these optimized similarity matrices with the drug-target interaction matrix into the Neighborhood Regularized Logistic Matrix Factorization (NRLMF) model.

Main Results:

• The proposed QLNRLMF model demonstrated superior performance compared to six existing methods (NetLapRLS, BLM-NII, WNN-GIP, KBMF2K, CMF, NRLMF).
• Achieved better prediction effects across four benchmark datasets: enzymes (E), nuclear receptors (NR), ion channels (IC), and G protein-coupled receptors (GPCR).
• Successfully optimized the fusion of heterogeneous information for improved DTI prediction accuracy.

Conclusions:

• The QLNRLMF model offers an effective approach for predicting drug-target interactions by intelligently fusing heterogeneous data.
• Optimizing information fusion through Q-learning significantly enhances DTI prediction performance.
• This method provides a valuable tool for accelerating novel drug discovery and development.