"Input a patient's, such as a gastric cancer patient's disease related multiomic data, and the AIGP platform can analyze these data to find a target for this gastric cancer patient, and generate an antibody or even a series of diverse proteins based on this target. Finally, the protein generated by one bond can also be returned to the automated laboratory for synthesis and functional verification." March 31, Song Le, CTO of Baitu Shengke (Beijing) Intelligent Technology Co., Ltd. (hereinafter referred to as Baitu Shengke), introduced to the reporter of Science and Technology Daily. Baitu Biotech's AIGP platform driven by the big model of life science was officially released a few days ago. Song Le said that what was mentioned above is only one of the functions of the AIGP platform, which can also generate a series of proteins that meet the requirements based on shape and physical and chemical properties; The other is to generate the corresponding target binding protein according to a protein target. For example, it can generate a protein for COVID-19, or even an enzyme for petroleum related molecules. It is understood that over the past two years, Baitu Biotech has constructed a cross modal large model "xTrimo" with 100 billion parameters in the field of life sciences. This large model learns the key laws of how proteins form and function, interact, combine, and regulate cell functions from cross species and cross modal life information, thereby cracking the natural language of life - proteins. Based on large models, AI has made significant progress in a series of task algorithms. In addition to being able to better complete basic tasks such as structural prediction, it is also starting to be able to design new proteins based on different questions and answer various life science questions. Currently, the AIGP platform has three types of functional modules, namely F2P (design/optimize proteins based on functional indicators such as structure, function, and exploitability), P2P (design antibodies and other proteins that bind to target proteins such as antigens in a specific manner), and C2P (specify cells, discover target proteins that regulate cell function, and design corresponding regulatory proteins). Previously, the AIGP platform has been undergoing internal testing for a period of time, contributing to the development of multiple high-performance warheads and new functional sensors in the innovative immune regulatory drug ImmuBot. Professor Dong Chen, an academician of the CAS Member and a famous immunologist, believes that AI can solve problems through the analysis and further deduction and application of big data. Song Le also stated that currently, according to the inputs and requirements of different modules, the AIGP platform can design and generate proteins with specific properties in a relatively short time. Starting from June this year, some of its functional modules will be further opened, allowing professional users to directly and independently use AI's protein generation capabilities in more research scenarios, stimulating more life science exploration. (Liao Xinshe)