GPSD: a hybrid learning framework for the prediction of phosphatase-specific dephosphorylation sites
2025

GPSD: A Tool for Predicting Phosphatase-Specific Dephosphorylation Sites

Sample size: 4393 publication Evidence: high

Author Information

Author(s): Han Cheng, Fu Shanshan, Chen Miaomiao, Gou Yujie, Liu Dan, Zhang Chi, Huang Xinhe, Xiao Leming, Zhao Miaoying, Zhang Jiayi, Xiao Qiang, Peng Di, Xue Yu

Primary Institution: Huazhong University of Science and Technology

Hypothesis

Can a hybrid learning framework improve the prediction of phosphatase-specific dephosphorylation sites?

Conclusion

The GPSD tool effectively predicts phosphatase-specific dephosphorylation sites using a hybrid learning framework.

Supporting Evidence

  • 4393 site-specific phosphatase–substrate relationships were manually curated.
  • A hybrid learning framework was developed integrating 10 sequence features and 3 machine learning methods.
  • 103 individual phosphatase-specific predictors were fine-tuned using transfer learning and meta-learning.
  • The GPSD tool is freely available for public use.

Takeaway

Scientists created a computer program that helps predict where proteins get their phosphate groups removed, which is important for understanding how cells work.

Methodology

The study involved curating phosphatase-substrate relationships and developing a hybrid learning framework using various machine learning methods.

Limitations

The predictions need further experimental validation and the tool primarily focuses on sequence characteristics around dephosphorylation sites.

Digital Object Identifier (DOI)

10.1093/bib/bbae694

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication