LSTMCNNsucc: A Bidirectional LSTM and CNN-Based Deep Learning Method for Predicting Lysine Succinylation Sites

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• Lysine succinylation is a key protein post-translational modification regulating cellular processes.
• Accurate identification of succinylation sites is essential for understanding its functional roles.
• Existing computational methods often overlook the semantic relationships between amino acid residues.

Purpose of the Study:

• To develop an advanced computational method for predicting protein succinylation sites.
• To incorporate semantic residue relationships into a deep learning model.
• To provide a user-friendly web server for practical application of the prediction tool.

Main Methods:

• A hybrid deep learning model integrating Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) was employed.
• The model was trained and validated on protein datasets to identify succinylation sites.
• Enrichment analysis was conducted on succinylated proteins to explore functional conservation.

Main Results:

• The proposed LSTM-CNN model achieved a Matthews correlation coefficient of 0.2508 on an independent test set.
• The method demonstrated superior performance compared to existing state-of-the-art computational approaches.
• Enrichment analysis revealed that succinylation functions are conserved across species, with some species-specific variations.

Conclusions:

• The developed deep learning model offers a significant advancement in predicting protein lysine succinylation sites.
• The findings underscore the conserved and divergent functional roles of succinylation across different species.
• A web server based on this method is available, facilitating research in protein post-translational modifications.