AI驱动的抗体设计迅速针对新的H5N1流感菌株,加速抗病毒的发展。 AI-driven antibody design rapidly targets new H5N1 flu strain, speeding up antiviral development.

根据2025年11月4日由Vanderbilt牵头在细胞杂志上公布的一项研究,AI-动力蛋白语言模型正在加速抗病毒抗体的创建。 AI-powered protein language models are speeding up the creation of antiviral antibodies, according to a Vanderbilt-led study published in Cell on November 4, 2025. 研究人员使用名为MAGE的模式,设计了针对病毒表面蛋白质的人体抗体,而无需依靠现有的模板,成功地消除了以前未见的H5N1菌株。 Using a model called MAGE, researchers designed human antibodies targeting viral surface proteins without relying on existing templates, successfully neutralizing a previously unseen H5N1 strain. 这种方法利用大赦国际来预测抗体病毒的相互作用,可以大大缩短治疗禽流感和RSV等新出现的威胁的发展时间。 This approach, which leverages AI to predict antibody-virus interactions, could drastically shorten development time for treatments against emerging threats like avian flu and RSV. 华盛顿大学的补充工作引入了RFantibody,这是一个开放源码的AI工具,使用扩散模型设计稳定的抗体粘合器,而麻省理工学院的BoltzGen则提供了另一种由AI驱动的针对以前不可药分子的方法。 Complementary work from the University of Washington introduced RFantibody, an open-source AI tool using diffusion models to design stable antibody binders, while MIT’s BoltzGen offers another AI-driven method for targeting previously undruggable molecules. 这些进展得到了NIH和ARPA-H的支持,突显了学术研究中日益增长的趋势,在学术研究中,开放源码的AI工具正在加速药物发现和扩大传染病、癌症和自免疫紊乱的潜在应用。 These advances, supported by NIH and ARPA-H, highlight a growing trend in academic research where open-source AI tools are accelerating drug discovery and expanding potential applications in infectious diseases, cancer, and autoimmune disorders.