BioGeometry Empowers Vazyme to Successfully Deliver Major Enzyme Engineering Project, Achieving 21-Fold Performance Boost in Just 60 Days!

Jul 16, 2025 19:06 CST Updated 19:06

BioGeometry

Large Molecule Drug Developer

A long-standing challenge in the IVD reagent field, often regarded as a "technical dead zone," is being quietly addressed by a Chinese biotech company.

Recently, BioGeometry announced that an important IVD reagent tool enzyme renovation project, conducted in collaboration with Nanjing Vazyme Biotech Co., Ltd. ("Vazyme"), has been successfully delivered, drawing attention within the industry. The project not onlyAchieved performance improvements that have never been reported in currently available literature and industrial practice., opening up a precise, efficient, and systematic new path for industrial-grade enzyme molecular design and multi-objective optimization.

Breaking the Low-Salt Catalysis "Curse," Generative AI Provides an Unprecedented Answer

The core issue of the project lies in the insufficient catalytic activity of the target enzyme at a specific salt ion concentration, which severely affects the purity and stability of the product.Previously, there was no effective solution in this field that successfully enhanced the salt ion tolerance of such enzymes.

BioGeometry, with its self-developed all-scenario atomic-level protein large model GeoFlow V2, managed to break this "curse" in just 60 days.

The team achieved all the expected goals through three rounds of iteration, and finally increased the product yield of the engineered enzyme under the target salt ion concentration to that of the wild-type enzyme.21 Times, greatly improving production efficiency. Not only that, the enzyme also fully meets or even exceeds standards in terms of specific activity, stability, and purity of the catalyzed product.

An Unprecedented Industry Gap, Why Did BioGeometry Achieve It?

It is reported that GeoFlow V2 is BioGeometry's new generation generative AI large protein model, and alsoThe World's First Generative Large Model Integrating "Structure Prediction + De Novo Design" for Proteins, which can be simultaneously used for two key tasks: protein structure prediction and de novo protein design. It has achieved comprehensive advancements in prediction accuracy, speed, and design rationality, surpassing mainstream industry models such as AlphaFold3 and RFDiffusion in one fell swoop.

For example, performance evaluations for predicting the structures of low-homology antigen-antibody complexes show that,GeoFlow V2's Top-1 Success Rate Reaches 45.19%, significantly surpassing industry-leading models such as Chai-1, Protenix, and AFM2.3, demonstrating a powerful ability to model interactions between biomolecules.

In addition, GeoFlow V2 supports complex scenarios such as specified epitopes, specified binding modes, specified antigen structures, and multi-conformation predictions, truly achieving "Quality by Design" at the source of innovation.

It can be said that this world-first model heralds the entry of AI + biomanufacturing into an intelligent and systematic era — by reconstructing the underlying paradigm of macromolecular design through generative AI.GeoFlow V2 will accelerate the R&D process of new drugs, vaccines, industrial enzymes, and other products, profoundly transforming the innovation landscape in biomedicine and biomanufacturing.

Acceleration in the Industrialization of AI Protein Design: Chinese Technology is Approaching the World's Cutting Edge

As the integration of AI and synthetic biology deepens, the industrialization of de novo protein design is accelerating. The speed and capability demonstrated by BioGeometry in this project send a clear signal to the industry: while achieving breakthroughs at the technological foundation, Chinese teams are rapidly打通从模型、算法到产业交付的全链路.

More enlightening is that this is not the first time BioGeometry has completed such a "quick, precise, and powerful" implementation project. Its self-developed platform, the GeoFlow large model, has already achieved industrial application in several high-barrier fields, including de novo antibody design, transmembrane protein solubilization modification, and multi-objective enhancement of enzyme performance.

It is reported that in the future, BioGeometry will continue to expand GeoFlow's application capabilities in key scenarios such as nucleic acid drugs, vaccine design, and diagnostic reagents, and join hands with more industry leaders to promote the systematic and large-scale development of a generative AI-driven biomanufacturing system.

In the global race to "reshape protein design with generative AI," this young Chinese company is becoming an increasingly indispensable name.

BioGeometry

Scan to follow

Visit the company's official website

https://www.biogeom.com