HPTRMF: Collaborative Matrix Factorization-Based Prediction Method for LncRNA-Disease Associations Using High-Order...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Predicting long noncoding RNA (lncRNA)-disease associations is crucial for understanding disease mechanisms.
• Existing matrix factorization methods struggle with the cold start problem and nonlinear data loss.

Purpose of the Study:

• To introduce HPTRMF, a novel matrix factorization approach designed to overcome limitations in lncRNA-disease association prediction.
• To enhance the accuracy and robustness of predicting associations between lncRNAs and diseases.

Main Methods:

• HPTRMF utilizes high-order perturbation to construct a correlation matrix, effectively addressing data sparsity and the cold start problem.
• A flexible trifactor regularization term is incorporated to capture complex similarity patterns in lncRNA and disease data, mitigating nonlinear data loss.

Main Results:

• HPTRMF demonstrated superior performance compared to nine state-of-the-art algorithms across three datasets.
• Validation through Leave-One-Out Cross-Validation (LOOCV) and Five-Fold Cross-Validation (5-Fold CV) confirmed HPTRMF's effectiveness.

Conclusions:

• HPTRMF offers a significant advancement in predicting lncRNA-disease associations, outperforming existing methods.
• The approach effectively handles data sparsity and nonlinearities, providing a more reliable tool for biological research.