人工智能助力药物虚拟筛选提速百万倍

Core Insights - The article discusses the development of DrugCLIP, an AI-driven ultra-high-throughput drug virtual screening platform that significantly enhances the speed and accuracy of drug discovery [1][2]. Group 1: DrugCLIP Platform - DrugCLIP improves screening speed by a factor of one million compared to traditional methods, achieving 10 trillion protein-ligand scoring in one day using 128 CPU cores and 8 GPUs [1]. - The platform has completed virtual screening projects covering approximately 10,000 protein targets and 20,000 protein pockets, analyzing over 500 million drug-like small molecules and identifying more than 2 million potential active compounds [2]. Group 2: Research and Development - The team successfully screened 1.6 million candidate molecules for the norepinephrine transporter, identifying about 100 high-scoring molecules, with 15 showing effective inhibition and 12 having better binding capabilities than a common antidepressant [2]. - The research results were published in the journal "Science," highlighting the potential for innovative drug discovery through genome-level virtual screening [1]. Group 3: Accessibility and Future Directions - The screening database created by DrugCLIP is now freely accessible to the global research community, providing data support for basic research and early drug discovery [2]. - The platform has served over 1,400 users, completing more than 13,500 screenings, and aims to accelerate the discovery of new targets and innovative drugs in areas such as cancer, infectious diseases, and rare diseases [2].