Director Marketing Communications

Virtual cells go multiscale to predict complex biology. Posted on July 24, 2026 By News Staff Virtual cell models that enable the prediction of cell behavior across scales and biological contexts are rapidly emerging at the forefront of drug discovery. Tom Sercu, PhD, vice president of AI and engineering at Biohub, points to a clinician-scientist studying a rare autoimmune disease as an example of how AI models could reshape translational medicine. Starting from a patient's genome, researchers could use virtual cells to predict how major immune cell types behave in disease versus healthy states. The result offers an invaluable tool across target and mechanism-of-action discovery, patient stratification, toxicity prediction, and therapeutic development. Yet, building a virtual cell is not an easy feat. "Transformative AI in biology does not come from algorithms alone, but when models are trained on large-scale, high-quality, openly accessible datasets," says Sercu. To capture complex biology, such data must span model systems and organisms, interventional and observational methods, and diverse cellular states. To support this mission, Biohub announced a $500 million commitment to the Virtual Biology Initiative in April. The five-year campaign will accelerate the generation of technologies and multi-modal datasets needed to power virtual cell models. Similar to how more than 253,000 experimentally determined molecular structures in the Protein Data Bank (PDB), assembled over five decades, became foundational training data for modern AI protein-structure prediction, Sercu sees an analogous moment for cellular biology. "We do not yet have the equivalent of the PDB for cells," he emphasized. "The Virtual Biology Initiative seeks to change that." Today's virtual cell developers reflect on what's needed for these models to predict complex biology and overhaul drug discovery. Single or bulk. Much of the industry has defined the virtual cell as transcriptome models that predict how perturbations alter gene expression across cellular contexts. Among the increasingly crowded ecosystem, Arc Institute's first-generation virtual cell model, STATE, predicts how stem cells, cancer cells, and immune cells respond to drugs, cytokines, or genetic perturbations. In March, billion-dollar-backed, Xaira Therapeutics unveiled X-Cell, the first scaling law demonstrator in the virtual cell domain, sizing up to a whopping 4.9 billion parameters. These models aim to generalize to unseen biological contexts by training on causal single-cell RNA sequencing (scRNA-seq) data. To train X-Cell, Xaira has spent its initial years building what the company describes as "the largest genome-wide CRISPRi Perturb-seq dataset ever reported." Named X-Atlas/Pisces, the dataset is composed of 25.6 million cells across seven screens and 16 biological contexts. Ginkgo Datapoints, the AI platform division of Ginkgo Bioworks, looks toward bulk transcriptomics rather than a single-cell approach. "Just like how models benefit from diversity in training data, we as an industry benefit from having diversity of approaches," said John Androsavich, PhD, general manager at Ginkgo Datapoints. The Datapoints team applies high-throughput automation to create diverse biological datasets, including cell perturbations, antibody developability, and ADME small molecule developability data, to support AI model training for life science partners. In March, Ginkgo Datapoints delivered the first data release of the Virtual Cell Pharmacology Initiative (VCPI). Approximately 2,280 small molecules were profiled in full dose response using DRUG-seq, a scalable arrayed transcriptomics assay measuring chemical perturbations. In contrast to scRNA-seq, which covers approximately 1,500 genes per cell, DRUG-seq captures nearly 10,000 genes per condition with higher signal-to-noise to optimize insights for pharmacology. Notably, VCPI has exclusively focused on THP-1, a human monocytic cell line widely used across immunology, oncology, and inflammatory disease research, to understand drug action. Across space. While the crowding around scRNA-seq has largely been driven by the pursuit of scale, "a cell is not only its RNA," tempers Hani Goodarzi, PhD, core investigator at Arc Institute. He emphasizes that cells are complex systems shaped by multiple layers of biology beyond gene expression alone, including protein abundance, chromatin state, spatial organization, metabolism, and post-translational regulation. A useful analogy comes from large language models (LLMs), which became powerful as text provided an exceptionally scalable substrate for training trillions of tokens. Yet, text alone is an incomplete representation of human communication. "The lesson is not that one modality is sufficient forever," says Goodarzi, "but that a single high-quality, scalable modality can support general representations when the training corpus is large." Emma Lundberg, PhD, co-founder and CSO at GenBio AI, is worried about the "streetlight effect." "We're scaling what we can and not necessarily what we should," she says. GenBio AI seeks to develop world models that cross multiscale biology. Instead of concentrating on one data modality, the company's so-called "AI-Driven Digital Organism" grows expertise in embedding, tokenizing, and training models across scales, from the molecular layer to regulatory networks. Rather than undergo internal data generation, GenBio AI focuses on public data and partnerships to power the company's models. Lundberg, who is also associate professor of bioengineering and pathology at Stanford University, argues that models can guide the field to which data modalities to pursue. As an example, models that incorporate biological priors, such as protein-protein interactions, can achieve noticeable improvements in predictive performance. Spatial and temporal data also capture critical dimensions of biological function that