ncRNA-disease association prediction based on sequence information and tripartite network

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Non-coding RNAs (ncRNAs) play crucial roles in biological processes and are linked to human diseases.
• Understanding ncRNA-disease associations is vital for disease diagnosis, treatment, and prevention.
• Limited computational methods exist for predicting these associations, particularly in pathogenesis.

Purpose of the Study:

• To develop and validate a novel computational method for predicting ncRNA-disease associations.
• To leverage biological sequence information and network analysis for improved prediction accuracy.
• To identify potential ncRNA-disease relationships for further experimental investigation.

Main Methods:

• Construction of a tripartite network integrating disease, target, and ncRNA information.
• Computation of prediction scores using pairwise similarity based on sequence expressions.
• Application of a multi-layer resource allocation technique to derive weights.
• Evaluation using 5-fold cross-validation with kernel parameter tuning.

Main Results:

• The proposed algorithm achieved an average AUC of 0.75 (without link cut) and 0.57 (with link cut).
• Performance surpassed previous methods evaluated under identical conditions.
• The method successfully predicted 23 ncRNA-disease associations validated by independent biological studies.

Conclusions:

• The developed computational method demonstrates high capability and accuracy in predicting ncRNA-disease associations.
• Integrating biological sequence information significantly enhances prediction performance.
• This work highlights the potential of computational approaches in uncovering novel ncRNA-disease relationships.