Block sparse Bayes-based fuzzy system for RNA N6-methyladenosine sites prediction.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• N6-methyladenosine (m6A) is a crucial RNA post-transcriptional modification impacting gene expression, regulation, and cell fate.
• m6A modifications are implicated in various diseases, necessitating accurate site identification for research.
• Current computational methods for m6A site prediction often overlook multi-tissue analysis within species.

Purpose of the Study:

• To develop a robust computational model for predicting m6A modification sites in RNA sequences.
• To address the limitation of existing models by considering RNA modifications across different tissues within the same species.
• To enhance the accuracy and generalizability of m6A site prediction models.

Main Methods:

• Development of a fuzzy system based on Block Sparse Bayesian Learning (BSBL), named BSBL-TSK-FS.
• Implementation of a Bayesian method with posterior probability output for sparse solutions and higher accuracy.
• Validation using five-fold cross-validation (5-CV) and cross-species testing.

Main Results:

• The BSBL-TSK-FS model achieved high precision (0.84–0.95) in predicting m6A sites across mouse, human, and rat tissues.
• The model demonstrated a 9.4% improvement in accuracy compared to state-of-the-art (SOTA) predictors.
• Cross-species tests confirmed the model's robustness and adaptability, highlighting its generalizability.

Conclusions:

• BSBL-TSK-FS is a powerful and accurate sequence-level m6A prediction model.
• The model effectively integrates multi-tissue and cross-species data, outperforming existing SOTA methods.
• This tool provides a reliable approach for understanding the complex landscape of RNA methylation modifications.