Predicting multiple types of MicroRNA-disease associations based on tensor factorization and label propagation

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are crucial regulators in disease pathogenesis.
• Current bioinformatics methods primarily focus on binary miRNA-disease associations, neglecting diverse association types.
• Existing miRNA-disease association datasets suffer from noise and incompleteness.

Purpose of the Study:

• To develop a novel method for predicting multiple types of miRNA-disease associations.
• To address the inherent noise and incompleteness in miRNA-disease association data.
• To improve the accuracy and scope of miRNA-disease association prediction.

Main Methods:

• Tensor robust principal component analysis (TRPCA) for cleaning and completing miRNA-disease association data.
• Gaussian interaction profile (GIP) kernel, disease semantic similarity, and miRNA functional similarity to construct integrated similarity networks.
• Label propagation integrating low-rank association tensor and biological similarity for iterative prediction.

Main Results:

• The proposed Tensor Factorization and Label Propagation (TFLP) method effectively handles noisy and incomplete data.
• TFLP demonstrates superior performance compared to state-of-the-art methods in predicting multiple miRNA-disease associations.
• The method successfully integrates tensor factorization and label propagation for enhanced prediction accuracy.

Conclusions:

• The TFLP method offers a robust approach for predicting complex miRNA-disease relationships.
• This work advances the field of bioinformatics by providing a more comprehensive prediction model.
• The findings have implications for understanding disease mechanisms and developing targeted therapies.