Biohub partners with Google, US to develop universal virtual cell

Biohub: Biohub, affiliated with the Chan Zuckerberg Initiative, advances biomedical research by integrating frontier AI models with experimental biology capabilities. It leads the Virtual Biology Initiative focused on generating standardized, multi-modal biological datasets to enable predictive modeling of living systems. In this news, Biohub serves as the central coordinator partnering with Google entities, Meta, and U.S. federal agencies to develop a universal virtual cell model. Google: Google develops advanced AI systems for scientific domains through DeepMind and Isomorphic Labs, focusing on protein structure, biology, and predictive modeling. It contributes AI expertise, data generation, and standardization efforts in collaborative initiatives. In the reported development, Google partners with Biohub and others to build data foundations enabling virtual cell models. Alex Rives: Alex Rives is head of science at Biohub, overseeing AI-driven biology research and emphasizing empirical data to connect computational models with physical biological outcomes. He has pioneered protein language models and stresses the challenges of scaling AI to full cellular complexity. Rives positions the virtual cell project as the start of a new paradigm where AI enhances the value of laboratory experiments. U.S. Government: The U.S. federal government advances scientific research through agencies including the Department of Energy and National Institutes of Health, which provide funding, datasets, and computing resources for large-scale biology and AI projects. These efforts prioritize open data resources to accelerate discovery across the scientific community. The government’s role in this partnership supports the creation of foundational data for AI models of cellular behavior. AI Data Challenge: Biology requires generating new empirical data from the physical world because much of the information needed for accurate AI models of cells does not yet exist. Open Science Balance: Commercial partners receive temporary exclusive access to developed data to incentivize participation, after which it becomes broadly available as an open resource. Virtual Experimentation: AI models of cells could allow scientists to test and prioritize experiments virtually, reserving costly lab work for the most promising hypotheses.