MFE-ACVP: anti-coronavirus peptide prediction based on multimodal feature extraction and ensemble learning

Area of Science:

• Biotechnology
• Computational Biology
• Drug Discovery

Background:

• The COVID-19 pandemic highlights the urgent need for effective antiviral therapies.
• Anti-coronavirus peptides (ACVPs) offer promising therapeutic potential due to their targeting specificity and low toxicity.
• Current computational models for ACVP identification are limited by insufficient validated data.

Purpose of the Study:

• To develop an advanced artificial intelligence framework, MFE-ACVP, for efficient identification of potential anti-coronavirus peptides.
• To address the challenge of limited data in ACVP discovery through innovative computational methods.
• To improve the accuracy and generalization ability of ACVP prediction models.

Main Methods:

• Developed an improved Generative Adversarial Network (GAN) with materialization constraints to generate high-quality ACVPs and mitigate data scarcity.
• Engineered a 100-dimensional cross-scale feature representation by integrating sequence, structural, evolutionary, and topological peptide features.
• Constructed an ensemble learning architecture combining five traditional machine learning models with deep neural networks (DNNs) for enhanced predictive performance.

Main Results:

• The MFE-ACVP framework demonstrated superior predictive performance on an independent validation set, achieving an Area Under the Curve (AUC) of 86.37%, Accuracy (ACC) of 77.62%, and Matthew's Correlation Coefficient (MCC) of 65.19%.
• MFE-ACVP outperformed existing models such as PreAntiCoV, iACVP, ACVPred, and ENNAVIA-C/D in terms of predictive accuracy and stability.
• A publicly accessible web server was developed to facilitate the screening of ACVPs.

Conclusions:

• The MFE-ACVP framework represents a significant advancement in computational approaches for identifying anti-coronavirus peptides.
• The study successfully addressed data insufficiency challenges in ACVP prediction using advanced AI techniques.
• The developed tool and methodology are expected to accelerate the discovery and development of novel antiviral peptide therapeutics.