SAPP: Structure Aware PTM Prediction.

Area of Science:

• Biochemistry
• Computational Biology
• Structural Biology

Background:

• Post-translational modifications (PTMs) are crucial for regulating diverse cellular processes, including signal transduction, cell growth, and differentiation.
• Accurate identification of PTM sites is essential for understanding cellular mechanisms and developing targeted therapies.
• Existing computational models often overlook structural contexts like intrinsically disordered regions and solvent accessibility, limiting prediction accuracy.

Purpose of the Study:

• To introduce SAPP (Structure-Aware PTM Prediction), a novel computational framework for PTM site prediction.
• To integrate protein structural information with sequence data for improved prediction.
• To demonstrate the utility of structure-based models in PTM prediction, including generalization to PTM types with limited data.

Main Methods:

• Developed SAPP, a Transformer-based model utilizing self-attention and cross-attention mechanisms.
• Integrated structural features derived from AlphaFold2 predictions with sequence information.
• Employed a phosphorylation-pretrained model for fine-tuning to other PTM types.

Main Results:

• SAPP effectively captures complex interactions between protein sequences and their structural states.
• The structure-aware approach significantly improves PTM prediction accuracy and biological relevance compared to sequence-only models.
• SAPP demonstrates strong generalization performance for PTM types with limited training data.

Conclusions:

• Structural information is critical for accurate PTM site prediction.
• SAPP represents a pioneering structure-based framework for PTM prediction, advancing the field.
• This work deepens the understanding of PTMs' biological significance and opens avenues for therapeutic development.