Recent Development of Machine Learning Methods in Microbial Phosphorylation Sites.

Area of Science:

• Microbiology
• Biochemistry
• Computational Biology

Background:

• Protein post-translational modifications, like phosphorylation, regulate vital cellular functions.
• Phosphorylation is critical in mycobacteria for processes including cell division and communication.
• High-precision mass spectrometry has identified numerous microbial phosphorylated proteins.

Purpose of the Study:

• To provide a comprehensive survey of existing machine learning (ML) predictors for microbial phosphorylation.
• To analyze key aspects of ML predictor development, including algorithms, feature selection, and utility.
• To identify limitations and future research directions for computational phosphorylation prediction.

Main Methods:

• Review of current microbial phosphorylation site databases.
• Analysis of state-of-the-art ML approaches, their working principles, and performance metrics.
• Examination of factors influencing predictor development: ML algorithms, feature selection, window size, and software utility.

Main Results:

• A review of available ML tools for predicting microbial phosphorylation sites.
• Discussion on the performance and limitations of current computational methods.
• Identification of essential components for developing effective ML predictors.

Conclusions:

• Machine learning approaches can overcome experimental limitations in identifying phosphorylation sites.
• Further development of ML predictors is needed to advance the study of microbial phosphorylation.
• This survey provides a foundation for future research in computational prediction of microbial phosphorylation.