Exaone Chemical Agent Deployed for Molecular Design of Obesity Drug
Joint Development of Oral Macrocyclic Peptide with D&D Pharmatech
Leveraging a Drug Discovery System Integrating Multiple Specialized AIs

LG is deploying its self-developed artificial intelligence (AI) for new drug development for the first time. The goal is to use the LG AI Research Center's proprietary drug design AI to create candidate substances for an oral obesity treatment. Unlike previous attempts by large corporations, which mainly focused on building platforms or adopting external technologies, LG is using an in-house model to directly design specific drugs. This is seen as a step forward in enhancing the capabilities of domestic AI-driven drug development.


According to the industry on June 18, the LG AI Research Center has begun joint development with D&D Pharmatech, a company specializing in new drug development, utilizing the 'Exaone Chemical Agent.' The Exaone Chemical Agent is a drug development–focused AI developed by the LG AI Research Center.


This move is noteworthy because LG AI Research Center, which had previously emphasized joint research with external organizations, is now putting its proprietary AI at the forefront of drug development. The process involves LG AI Research Center designing and discovering candidate substances, which D&D Pharmatech then synthesizes, evaluates, and develops into oral formulations. D&D Pharmatech validates the candidates designed by LG, and the results are fed back into the AI to further advance the model.


LG AI Research Center Enters Oral Obesity Drug Race with Specialized AI for Drug Development View original image

The first drug to utilize the Exaone Chemical Agent is confirmed to be an oral obesity treatment aimed at preventing muscle loss during the course of administration. The objective is to address the limitations of existing obesity medications, which often result in both weight and muscle loss, by reducing fat while preserving muscle mass.


Multiple AI models are used in the process of discovering candidate substances. Literature and patent analysis, as well as molecular structure generation, are handled by 'Exaone Discovery,' while molecular design is managed by the Exaone Chemical Agent. At the clinical stage, 'Exaone Path' is also employed to interpret pathology images and assist in patient selection.


The core strategy of the LG AI Research Center is the combination of multiple models. Historically, global AI-driven drug development has been led by single models that maximize specific functions. For example, 'AlphaFold,' developed by Demis Hassabis, CEO of Google DeepMind and recipient of the 2024 Nobel Prize in Chemistry, rapidly predicts protein structures, thereby increasing the efficiency of searching for drug candidates. In the same year, 'RFdiffusion'—developed by Professor David Baker's research team at the University of Washington, who shared the Nobel Prize—went a step further by introducing AI technology that designs protein structures with desired functions.


Just as RFdiffusion performs protein structure design, the Exaone Chemical Agent handles the structural design and generation of candidate compounds for new drugs. The LG AI Research Center plans to enhance discovery efficiency by sequentially linking other specialized models, such as those for literature analysis and target identification.


This 'integrated strategy' will be tested in the development of the current obesity drug, as the targeted candidate substance is a 'macrocyclic peptide,' which is particularly challenging to design. Peptides are chains of amino acids that regulate biological functions. Macrocyclic peptides, with their closed molecular structures, are more stable than traditional peptides. They can also bind precisely to specific targets like antibodies while offering greater potential for oral administration. However, because the two ends of the molecule must be accurately connected and the possible combinations of structures and functions are vast, finding candidate substances using traditional methods is difficult.


The outcome will likely depend on whether such complex structures can be accurately crafted. Although Professor Baker's team recently expanded RFdiffusion and published macrocyclic peptide design techniques in academia, there have not yet been any cases where a macrocyclic drug designed by AI has reached clinical approval.



This is not LG's first foray into drug development. Its affiliate, LG Chem, produced 'Factive,' the first domestically developed antibiotic to receive approval from the U.S. Food and Drug Administration (FDA) in 2003. An official from LG AI Research Center stated, "Through this obesity drug, we aim to demonstrate our capabilities in drug design with tangible results," adding, "The experience gained here can be applied to the development of treatments for other diseases as well."


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

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing