Systematic characterization and prediction of post-translational modification cross-talk

Area of Science:

• Proteomics and Molecular Biology
• Computational Biology and Bioinformatics

Background:

• Post-translational modifications (PTMs) regulate protein function.
• PTM cross-talk, involving interactions between multiple PTM sites on a single protein, is critical for determining functional outcomes.
• Systematic characterization of PTM cross-talk properties is needed.

Purpose of the Study:

• To systematically characterize the properties of PTM cross-talks.
• To develop a computational model for predicting PTM cross-talks.
• To provide a tool for prioritizing PTM cross-talk candidates for functional validation.

Main Methods:

• Collected 193 PTM cross-talk pairs from 77 human proteins.
• Analyzed location preference and co-evolution of PTM sites.
• Developed a naïve Bayes classifier integrating features to predict PTM cross-talks.
• Validated the model using 10-fold cross-validation.

Main Results:

• PTM cross-talks occur preferentially among nearby sites, particularly in disordered protein regions.
• PTM cross-talk pairs exhibit co-evolutionary patterns.
• The integrated prediction model achieved an Area Under the ROC Curve of 0.833, outperforming individual features.
• The model's performance is robust to data biases.

Conclusions:

• PTM cross-talks have predictable properties related to site proximity and co-evolution.
• An integrated computational approach effectively predicts PTM cross-talks.
• The developed web server facilitates large-scale identification of potential PTM cross-talks for further research.