Weighted matrix factorization on multi-relational data for LncRNA-disease association prediction.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long non-coding RNAs (lncRNAs) are implicated in critical biological processes and human diseases.
• Accurate identification of lncRNA-disease associations is essential for disease research.
• Existing data integration models for lncRNA-disease association prediction have limitations.

Purpose of the Study:

• To develop a novel data integrative model for precise identification of lncRNA-disease associations.
• To overcome the limitations of current methods that project heterogeneous data onto homologous networks.
• To propose a model that respects the intrinsic structures of individual data sources while fusing them.

Main Methods:

• Introduced a weighted matrix factorization model on multi-relational data (WMFLDA).
• Utilized a heterogeneous network to capture inter(intra)-associations between genes, lncRNAs, and Disease Ontology terms.
• Cooperatively decomposed weighted association matrices into low-rank matrices and jointly optimized matrices and weights.

Main Results:

• WMFLDA demonstrated significantly improved performance compared to existing data integrative solutions.
• The model achieved better results across various experimental settings and evaluation metrics.
• WMFLDA effectively preserved the intrinsic structures of individual data sources during fusion.

Conclusions:

• WMFLDA offers a superior approach for predicting lncRNA-disease associations.
• The model's ability to respect and fuse heterogeneous data structures enhances prediction accuracy.
• This method provides a valuable tool for advancing research in lncRNA-related diseases.