LG AI Research has signed a joint research agreement with Professor Min-Kyung Baek of the School of Life Sciences at Seoul National University to develop "next-generation protein structure prediction AI" at the Global Lounge of LG Science Park in Magok, Seoul, according to PR Newswire.
Proteins are key biomolecular substances in various activities of the human body, and their structure prediction technology is of great significance for the construction of "digital cell atlas". It can help researchers accurately identify the cause of the disease and greatly accelerate the process of new drug development. At present, many technology giants around the world have made some achievements in protein prediction AI technology, but most of the existing technologies focus on the prediction of single protein structure. However, in reality, there are many "polymorphic" proteins that exhibit multiple forms due to environmental and chemical changes, and the prediction of their structures has become a major problem.
The two sides of this cooperation complement each other's advantages and have strong strength. Professor Min-Kyung Baek, a world-renowned expert in protein structure prediction, has co-developed "RoseTTAFold" with Professor David Baker of the University of Washington, who won the 2024 Nobel Prize in Chemistry for his AI-based research. LG AI Research has grown rapidly since its inception in December 2020, successfully developing EXAONE, a multimodal AI model with 300 billion parameters, in 2021. The model has powerful language and visual data processing capabilities, which provides strong technical support for the study of protein structure prediction.
Their collaboration has a clear goal and plans to develop an AI model for protein polymorphism structure prediction by the end of the year. If successful, this model will push the boundaries of existing technologies and enable scientists to gain a deeper understanding of biological processes to accelerate drug development. In particular, major breakthroughs are expected to be made in overcoming difficult diseases such as Alzheimer's disease, helping to discover pathogenic factors and develop targeted new drugs.
From the perspective of corporate strategy, Gu Guangmo, Chairman and CEO of LG Group, clearly expressed the great importance of the bio business in his 2025 New Year's speech. He expects to drive the biotech business as one of the key drivers of the company's future growth through healthcare solutions and innovative medicines. This partnership is a key step for LG to implement its ABC (AI, Bio, Cleantech) strategy, which will help enhance the company's competitiveness in the global market and achieve sustainable development.
Protein structure prediction AI technology has shown great potential and broad prospects in the current market. From a pharmaceutical development perspective, the traditional drug development process is long and costly, taking an average of 10-15 years and investing billions of dollars. With the help of protein structure prediction AI technology, researchers can quickly identify potential drug targets, greatly shortening the R&D cycle and reducing R&D costs. According to relevant data, the use of AI technology to assist drug development is expected to shorten the R&D cycle by 30% to 50% and reduce the cost by 20% to 40%. As a result, major pharmaceutical companies are investing more in this area, and the demand for AI technology for protein structure prediction continues to grow worldwide.
Figure: LG creates a new generation of AI prediction of protein structure to help drug development (Source: PR Newswire)
In the field of biotechnology, protein structure prediction AI technology also provides key support for emerging technologies such as gene editing and cell therapy. Accurate protein structure prediction can help researchers better understand the relationship between genes and proteins, optimize gene editing strategies, and improve the efficacy and safety of cell therapy. As the biotechnology market continues to expand, so does the reliance on AI technology for protein structure prediction.
From the perspective of market competition, many enterprises and research institutions have emerged in the field of protein structure prediction AI. In addition to large enterprises like LG, some biotech-focused startups have also risen rapidly with their unique technological advantages. For example, AlphaFold, developed by DeepMind in the United Kingdom, has made a major breakthrough in the field of protein structure prediction, and its prediction of protein structure has greatly improved the accuracy of its prediction of protein structure, setting a new benchmark for the industry. The existence of these competitors not only promotes the rapid development of technology, but also intensifies the degree of competition in the market. However, there is still an unmet demand for more advanced and accurate protein structure prediction technologies, especially in predicting the structure of "polymorphic" proteins. LG's joint project with Seoul National University is aimed at this market demand, and it is expected that with its unique technological advantages and innovative research ideas, it will stand out in the fierce market competition and occupy market opportunities.
So, how does protein structure prediction AI work? In simple terms, proteins are made up of chains of amino acids that fold into complex three-dimensional structures that often depend on this unique three-dimensional structure. Protein structure prediction AI uses deep learning algorithms to learn and analyze large amounts of known protein structure and amino acid sequence data. By simulating the physical and chemical interactions during protein folding, such as hydrogen bonding, van der Waals forces, electrostatic interactions, and more, AI models are able to predict the structure of unknown proteins. During the learning process, the model continuously adjusts its parameters to improve its understanding of the structural features of the protein. When a new protein amino acid sequence is inputted, the model predicts the most likely 3D structure of the protein based on the learned rules and patterns, providing key information for subsequent drug development and disease research.
