Recently, the startup Shuimu Molecular completed a Series A funding round of tens of millions of dollars. Incubated by Tsinghua University's Institute for Artificial Intelligence Industry Research (AIR), the company aims to construct a large-scale foundational model and a next-generation conversational biopharmaceutical research assistant. The company's products will serve various stages of drug development, including early-stage research, target discovery, molecular design optimization, clinical trial design, and drug repurposing.
Shuimu Molecular focuses on the research and application of large-scale models for the biopharmaceutical vertical industry. In April of this year, Professor Zaiqing Nie's team at AIR open-sourced the lightweight research version of BioMedGPT1.6B, which validated the compression of literature, molecules, proteins, sequencing, knowledge graphs, and other data into a unified multimodal framework. This enables the model to possess the ability of "holistic understanding" and outperforms single-domain models in tasks such as molecular property prediction, drug-target affinity prediction, property prediction, drug sensitivity prediction, molecular-text cross-modal retrieval, and molecular-text cross-modal information generation. The feasibility of the technology has been validated at a research level. Shuimu Molecular will collaborate with the AIR development team to build an industry-ready foundational model base. Additionally, to meet the high confidentiality requirements of the biopharmaceutical industry for private data and operations, the company will provide customers with a private deployment solution, supporting model fine-tuning and deployment based on private data in an easy-to-use, secure, and low training-cost manner. In terms of applications, leveraging the capabilities of the foundational model, the company has successfully developed ChatDD (Chat Drug Design), the world's first next-generation conversational biopharmaceutical research assistant prototype. ChatDD possesses abilities such as natural interaction, intelligent inspiration, and tool invocation, serving as an invaluable intelligent assistant (Co-Pilot) for researchers in various stages. For more details, please click the link below：https://air.tsinghua.edu.cn/info/1007/2060.htm