BioGeometry Launches GeoFlow V2: The World’s First All-Scenario Atomic-Level AI Protein Design Platform, Pioneering a New Era in AI-Driven Drug Discovery

—The application for each award has started successively, stay tuned—

5Month29The other day, BioGeometry announced its cooperation with Hanhe Biotech. The two parties will focus on generative artificial intelligence (AI) Driven enzyme design, enzyme mining, strain modification, metabolic pathway optimization, cell factory design and other core technologies to carry out in-depth collaboration, promoting“AI+Synthetic Biology”The deep integration application.

Despite being officially operational for less than three years, BioGeometry has becomeApplying Generative AI to Protein DesignThe leading enterprise, whose core team is generativeAIA pioneer in the field of molecular generation, as early as2021Years ago, the diffusion model was applied to the generation of three-dimensional molecular structures, and its key paper is2022YearAITop Ranked by Field Citations50One of the papers.

The World's First Full-Scenario Atomic-Level Protein Large Model

4Month16On [Date], BioGeometry officially released a milestoneGeoFlow V2——The world's first full-scenario atomic-level protein large model. This model unifies the two core capabilities of protein structure prediction and de novo design for the first time at atomic-level precision, providing a one-stop solution for biomacromolecules from understanding to creation.

GeoFlow V2Marks a significant model evolution for BioGeometry. It is the first to achieve the fusion and unification of two core capabilities—protein structure prediction and de novo design—at the atomic level expression, building a full-scenario, universal protein design platform. Essentially,GeoFlow V2No longer just a“Predictor”Or“Generator”, but truly versatile proteins“Design Engine”。

At the practical application level,GeoFlow V2It has already supported the implementation of multiple commercial projects, especially in the fields of antibody drug design, functional protein design, and enzyme-directed evolution. The model can be effectively applied to high-difficulty tasks such as de novo antibody design for specified epitopes, affinity optimization, and structural prediction of multi-conformation complexes.

AlthoughAlphaFold3AndRFDiffusionThese are all outstanding achievements in the industry, but most mainstream models currently focus on single capability directions, such as structure prediction, structure design, sequence design, or function prediction, each operating independently. Looking back at text, images, etc.AIThe evolutionary process in the field also witnessed similar phenomena in its early stages (for example, different models were used for sentiment analysis and text classification), but later evolved into...GPTSuch unified text models have now evolved into multimodal large models capable of understanding and generating both text and images, and even visual language action models that unify vision, text, and behavioral decision-making.

In the life sciences field,GeoFlow V2Is the first large model to integrate structure prediction and de novo design into a unified atomic-level architecture, solving“The limitation of "can only predict" or "can only generate."GeoFlow V2On a low-homology antigen-antibody datasetTop-1The prediction success rate reaches45%, has surpassedChai-1、Protenix、AFM2.3And other models with the highest recognition in the industry.

In addition,GeoFlow V2With powerful multi-modal design capabilities, it not only supports protein design but also handlesDNA/RNA, small molecules and other underlying inputs to meet task requirements in different scenarios. This makes it a“Free Engine for Molecular Design”, its advantages do not come from the simple stacking of a few independent modules, but rather from achieving true breakthroughs in multiple dimensions such as performance, efficiency, and scope of application, pushing protein design forward to“From Understanding to Creation”The era of full-process intelligence.

BioGeometry realized the generativeAIThe huge potential of large models in the field of protein design. Therefore, instead of dispersing efforts on building numerous small models, the company focused early on constructing a multimodal, general-purpose, all-atom large model, deeply integrating with high-throughput wet lab platforms for rapid iteration. This strategic choice ultimately led toGeoFlow V2The birth and a series of successfully implemented project outcomes.

"Integrated Dry and Wet" Two Major Platforms

BioGeometry always insistsAdopting the "Dry-Wet Integration" Strategy`, whose core`The technology platform consists of two parts:

• GeoBiologicsLarge Model Platform:Equipped withGeoFlow V2GenerativeAILarge models provide atomic-level protein design capabilities. The platform supports multi-dimensional modeling analysis from sequence prediction, structure generation, and function optimization to drug affinity, immunogenicity, expressibility, and more.

• High-Throughput Wet Lab Platform:Covering the entire process of automated construction, expression, purification, and functional verification to ensure that design results can be quickly validated, feedback obtained, and optimizations made, thereby feeding back into the process.AIModel.

This dry-wet closed-loop system not only significantly improves the speed and accuracy of model iteration but also helps customers achieve seamless integration.“Artificial Intelligence Design”And“Wet Lab Validation”. This system has been successfully applied in multiple practical projects recently, such as the solubilization modification of transmembrane proteins, multi-target enzyme evolution, and de novo design of functional proteins, accomplishing tasks that traditional computational methods found difficult to突破.

The practice of BioGeometry clearly outlinesAIThe Future of Pharmaceutical Development: An Intelligent R&D System Driven by a Unified, Powerful Atomic-Level Intelligent Design Engine and Seamlessly Integrated with High-Throughput Experimentation.GeoFlow V2Achieved“From Understanding to Creation”The closed-loop system represents not only a technological leap but also a profound reconstruction of the essence of drug development. It signifies that future drug discovery will increasingly begin with precise digital design and undergo rapid iteration and optimization through efficient experimentation. BioGeometry stands at the forefront of this transformation, with its pioneering platform accelerating the comprehensive arrival of this intelligent era of drug development.

PharmaCube Pro

Insights into Global Biomedical Frontiers

Empowering the Transformation of China's Indigenous Innovation Achievements

Media cooperation: 15895423126