Answers are grounded in this paper and cite their sources.
Antimicrobial resistance (AMR) has become a significant clinical crisis worldwide, resulting in substantial economic losses and posing a threat to human health. Because antimicrobial peptides (AMPs) can interact with membrane bilayers and inhibit microbial growth, they represent a promising therapeutic candidate to overcome AMR. Despite their potential, AMP development is strongly constrained due to the high costs and inefficiencies in identifying novel and effective AMPs, particularly those with specific functional activities against diverse microbial species. To address these limitations, I developed a comprehensive ensemble learning framework for both AMP identification and functional activity prediction. This approach integrates four amino acid encoding methods (one-hot encoding, BLOSUM62 substitution matrix, AAIndex physicochemical properties, and pseudo amino acid composition) with ensemble learning algorithms to predict 14 different antimicrobial activities, including antibacterial, antifungal, antiviral, anticancer, and other specialized functions. I also evaluated various advanced machine learning algorithms across different encoding approaches and built comprehensive model combinations for each task. The experimental results showed that my ensemble learning-based model achieved the highest performance, with an accuracy of 97.66% for AMR prediction. Additionally, the ensemble approach demonstrated superior performance compared to other individual models. It is able to achieve balanced accuracy across various functional activity tasks. In conclusion, my proposed method offers a practical approach for AMP discovery, helping to address the AMR crisis.
No takes yet. Share an insight, caveat, or question.
Zhirui Liu (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: