A Computational Model to Predict the Causal miRNAs for Diseases.

Area of Science:

• Biochemistry
• Genetics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are crucial noncoding RNAs involved in various diseases.
• Existing databases identify miRNA-disease associations but not causality.
• Determining causal roles of miRNAs is vital for understanding disease mechanisms.

Purpose of the Study:

• To manually curate causal miRNA-disease associations.
• To develop a computational model (MDCAP) for predicting novel causal miRNA-disease associations.
• To enhance the Human miRNA Disease Database (HMDD) with causal information.

Main Methods:

• Manual curation of causal miRNA-disease associations.
• Development and application of the MiRNA-Disease Causal Association Predictor (MDCAP) model.
• Validation using ROC analysis (independent test and 10-fold cross-validation) and case studies.

Main Results:

• Collected 6,667 causal miRNA-disease associations involving 616 miRNAs and 440 diseases.
• MDCAP achieved high predictive performance with an AUC of 0.928 (independent test) and 0.925 (10-fold CV).
• Case studies on myocardial infarction and hsa-mir-498 demonstrated the model's biomedical significance.

Conclusions:

• The MDCAP model effectively predicts causal miRNA-disease associations.
• This work provides a valuable resource for understanding miRNA roles in disease pathogenesis.
• The findings highlight the importance of causal inference in miRNA-disease research.