Related Experiment Video
Updated: Jun 6, 2025
A Fast and Quantitative Method for Post-translational Modification and Variant Enabled Mapping of Peptides to Genomes
Published on: May 22, 2018
GPS-pPLM: A Language Model for Prediction of Prokaryotic Phosphorylation Sites
Chi Zhang1, Dachao Tang1, Cheng Han1
1Department of Bioinformatics and Systems Biology, MOE Key Laboratory of Molecular Biophysics, Hubei Bioinformatics and Molecular Imaging Key Laboratory, College of Life Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, China.
The GPS-pPLM tool accurately predicts prokaryotic phosphorylation sites (p-sites) using deep learning. This updated server enhances the study of protein posttranslational modifications (PTMs) in bacteria.
Area of Science:
- * Molecular Biology
- * Bioinformatics
- * Computational Biology
Background:
- * Protein phosphorylation is a crucial posttranslational modification (PTM) in prokaryotes, regulating diverse biological processes.
- * Accurate prediction of phosphorylation sites (p-sites) is essential for understanding prokaryotic cell function.
- * Existing prediction tools require improvement for prokaryotic systems.
Purpose of the Study:
- * To introduce an updated online server, the group-based prediction system for prokaryotic phosphorylation language model (GPS-pPLM).
- * To develop a highly accurate predictor for prokaryotic phosphorylation sites using advanced deep learning techniques.
- * To provide a valuable resource for researchers studying prokaryotic PTMs.
Main Methods:
- * Employed deep learning models, including transformer and deep neural networks, for p-site prediction.
- * Integrated 10 sequence and contextual features using a dataset of 44,839 nonredundant p-sites from 16,041 prokaryotic proteins.
- * Developed general models for O-phosphorylation and N-phosphorylation, fine-tuned into 6 residue-specific and 134 species-specific predictors.
Main Results:
- * The GPS-pPLM demonstrated superior accuracy in predicting prokaryotic O-phosphorylation p-sites compared to existing tools.
- * Generated 140 specialized predictors (6 residue-specific, 134 species-specific) for enhanced prediction.
- * Integrated predicted p-sites with experimental evidence, 3D structures, and disorder tendencies from 22 public resources.
Conclusions:
- * GPS-pPLM offers a significant advancement in predicting prokaryotic phosphorylation sites.
- * The tool provides valuable annotations, aiding functional and structural analysis of prokaryotic proteins.
- * The freely accessible online service supports academic research in microbiology and molecular biology.
