A method for evaluating of RNA's coding potential using the interaction effects of open reading frames and...

Area of Science:

• Bioinformatics
• Molecular Biology
• Genomics

Background:

• Distinguishing long non-coding RNAs (lncRNAs) from protein-coding RNAs (pcRNAs) is crucial for understanding gene regulation in development and disease.
• Existing computational methods for RNA coding potential evaluation rely on various sequence properties but often require extensive tuning.

Purpose of the Study:

• To develop a novel, accurate, and easily applicable method for distinguishing lncRNAs from pcRNAs.
• To design sequence-derived features that are robust and do not necessitate hyperparameter optimization.

Main Methods:

• A series of features were designed based on the effects of open reading frames (ORFs) on RNA sequence interactions and site electrical intensity.
• A classification model was constructed using these designed features.
• Auxiliary features were combined with the primary features to further enhance prediction accuracy.

Main Results:

• The single model built on the designed features achieved a strong classifier performance with an accuracy between 82% and 89%.
• Combining auxiliary features with the designed features resulted in prediction accuracy comparable to or exceeding top-performing classification tools.
• The developed method demonstrated species insensitivity, making it broadly applicable across different organisms.

Conclusions:

• The proposed feature design and classification approach offers an accurate and user-friendly method for differentiating lncRNAs from pcRNAs.
• The species-insensitivity and lack of hyperparameter tuning requirements facilitate widespread adoption in RNA research.
• Identified feature correlations provide insights for future investigations into RNA coding potential and function.