Imagine exploring the human genome to find new drugs at lightning speed ⚡️—that's the promise of DrugCLIP, an AI-driven screening platform unveiled yesterday by researchers in the Chinese mainland.
Developed by Professor Lan Yanyan's team at the Joint Center for Computational Health—set up by Tsinghua University's Institute for AI Industry Research (AIR) and the Beijing Academy of Artificial Intelligence (BAAI)—DrugCLIP was published on January 9 in the journal Science.
Chemical genomics aims to find drug candidates for all ~20,000 proteins encoded by our genome. While about 90% of these proteins are linked to disease, most remain "undruggable" because traditional screening methods would take centuries to explore the massive search space.
Instead of simulating molecule–protein fits step by step, DrugCLIP creates a "vectorized binding space" that turns proteins and small molecules into mathematical vectors. Using deep contrastive learning, it transforms complex biochemical interactions into a fast vector retrieval problem—achieving a millionfold speed boost compared with conventional tools 🚀.
First released in June 2025, DrugCLIP is free and open to researchers worldwide. Over 1,000 teams have already completed tens of thousands of large-scale screening tasks across the globe 🌎.
Wang Xiaodong—an academician of the Chinese Academy of Sciences and director of the National Institute of Biological Sciences in Beijing—calls DrugCLIP "transformative," saying it not only accelerates discovery but also expands the chemical space of potential compounds and lowers the barrier to pharmaceutical innovation.
Experts say this marks the start of a "post-AlphaFold era" in drug discovery, defined by large-scale, systematic exploration and open collaboration. The future of new drug development just got a major upgrade!
With DrugCLIP, the hunt for new therapies could move from decades to days. Keep an eye on this space for the next wave of biotech breakthroughs 🔬✨
Reference(s):
Chinese scientists use AI to boost drug screening by a millionfold
cgtn.com
