PPSNO: A Feature-Rich SNO Sites Predictor by Stacking Ensemble Strategy from Protein Sequence-Derived Information

Area of Science:

• Biochemistry
• Proteomics
• Bioinformatics

Background:

• Protein S-nitrosylation (SNO) is a critical post-translational modification impacting protein stability, activity, localization, and function.
• Accurate prediction of SNO sites is essential for elucidating biological mechanisms and disease pathways.

Purpose of the Study:

• To develop a highly accurate computational predictor, named PPSNO, for identifying protein S-nitrosylation sites.
• To enhance the prediction of SNO sites by integrating multiple machine learning techniques through stacked ensemble learning.

Main Methods:

• Construction of benchmark datasets by collecting SNO sites from diverse sources.
• Application of various feature extraction techniques on protein sequences.
• Development of the PPSNO predictor using a two-layer stacked ensemble model integrating extracted features.

Main Results:

• PPSNO achieved high predictive performance with 92.8% accuracy, 96.1% AUC, 81.3% MCC, 85.6% F1-score, 79.3% SN, 97.7% SP, and 92.2% AP.
• Comparative analysis demonstrated that PPSNO significantly outperformed existing SNO site predictors (PSNO, PreSNO, pCysMod, DeepNitro, RecSNO, Mul-SNO).
• Visualizations including ROC curves, PR curves, and radar plots confirmed the superior performance of PPSNO.

Conclusions:

• Fused protein sequence features and stacked ensemble models markedly improve the accuracy of SNO site prediction.
• The PPSNO predictor offers a valuable tool for advancing the understanding of cellular processes and disease mechanisms.
• The developed codes and data are publicly available for further research and application.