Hit Identification Driven by Combining Artificial Intelligence and Computational Chemistry Methods: A PI5P4K‑β Case Study

Computer-aided drug design (CADD), especially artificial intelligence-driven drug design (AIDD), is increasingly used in drug discovery. In this paper, a novel and efficient workflow for hit identification was developed within the ID4Inno drug discovery platform, featuring innovative artificial intelligence, high-accuracy computational chemistry, and high-performance cloud computing. The workflow was validated by discovering a few potent hit compounds (best IC50 is ∼0.80 μM) against PI5P4K-β, a novel anti-cancer target. Furthermore, by applying the tools implemented in ID4Inno, we managed to optimize these hit compounds and finally obtained five hit series with different scaffolds, all of which showed high activity against PI5P4K-β. These results demonstrate the effectiveness of ID4inno in driving hit identification based on artificial intelligence, computational chemistry, and cloud computing.计算机辅助药物设计（Computer-aided drug design, CADD），尤其是人工智能驱动的药物设计（Artificial Intelligence-Driven Drug Design, AIDD），在药物发现领域的应用日益普及。本文在ID4Inno药物发现平台内开发了一种新颖高效的命中物识别工作流，该工作流融合了创新性人工智能技术、高精度计算化学方法与高性能云计算能力。本研究通过发现针对新型抗癌靶点磷脂酰肌醇5-磷酸4-激酶β（PI5P4K-β）的数种强效命中化合物（最佳半最大抑制浓度IC50约为0.80 μM），对该工作流的有效性进行了验证。此外，借助ID4Inno平台内置的工具，我们成功优化了上述命中化合物，最终获得五个具有不同母核结构的命中化合物系列，所有系列均对PI5P4K-β表现出优异的活性。上述结果证实了ID4Inno平台在基于人工智能、计算化学与云计算开展命中物识别工作方面的有效性。