sequence data alone cannot resolve. According to the Human Protein Atlas, roughly 60% of human genes encode proteins that localize to multiple cellular compartments, often carrying out distinct functions depending on context. In a May preprint posted on bioRxiv, Lundberg and colleagues introduced ProtiCelli, a deep generative model that visualizes the spatial organization of nearly the entire proteome within individual cells. By training on 1.23 million images from the Human Protein Atlas, the model simulates microscopy images for 12,800 human proteins while also generalizing to unseen cell types and drug perturbations absent from training. Through time. Cellular Intelligence is developing a universal virtual cell signaling model designed to simulate cell-state transitions over time, with the goal of expanding the possibilities of regenerative medicine. By learning the underlying "grammar" through which sequences of signaling cues drive cell differentiation, these models aspire to enable the on-demand generation of any cell type. Less than one percent of known human cell types can be reliably produced for downstream applications in cell therapy. As only 20 fundamental molecular signaling pathways give rise to thousands of cell states, researchers face an unfathomably large search space when engineering a particular cell type. "Every cell that we discover or optimize opens a slew of potential applications," said Micha Breakstone, CEO and co-founder of Cellular Intelligence. "One could spend a decade and tens of millions of dollars on painstaking trial-and-error to differentiate a new cell type, or solve this problem in one fell swoop, much like AlphaFold for the protein folding challenge." The company's platform leverages a semi-permeable capsule technology, which selectively retains cells and large analytes while being freely accessible to media, enzymes, and reagents. The method enables high-throughput assays combining live-cell culture with genome-wide readouts. Millions of time-varying signal combinations are tested on human stem cell differentiation in parallel, providing 1,000 times higher efficiency than traditional methods. In May, Cellular Intelligence advanced as a Phase II-ready clinical company after entering an agreement with Novo Nordisk to acquire STEM-PD, an allogeneic cell therapy program for Parkinson's disease with Fast Track Designation. The deal comes six months after Novo announced its strategic exit from the cell therapy space. The start-up's AI cell signaling models will address protocol development, one of the biggest obstacles preventing cell therapies from clinical impact. "Novo selected Cellular Intelligence as the right partner because the next major challenge for complex cell therapy programs is not only the biology," says Breakstone. "It is manufacturing scale-up, comparability, clinical logistics, and commercial readiness." As the diversity of virtual cell models targets new dimensions of complex biology, every approach takes another step closer toward clinical impact.

Cellular Intelligence has appointed two senior executives to its leadership team as the AI-driven biotech company prepares for its next growth phase. Jonathan Alspaugh joins as chief strategy officer, bringing over 15 years of biopharma experience, including roles at RyCarma Therapeutics and Aeglea BioTherapeutics. Adam Weinroth becomes chief marketing officer, with more than 20 years scaling AI and deep-tech platforms, most recently at Form Bio. The Boston-based company is building a foundation model of cell signalling to advance therapeutic applications. In May, it acquired STEM-PD, a Phase 2-ready Parkinson's disease cell therapy programme, from Novo Nordisk. Cellular Intelligence has raised over $70 million from investors including Khosla Ventures, Chan Zuckerberg Initiative, Novo Nordisk, and AMD Ventures.

Novo Nordisk hands a Parkinson's cell therapy to Cellular Intelligence, a startup trying to design cell behavior. Cellular Intelligence has acquired global rights to a clinical-stage Parkinson's cell therapy, from Novo Nordisk, which took an equity stake in the startup and retained milestone and royalty rights. Cellular Intelligence, formerly Somite AI, has raised over $60 million from Khosla Ventures, AMD Ventures, CZI, SciFi VC and others to build foundation models that predict cell behavior across millions of perturbation conditions. The company was incorporated in 2023. The deal comes as cell and gene therapy venture funding thaws after falling from its 2021 peak. Under the agreement, Cellular Intelligence secured global rights to STEM-PD, an allogeneic stem cell-derived therapy designed to replace the dopamine-producing neurons that Parkinson's patients lose. Novo Nordisk took an equity stake in the company and retained milestone and royalty rights. "Our mission is to transform cell biology from trial and error into an engineering discipline," said Micha Breakstone, Ph.D., the company's co-founder and CEO, in an interview. The deal with Novo fulfills something of a longstanding mission of Breakstone's. "I told my wife that [the day he learned of the partnership] is probably the very best day in my career," Breakstone said, "because for the first time it felt that I was much, much closer to the ultimate goal, which is reducing suffering and touching patients' lives." The Novo connection grew out of the broader network Cellular Intelligence has assembled, spanning tech-bio investors, academic biology and large-pharma relationships. "We've engaged with Novo for the last five months, give or take," Breakstone said. Breakstone had known Jacob Petersen, a longtime Novo Nordisk executive. "I had reached out about a year prior, or maybe a little less, to learn about the great industry leaders, and he had immediately captivated me with his vision and his deep understanding of the field," he said. Unmet medical need in Parkinson's remains significant. From left: Nuno Mendonça, Cellular