Drug-target interaction prediction using unifying of graph regularized nuclear norm with bilinear factorization

Area of Science:

• Computational Biology
• Bioinformatics
• Drug Discovery

Background:

• Wet-lab experiments for drug-target interaction (DTI) identification are resource-intensive.
• Computational prediction of DTIs is crucial for accelerating drug discovery.
• Predictive models reduce the search space for potential drug-target interactions.

Purpose of the Study:

• To develop an efficient computational method for predicting drug-target interactions.
• To improve the accuracy and reduce the cost of identifying potential drug candidates.
• To leverage matrix factorization and graph-based approaches for DTI prediction.

Main Methods:

• A novel approach unifying matrix factorization and nuclear norm minimization for low-rank DTI prediction.
• Utilizing Rank-Restricted Soft Singular Value Decomposition (RRSSVD) for bilinear factorization.
• Encoding drug-target adjacencies using graphs, incorporating drug-drug and target-target similarities.

Main Results:

• The proposed method effectively identifies low-rank drug-target interactions.
• Performance evaluation on benchmark datasets (Enzymes, Ion Channels, GPCRs, Nuclear Receptors) demonstrated significant improvements.
• The method showed enhanced accuracy compared to existing state-of-the-art techniques.

Conclusions:

• The unified computational approach offers a more efficient and accurate alternative to traditional wet-lab DTI identification.
• The method successfully predicts drug-target interactions, aiding in the drug discovery pipeline.
• The findings suggest a promising direction for computational drug discovery research.