Predicting protein oxidation sites with feature selection and analysis approach.

Area of Science:

• Biochemistry
• Computational Biology
• Bioinformatics

Background:

• Protein oxidation is a common post-translational modification with significant physiological and pathological roles.
• Experimental determination of protein oxidation sites is costly and time-consuming.
• In silico methods are needed for rapid and effective identification of protein oxidation sites.

Purpose of the Study:

• To develop a computational method for identifying protein oxidation sites.
• To identify optimal features for predicting protein oxidation.

Main Methods:

• Utilized the nearest neighbor algorithm.
• Incorporated maximum relevance minimum redundancy and incremental feature selection.
• Selected 16 optimal features from an initial set of 735.

Main Results:

• Identified 16 optimal features, including conservation scores, amino acid factors, and solvent accessibility.
• Achieved an overall prediction success rate of 75.82%.
• The selected features offer insights into protein oxidation mechanisms.

Conclusions:

• The developed in silico method is promising for practical applications in identifying protein oxidation sites.
• The identified features can aid in understanding the mechanisms of protein oxidation.