机器学习与深度学习在疫苗靶点选择、设计、开发及表征中的应用
该综述描述了机器学习和深度学习如何辅助理性疫苗设计,涵盖疫苗靶点选择、设计、开发与表征等环节。文中详述了表位预测、抗原性评分预测、免疫原结构设计、免疫系统建模、疫苗配方优化、计算机模拟临床试验、疫苗安全性与过敏原性预测及组学数据整合等计算工具。文章还概述了数据可用性的局限性,并认为基于ML和DL的新技术将应对最具挑战性的传染病问题。
要点
- ML和DL模型可用于表位预测、抗原性评分及免疫原结构设计等疫苗设计关键步骤。
- 计算工具覆盖疫苗配方优化、计算机模拟临床试验及安全性与过敏原性预测。
- 数据可用性是当前ML和DL应用于疫苗设计的主要局限性之一。
核心结论
- 基于ML和DL的新技术有望应对最具挑战性的传染病问题并惠及全球人群。
研究设计与样本
- 综述类文章,检索范围未在摘要中具体报告。
主要发现
- ML和DL模型已应用于表位预测、抗原性评分预测、免疫原结构设计、免疫系统建模、疫苗配方优化、计算机模拟临床试验、疫苗安全性与过敏原性预测及组学数据整合。
临床怎么用
- 可借助ML和DL计算工具辅助疫苗靶点选择、设计、开发与表征等环节。
局限性
- 数据可用性存在局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
Vaccination is an effective approach that saves the lives of millions of people. Computer tools and technologies have recently helped design novel, safe, effective, and fastest vaccines. Among these tools and technologies, artificial intelligence (AI)-enabled tools and technologies, especially machine learning (ML) and deep learning (DL), with their incredible breakthroughs, have become the key foundation of rational vaccine design. The review aims to describe how ML and DL assist in rational vaccine design. Furthermore, the paper illustrates the vaccine's target selection, design, development, and characterization. The review describes details of different ML and DL models in vaccine design. It also illustrates computational tools, including ML and DL, in key steps of vaccine design, development, and characterizations, such as epitope prediction, antigenicity score prediction, immunogen structure designing, immune system modeling, vaccine formulation optimization, in silico clinical trials of vaccines, vaccine safety, allergenicity prediction, and omics data integration. The article also outlines the limitations of data availability. ML and DL-based new technologies will address the most challenging infectious disease problems and further help people throughout the globe. The article describes about ML's and DL's roles in vaccine development processes, such as vaccine target selection, design, development, and characterization. It will help to fight against infectious disease problems and further help people throughout the globe.
引用
- Bhattacharya M, Chatterjee S, Das A, et al. Machine learning (ML) and deep learning (DL) in vaccine target selection, design, development and characterization. Vaccine. 2025. PMID: 41371096. doi:10.1016/j.vaccine.2025.128051
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。