DrugCLIP：靶向人类全基因组 药物筛选提速百万倍

Core Insights - Tsinghua University has developed an AI-driven ultra-high-throughput drug virtual screening platform called DrugCLIP, which significantly enhances the speed of new drug screening by a factor of one million and achieves full coverage of human genome-level targets [1][2] Group 1: Breakthrough in Drug Screening - The traditional drug development process faces challenges in matching small molecules to disease targets, with only 10% of potential targets currently being utilized [2] - DrugCLIP transforms the screening process by converting proteins and small molecules into computer-readable signals, allowing for automatic and precise matching without the need for gradual simulation [2][3] - DrugCLIP can perform 31 trillion matching calculations in a single day on a standard high-performance computer, completing the screening of 1 million candidate molecules in just 0.02 seconds [3] Group 2: Practical Applications and Validation - In practical applications, DrugCLIP demonstrated its accuracy by screening 1.6 million candidate molecules for depression-related targets, identifying 100 potential molecules, with 15% showing effective action on the target [4] - The platform successfully identified potential small molecules for a target related to tumors and Parkinson's disease, where no suitable small molecules had been found before, validating its broad applicability [4] Group 3: Future Prospects and Collaboration - DrugCLIP aims to collaborate with research and industry partners to accelerate the discovery of new targets and first-in-class drugs, focusing on areas such as cancer, infectious diseases, and rare diseases [5] - The platform will continue to optimize its engine performance and expand its support modalities to build a more intelligent, efficient, and inclusive global drug innovation ecosystem [5]