DeepPTMPred: a multi-modal deep learning framework for accurate prediction of protein post-translational modification...

Area of Science:

• Biochemistry
• Computational Biology
• Proteomics

Background:

• Post-translational modifications (PTMs) are crucial for regulating diverse cellular functions.
• Traditional experimental methods for identifying PTM sites are accurate but labor-intensive and time-consuming.

Purpose of the Study:

• To develop a novel computational model for predicting 17 types of PTMs.
• To integrate multi-modal data and AlphaFold predictions for enhanced PTM site prediction accuracy.

Main Methods:

• Utilized an enhanced CNN-transformer architecture to analyze sequence dependencies.
• Incorporated structural features and evolutionary patterns for spatial and global context.
• Employed multi-modal data integration and AlphaFold predictions.

Main Results:

• Achieved high AUC scores: 96.5% for hydroxylation, 91.6% for malonylation, 91.0% for O-linked glycosylation, and 89.5% for phosphorylation.
• Successfully identified known tau phosphorylation sites.
• Detected novel phosphorylation sites linked to early Alzheimer's disease pathology.

Conclusions:

• The developed model offers a highly accurate and efficient method for predicting PTM sites.
• This work advances the understanding of PTMs and their roles in cellular processes.
• The model shows potential for future research in PTM prediction and functional annotation, particularly in neurodegenerative diseases.