DeepSuccinylSite: a deep learning based approach for protein succinylation site prediction

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• Protein succinylation is a significant post-translational modification (PTM) occurring on lysine residues.
• Succinylation alters protein charge and is linked to gene activity and cellular respiration.
• This PTM may play a role in genomic regulation, repair, and disease etiology.

Purpose of the Study:

• To develop a novel prediction tool for identifying protein succinylation sites.
• To leverage deep learning and embedding methodologies for enhanced prediction accuracy.

Main Methods:

• Development of DeepSuccinylSite, a computational tool utilizing deep learning.
• Application of embedding techniques for primary protein structure analysis.
• Validation using an independent set of experimentally identified succinylation sites.

Main Results:

• DeepSuccinylSite achieved high performance metrics: 79% sensitivity, 68.7% specificity, and 0.48 MCC.
• The model demonstrated a robust area under the ROC curve of 0.8.
• Comparative analysis showed DeepSuccinylSite significantly outperformed existing succinylation prediction tools.

Conclusions:

• The developed method offers a robust and complementary approach for studying protein succinylation.
• DeepSuccinylSite facilitates advanced exploration of succinylation's role in biological processes.