Identifying protein succinylation sites using generative transformer and a two-dimensional representation with a deep...

Area of Science:

• Biochemistry
• Computational Biology
• Proteomics

Background:

• Protein succinylation is a critical post-translational modification regulating cellular functions.
• Accurate identification of succinylation sites is essential for understanding protein roles and developing targeted therapies.

Purpose of the Study:

• To develop an intelligent computational model, iSucc-SnCNs, for accurate prediction of protein succinylation sites.
• To enhance protein function analysis and drug discovery through improved succinylation site identification.

Main Methods:

• Utilized ProtGPT2 for protein sequence encoding.
• Extracted structural features using SMR and PSSM matrices (SMR-HOG, SMR-DCT, PSSM-DWT).
• Employed BTGA+KNN for feature selection and a self-normalized capsule neural network (Sn-CapsNet) for prediction.

Main Results:

• Achieved 92.92% accuracy and 0.96 AUC with iSucc-SnCNs.
• Outperformed traditional models by 17% in prediction accuracy.
• Demonstrated robust generalization on independent datasets (Ind-I and Ind-II) with significant performance improvements.

Conclusions:

• iSucc-SnCNs provides a robust and efficient framework for large-scale succinylation site prediction.
• The model significantly advances protein function analysis in the context of drug discovery.
• Highlights the potential of advanced computational models in post-translational modification research.