Prediction of protein-coding small ORFs in multi-species using integrated sequence-derived features and the random...

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Evolutionary Biology

Background:

• Small open reading frames (sORFs) encode functional proteins and contribute to gene birth and evolutionary innovation.
• Accurate identification of protein-coding sORFs is essential for understanding their roles in biological systems.
• Existing prediction methods may have limitations in accuracy and scope.

Purpose of the Study:

• To develop a reliable prediction method for protein-coding sORFs based on sequence-derived features.
• To assess the prediction performance against established methods.
• To provide a user-friendly tool for sORF analysis.

Main Methods:

• Integrative analysis of sequence-derived features for sORF prediction.
• Development of domain-specific models for eukaryotes and prokaryotes.
• Correlation analysis with Codon Adaptation Index (CAI) to estimate potential expression.
• Web-server implementation for accessible prediction.

Main Results:

• The proposed method demonstrates superior or comparable performance to nine other prevalent prediction methods.
• Sequence features of protein-coding sORFs show significant differences between eukaryotic and prokaryotic domains.
• Domain-specific models improve prediction accuracy for cross-domain sORFs.
• A strong correlation was observed between predicted possibility and CAI, indicating reliable expression estimation.

Conclusions:

• The developed method offers a reliable tool for predicting protein-coding sORFs.
• Understanding domain-specific sequence features is crucial for accurate cross-domain prediction.
• The provided web-server facilitates research on sORFs and their functional roles.