DF-Phos: Prediction of Protein Phosphorylation Sites by Deep Forest

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Phosphorylation is a critical post-translational modification (PTM) vital for protein function and experimental design.
• Predicting phosphorylation sites is essential for understanding cellular processes.
• Deep learning methods show promise in improving phosphorylation site prediction accuracy.

Purpose of the Study:

• To introduce DF-Phos, a new computational tool for predicting phosphorylation sites.
• To evaluate the performance of the Deep Forest algorithm for phosphosite prediction.

Main Methods:

• DF-Phos employs a Deep Forest framework.
• Input features are derived from the CkSAApair method.
• Performance was assessed using 10-fold cross-validation.

Main Results:

• The Deep Forest method demonstrated superior performance compared to other available prediction methods.
• DF-Phos achieved high accuracy in predicting phosphorylation sites.

Conclusions:

• Deep Forest is a powerful and effective method for phosphosite prediction.
• DF-Phos offers a valuable, freely available resource for PTM research.