Stacking based ensemble learning framework for identification of nitrotyrosine sites.

Area of Science:

• Biochemistry
• Computational Biology
• Proteomics

Background:

• Protein nitrotyrosine is a key post-translational modification impacting biological functions and diseases.
• Accurate identification of nitrotyrosine sites is vital for understanding biological processes and disease mechanisms.

Purpose of the Study:

• To introduce iNTyro-Stack, a novel computational tool for accurate identification of protein nitrotyrosine sites.
• To leverage machine learning for predicting nitrotyrosine modification sites.

Main Methods:

• Developed iNTyro-Stack, a machine learning model utilizing a stacking algorithm.
• Employed a feature map combining amino acid composition encoding schemes (k-spaced amino acid pairs, tri-peptide composition).
• Utilized recursive feature elimination for significant feature selection and evaluated performance via k-fold cross-validation and independent testing.

Main Results:

• iNTyro-Stack achieved 86.3% accuracy and 72.6% MCC in cross-validation.
• Demonstrated generalization capability on an imbalanced independent test set with 69.32% accuracy.
• Outperformed existing state-of-the-art methods in identifying protein nitrotyrosine sites.

Conclusions:

• iNTyro-Stack provides an accurate and effective computational approach for identifying protein nitrotyrosine sites.
• The tool aids in elucidating biological functions and diseases associated with nitrotyrosine modification.
• The methodology and results are reproducible via a publicly available GitHub repository.