AI辅助发现新型抗MRSA抗生素：4-羟基-2,5-二氢噻唑衍生物的设计、合成与评价

《European Journal of Medicinal Chemistry》：4-hydroxy-2, 5-dihydrothiazole derivatives as a new class of small-molecule antibiotics for MRSA: AI-integrated design, chemical synthesis and biological evaluation

本文报道了通过自主开发的AI软件AutoMolDesigner进行虚拟筛选，发现以4-羟基-2,5-二氢噻唑为核心骨架的新型小分子抗生素，其中先导化合物C8对耐甲氧西林金黄色葡萄球菌（MRSA）展现出强效抗菌活性（MIC: 0.5 μg/mL）且无细胞毒性，并通过抑制DNA旋转酶（Gyrase）机制发挥作用，在体内外模型中均表现出良好的治疗效果，为应对抗生素耐药性提供了新策略。

　　

Hit compound B1 identified by AutoMolDesigner-based virtual screening

基于AutoMolDesigner的虚拟筛选所鉴定的先导化合物B1

The workflow of AI-integrated virtual screening is schematically shown in Fig. 2A. Firstly, we used the open-source software, AutoMolDesigner (V1.1) developed by our research group and S. aureus-related datasets (26,428 molecules) curated from ChEMBL23 to build machine learning-based prediction models. The model performed very well, with area under curve (AUC) of 0.914 and Matthews...

人工智能（AI）整合虚拟筛选的工作流程如图2A所示。首先，我们使用本课题组开发的开源软件AutoMolDesigner（V1.1）以及从ChEMBL23整理的与金黄色葡萄球菌（S. aureus）相关的数据集（26,428个分子）来构建基于机器学习的预测模型。该模型表现非常出色，曲线下面积（AUC）达到0.914，马修斯...

Conclusion

结论

Antibiotic resistance poses an increasingly serious threat to global public health. Developing new classes of small-molecule antibiotics is one of the most important ways to address the issue of antibiotic resistance. In this study, with the AI-based antibacterial activity prediction model developed by AutoMolDesigner, we identified compound B1 (SPECS ID: AG-670/15296041) featuring a 4-hydroxy-2,5-dihydrothiazole scaffold as a moderately active small-molecule antibiotic for S. aureus (ATCC...

抗生素耐药性对全球公共卫生构成了日益严重的威胁。开发新型小分子抗生素是解决抗生素耐药性问题最重要的途径之一。在本研究中，利用AutoMolDesigner开发的基于人工智能（AI）的抗菌活性预测模型，我们鉴定出化合物B1（SPECS ID: AG-670/15296041），其以4-羟基-2,5-二氢噻唑为骨架，是一种对金黄色葡萄球菌（S. aureus）（ATCC...）具有中等活性的小分子抗生素。