AIPs-SnTCN: Predicting Anti-Inflammatory Peptides Using fastText and Transformer Encoder-Based Hybrid Word Embedding...

Area of Science:

• Biochemistry
• Computational Biology
• Pharmacology

Background:

• Inflammation is a critical biological response to injury or infection, but prolonged inflammation leads to chronic diseases.
• Current wet-laboratory treatments for inflammation are expensive, time-consuming, and can harm healthy cells.
• Peptide therapeutics offer high specificity for targeting diseased cells, presenting a promising alternative.

Purpose of the Study:

• To develop a highly accurate computational model for predicting anti-inflammatory peptides.
• To leverage advanced machine learning techniques for peptide-based drug discovery.
• To provide a more efficient and cost-effective approach for identifying potential anti-inflammatory agents.

Main Methods:

• Peptide samples were encoded using word embedding (skip-gram, BERT) and conjoint triad features (CTF).
• A fused vector combined word embedding and sequential features to overcome individual limitations.
• Support Vector Machine-Recursive Feature Elimination (SVM-RFE) optimized feature selection.
• An improved self-normalized temporal convolutional network (SnTCN) was employed for model training.

Main Results:

• The AIPs-SnTCN model achieved 95.86% accuracy and 0.97 AUC on training data.
• On an independent dataset, the model attained 92.04% accuracy and 0.96 AUC.
• AIPs-SnTCN significantly outperformed existing models, showing ~19% higher accuracy and ~14% higher AUC.

Conclusions:

• AIPs-SnTCN is a reliable and effective computational tool for predicting anti-inflammatory peptides.
• The model demonstrates significant potential to accelerate pharmaceutical design and research in academia.
• This approach offers a valuable advancement in the development of targeted peptide therapeutics for inflammatory conditions.