PseUI: Pseudouridine sites identification based on RNA sequence information.

Area of Science:

• Molecular Biology
• Bioinformatics

Background:

• Pseudouridylation is a prevalent RNA modification impacting cellular processes.
• Accurate identification of pseudouridine (Ψ) sites is crucial for understanding RNA function.
• Existing experimental methods for Ψ site detection are inefficient and costly.

Purpose of the Study:

• To develop a computational model for accurate and efficient pseudouridine (Ψ) site identification in RNA sequences.
• To improve upon the predictive accuracy of existing computational methods for Ψ site detection.

Main Methods:

• Generated five types of RNA sequence features: nucleotide composition (NC), dinucleotide composition (DC), pseudo dinucleotide composition (pseDNC), position-specific nucleotide propensity (PSNP), and position-specific dinucleotide propensity (PSDP).
• Employed a sequential forward feature selection strategy to identify a discriminative feature subset.
• Utilized a support vector machine (SVM) to build the PseUI prediction model.
• Validated model generalization using jackknife and independent validation tests.

Main Results:

• The PseUI model demonstrated higher accuracy and stability compared to previously published models for Ψ site identification across three species (H. sapiens, S. cerevisiae, M. musculus).
• The model achieved superior performance on benchmark datasets.
• A user-friendly web server for PseUI was developed and made accessible.

Conclusions:

• PseUI is a novel and effective predictor for identifying pseudouridine (Ψ) sites in RNA sequences.
• The proposed model outperforms existing state-of-the-art methods.
• PseUI is expected to be a valuable tool for RNA pseudouridine site identification in academic research.