MSF-UBRW: An Improved Unbalanced Bi-Random Walk Method to Infer Human lncRNA-Disease Associations.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long non-coding RNAs (lncRNAs) play crucial roles in human diseases.
• Accurate identification of lncRNA-disease associations (LDAs) is vital for understanding disease mechanisms.
• Existing computational models have limitations in discovering novel LDAs.

Purpose of the Study:

• To develop a novel computational method, MSF-UBRW, for predicting potential lncRNA-disease associations (LDAs).
• To enhance the accuracy and reliability of LDA prediction by integrating multiple similarity measures.
• To validate the method's effectiveness through rigorous cross-validation and case studies.

Main Methods:

• A novel method, MSF-UBRW (multiple similarities fusion based on unbalanced bi-random walk), was designed.
• Linear fusion of functional similarity and Gaussian Interaction Profile kernel similarity for lncRNAs and diseases.
• Prediction of potential LDAs using an unbalanced bi-random walk approach.
• Validation using leave-one-out cross-validation (LOOCV) and 5-fold cross-validation (5-fold CV).

Main Results:

• The MSF-UBRW method demonstrated high prediction accuracy.
• Achieved an Area Under the Curve (AUC) of 0.9391 in LOOCV.
• Achieved an AUC of 0.9183 (±0.0054) in 5-fold CV, confirming reliable performance.
• Case studies successfully identified new LDAs for three common diseases.

Conclusions:

• MSF-UBRW effectively predicts novel lncRNA-disease associations.
• The integration of multiple similarities significantly improves prediction performance.
• MSF-UBRW offers a reliable tool for advancing research in lncRNA and disease studies.