Intelligence's chief medical officer; Malin Parmar, professor of cellular neuroscience at Lund University; Agnete Kirkeby, who led preclinical development of STEM-PD; and Micha Breakstone, Cellular Intelligence's co-founder and CEO. (Photo courtesy of Cellular Intelligence) STEM-PD is a Fast Track-designated asset with IND clearance that targets a significant unmet need. While Parkinson's disease has been medically recognizable for more than two centuries, since James Parkinson described "shaking palsy" in 1817, its therapeutic story has moved slowly. Levodopa, introduced for Parkinson's in 1970, remains the benchmark treatment for motor symptoms. "There are a lot of symptomatic treatments," said Nuno Mendonça, M.D., a board-certified neurologist who recently joined Cellular Intelligence as chief medical officer. "You take them and you improve some of your motor symptoms, but the underlying process goes on. Most of the investigation is devoted to disease modification, and most of it fails." Cell therapy, he said, works on a different principle entirely: "You're basically substituting what the patients are missing." More than 20 treatments have been approved since 2015, many involving new formulations, infusion systems or device refinements such as adaptive deep brain stimulation. The Michael J. Fox Foundation is tracking 151 treatments in clinical testing and has funded $3 billion in research. The economic burden of Parkinson's disease and atypical parkinsonism in the U.S. reached more than $82 billion in 2024, surpassing the $79 billion previously projected for 2037. Yet no approved therapy slows or stops the underlying neurodegeneration. One of the field's most closely watched strategies, targeting alpha-synuclein with monoclonal antibodies, has produced mixed and often disappointing mid-stage results. STEM-PD: From Lund to the clinic. The program Cellular Intelligence is acquiring grew out of more than a decade of research led from Lund University in Sweden, where neuroscientist Malin Parmar, a professor of cellular neuroscience, has developed methods to turn embryonic stem cells into the dopaminergic neurons that Parkinson's patients progressively lose. The STEM-PD trial is a broader academic and clinical collaboration involving Lund University, Skåne University Hospital, University of Cambridge, Cambridge University Hospitals NHS Foundation Trust, Imperial College London and Novo Nordisk. Dopamine-producing neurons, concentrated in a brain region called the substantia nigra, produce dopamine, the neurotransmitter essential for coordinating movement. The therapy entered a first-in-human trial in Sweden in February 2023, with development funded by national and European agencies as well as Novo Nordisk. Nature Medicine named it one of 11 clinical trials expected to shape medicine in 2024. Mendonça arrived through the deal itself. Breakstone said he was initially brought in as a diligence consultant. Mendonça had previously led late-stage EMEA clinical development of Zolgensma, a gene therapy for spinal muscular atrophy, at Novartis Gene Therapies. Six hours versus ten. Cell replacement enables the replacement of the dopamine-producing cells Parkinson's destroys. Manufacturing provides the translation layer. The same cell type has to be produced reproducibly, at clinical quality and in a form surgical teams can administer. That is one area where Cellular Intelligence's AI platform can help. Johanne Press Wegeberg, Ph.D., a Novo Nordisk cell therapy scientist who worked on the STEM-PD program, with Micha Breakstone, Ph.D., co-founder and CEO of Cellular Intelligence. (Photo courtesy of Cellular Intelligence) Stem cell-derived therapies depend on differentiation protocols, or "recipes," that guide pluripotent cells through timed exposures to growth factors and other signals until they acquire the desired identity. Research in human pluripotent stem cell models has shown that signaling history, including duration, can help shape cell fate. That makes protocol design a natural target for Cellular Intelligence's predictive platform. "The protocols that are used for differentiation of cells from pluripotency into any cell fate are extremely sensitive to very minor changes and tweaks," Breakstone said. "Very slight tweaks can end up in outsized deltas in terms of the profile of the cell. You can imagine that an exposure of six hours versus 10 hours to a certain biological growth factor might produce a very different viability window." Cellular Intelligence claims its platform can track those shifts in a way conventional approaches cannot. "Unlike any other company, we're able to track cells over time," Breakstone said. "Our data is temporally resolved. It has context. We know what happens to the cells over time, and we're able to show that those contexts actually deeply matter." He compared the approach to the trajectory of large language models: "This move from static perturbations to temporally resolved inputs and outputs seems to follow the same scaling laws that have brought about this latest revolution in AI with large language models." In Breakstone's example, a 10% increase in viability window would give operators 10% more time between extracting cells from reactors and filling vials, meaning more filled vials or roughly 9% lower cost of goods. Longer viability could also make the injection procedure easier to administer, he said. "Learning about how to ever-so-slightly change the recipe, the protocol of cell differentiation, has a very large impact on attributes of the cells, such as purity, viability, their potentially engrafting properties, and other topics," Breakstone said. "We're placing cells in patients' brains, and you want those cells to be of the best quality," Mendonça said. "You want to be able to manufacture them as well as you can, with as streamlined a process as you can, as off-the-shelf as you can, so that you can then launch it into the unmet clinical need that is PD."