SPLHRNMTF: robust orthogonal non-negative matrix tri-factorization with self-paced learning and dual hypergraph...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) play crucial roles in human diseases.
• Understanding miRNA-disease associations aids in elucidating disease pathogenesis.
• Traditional experimental methods for identifying these associations are resource-intensive.

Purpose of the Study:

• To develop a novel computational model for predicting miRNA-disease associations.
• To enhance the accuracy and efficiency of miRNA-disease association prediction.
• To provide a complementary tool to experimental methods in biological research.

Main Methods:

• Proposed a robust orthogonal non-negative matrix tri-factorization (NMTF) model with self-paced learning and dual hypergraph regularization (SPLHRNMTF).
• Employed non-linear fusion for comprehensive miRNA and disease similarity.
• Utilized weighted k-nearest neighbor profiles to correct false-negative associations and incorporated L1 norm for residual error calculation.
• Integrated self-paced learning to prevent local optima and applied hypergraph regularization to capture high-order relationships.

Main Results:

• SPLHRNMTF achieved higher average AUC values compared to baseline models in 5-fold cross-validation experiments.
• Case studies on breast and lung neoplasms confirmed the model's accuracy.
• Identified potential miRNA-disease associations with significant biological relevance.

Conclusions:

• SPLHRNMTF is an effective computational tool for predicting miRNA-disease associations.
• The model demonstrates superior performance and accuracy over existing methods.
• The findings contribute to a deeper understanding of disease mechanisms involving miRNAs.