Prediction of miRNA-disease association based on multisource inductive matrix completion

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are key regulators in cellular processes and human diseases.
• Accurate prediction of miRNA-disease associations is vital for clinical applications like diagnosis and prognosis.
• Existing methods may not fully capture complex interactions between miRNAs and diseases.

Purpose of the Study:

• To develop and validate a novel computational model, Autoencoder Inductive Matrix Completion (AEIMC), for predicting potential miRNA-disease associations.
• To integrate diverse biological data sources for enhanced prediction accuracy.
• To demonstrate the utility of AEIMC in identifying novel miRNA-disease relationships.

Main Methods:

• The AEIMC model utilizes autoencoders to extract abstract feature representations from multiple similarity networks.
• It incorporates miRNA functional similarity, miRNA sequence similarity, disease semantic similarity, disease ontology similarity, and Gaussian interaction kernel similarity.
• Inductive matrix completion is employed for the final prediction of miRNA-disease associations.

Main Results:

• The AEIMC model demonstrated high effectiveness in predicting miRNA-disease associations, validated through cross-validation and stratified threshold evaluation.
• Ablation experiments confirmed the significant contribution of sequence and ontology similarities to the model's performance.
• Case studies further supported the model's capability in identifying relevant miRNA-disease links.

Conclusions:

• The AEIMC model provides a robust and effective approach for predicting miRNA-disease associations.
• Integrating multiple similarity networks and employing autoencoders enhances the accuracy and interpretability of predictions.
• This computational tool has the potential to accelerate the discovery of disease-related miRNAs and inform therapeutic strategies.