Naïve Bayes classifier with feature selection to identify phage virion proteins

Area of Science:

• Proteomics
• Bioinformatics
• Virology

Background:

• Understanding phage virion protein composition is crucial for elucidating viral functions.
• Experimental identification of virion proteins is labor-intensive and costly.
• Developing computational methods is essential for high-throughput phage proteomics.

Purpose of the Study:

• To propose a novel computational method for predicting phage virion proteins.
• To enhance the efficiency and reduce the cost of phage proteomics research.

Main Methods:

• A Naïve Bayes classifier was utilized for prediction.
• Amino acid composition and dipeptide composition were used as features.
• A feature selection technique was employed to optimize informative features.

Main Results:

• The proposed method achieved 79.15% accuracy in classifying phage virion and nonvirion proteins.
• The Naïve Bayes method outperformed other state-of-the-art classifiers.
• Optimized features improved classification performance.

Conclusions:

• The developed computational method is effective and promising for high-throughput phage proteomics.
• This approach provides a valuable tool for accelerating the study of phage virion proteins.
• The method offers a cost-effective and efficient alternative to traditional experimental techniques.