KIST Research Team Proposes the Most Innovative Candidate Substance for COVID-19 Treatment

Dr. Geunwan Park, KIST

The Korea Institute of Science and Technology (KIST, President Oh Sang-rok) announced that Dr. Park Geun-wan's team at the Natural Product Systems Biology Research Center was selected as one of the top four teams worldwide in the COVID-19 virus target drug candidate discovery category at the 3rd CACHE Challenge. The CACHE Challenge, a drug candidate prediction competition, showcases AI technologies with the highest performance, and all data obtained from the competition are publicly released to support subsequent drug development research. This community-centered drug development paradigm is similar to CASP, the protein structure prediction competition in which AlphaFold participated.

CACHE (Critical Assessment of Computational Hit-finding Experiments) is an international consortium established in 2021 to fairly evaluate the predictive performance of various AI methods for discovering drug candidates. It collaborates with global pharmaceutical companies such as AstraZeneca, Bayer, and Boehringer Ingelheim, and is supported by the Canadian government and the U.S. National Institutes of Health (NIH).

This CACHE Challenge aimed to develop drugs to suppress deadly infectious diseases like the COVID-19 virus, with 23 finalist teams from 11 countries competing over approximately two years. The participating teams proposed a total of 1,739 drug candidate compounds using computer-based AI prediction technologies, with finalists selected in January 2023 and the final results announced in December 2024.

KIST compound prediction model utilized in the Cache challenge.

The KIST research team utilized an AI model called ECBS (Evolutionary Chemical Binding Similarity), which they developed in-house for natural product drug development. ECBS is a novel AI model that improves compound prediction accuracy by leveraging evolutionary information of disease target proteins and is currently used at KIST's Gangneung branch Natural Products Research Institute to discover natural product drug candidates.

In the finals, the evaluation comprehensively considered not only experimental binding data but also compound binding affinity, physicochemical properties, and chemical structure originality. As a result, research teams from KIST (South Korea), the University of Ottawa (Canada), the University of British Columbia (Canada), and Freie Universit?t Berlin (Germany) were selected as the final winners. The teams presented innovative small molecules targeting the Nsp3 protein of the COVID-19 virus, demonstrating the excellence and international competitiveness of KIST's AI-based drug development technology.

KIST expressed expectations that through joint research with domestic and international research institutions and pharmaceutical companies, they will be able to lead AI-based drug development research. Furthermore, it is anticipated that natural product drug development will open possibilities for creating new industries in previously overlooked fields such as rare diseases and intractable disease treatments.

Dr. Park Geun-wan of KIST stated, “This competition has shown the potential for AI-driven drug development research to contribute to solving deadly infectious diseases like coronavirus,” adding, “Based on the results of this competition, we plan to conduct drug development research with global competitiveness by promoting research collaboration and information exchange with domestic and international institutions.”

This research was supported by the Ministry of Science and ICT through KIST’s major projects and the Ministry of Oceans and Fisheries’ National BioResearch Resources Advancement Project (RS-2021-KS211526). The results of this competition were announced on the official CACHE website.

This content was produced with the assistance of AI translation services.