FuncPhos-STR

FuncPhos-STR predicts functional phosphosites and their regulatory mechanisms across the human proteome by integrating AlphaFold protein structures, structural dynamics, sequence evolution, and protein-protein interaction information.

Key Features:

• Structural and dynamics integration: Incorporates structural topology and dynamics features from AlphaFold protein structures to characterize functional phosphosites.
• Sequence evolutionary features: Integrates phosphosite sequence evolution information from the reported FuncPhos-SEQ framework.
• Protein-protein interaction network features: Uses PPI information to capture network-level properties of phosphosites.
• Multi-dimensional feature integration: Combines structural and dynamics, sequence evolutionary, and PPI network features from different protein dimensions.
• Performance benchmarking: Achieves the best AUC value of 0.855 compared to other reported models.
• Pocket region accessibility analysis: Identifies phosphosites inside pocket regions that receive higher functional scores, supporting their potential regulatory mechanisms.

Scientific Applications:

• Functional phosphosite identification: Accelerates functional assignment of unexplored phosphosites in the human proteome.
• Allosteric regulation mechanism elucidation: Facilitates understanding of molecular mechanisms by which phosphorylation regulates protein functions through structural and dynamics characterization.
• Structural visualization: Enables structural visualization of human proteome-level phosphosites and their functional predictions.

Methodology:

FuncPhos-STR was developed based on the FuncPhos-SEQ framework and uses an integrated deep neural network that combines structural and dynamics information from AlphaFold protein structures with sequence evolutionary and PPI network features.