TransPTM: a transformer-based model for non-histone acetylation site prediction

Area of Science:

• Biochemistry
• Computational Biology
• Genomics

Background:

• Protein acetylation is a crucial post-translational modification (PTM) impacting numerous biological processes.
• Existing computational tools often lack specialized datasets and predictors for non-histone acetylation sites.

Purpose of the Study:

• To create a benchmark dataset for non-histone acetylation sites.
• To develop and validate a novel computational predictor for non-histone acetylation sites.

Main Methods:

• Construction of the NHAC benchmark dataset with varying sequence lengths.
• Development of TransPTM, a transformer-based neural network model utilizing ProtT5 embeddings and a graph neural network framework.
• Benchmarking TransPTM against state-of-the-art tools for non-histone acetylation site prediction.

Main Results:

• TransPTM demonstrates competitive performance in predicting non-histone acetylation sites.
• The study provides a valuable dataset, addressing a gap in non-histone acetylation research.
• The model's performance surpasses existing state-of-the-art tools.

Conclusions:

• TransPTM offers a robust tool for non-histone acetylation site prediction.
• The developed dataset and model enhance understanding of PTM mechanisms.
• This work provides a foundation for identifying drug targets for related diseases.