Cellarity, a biotechnology company focused on developing Cell State-Correcting therapies, has published a groundbreaking study in the journal Science detailing a novel framework that integrates high-dimensional transcriptomics with artificial intelligence to revolutionize drug discovery. The research presents a reproducible and generalizable approach to leveraging AI and multi-omics data to identify therapeutic candidates that restore normal cellular function in complex diseases. At the core of Cellarity’s platform is the understanding that diseases are not driven by single gene mutations but by disruptions in the intricate network of cellular states. The company’s approach maps these states at single-cell resolution using advanced transcriptomic data, capturing the dynamic interplay of gene expression, signaling pathways, and regulatory mechanisms. By applying generalizable AI models, Cellarity links chemical compounds to disease biology, enabling the design of drugs that correct dysfunctional cell states rather than targeting isolated proteins. The study highlights a key innovation: an active, lab-in-the-loop deep learning framework that iteratively refines predictions based on experimental outcomes. This method significantly outperforms traditional phenotypic screening, improving the recovery of phenotypically active compounds by 13- to 17-fold. This dramatic enhancement addresses long-standing challenges in drug discovery, where success rates have stagnated due to oversimplified target-centric models. Parul Doshi, Cellarity’s Chief Data Officer, emphasized the importance of a comprehensive view of cellular states: “We believe a comprehensive view of the cell state will help us create better therapies that can correct the foundational mechanisms of disease.” The platform’s success is underscored by its first clinical candidate, CLY-124, an oral therapy currently in Phase 1 trials for sickle cell disease. CLY-124 works by modulating globin gene expression to restore healthy hemoglobin production, representing a new class of treatment for the condition. Jim Collins, Ph.D., Termeer Professor at MIT and co-founder of Cellarity, noted the shift from single-target to systems-level drug discovery: “Diseases are generally driven by more complex interplay than just a single gene mutation. By analyzing phenotypic connections and polypharmacology, this platform offers strong potential to accelerate discovery and deliver effective oral therapeutics.” In a significant move to advance the broader scientific community, Cellarity is releasing three open-source datasets alongside the publication. The first is a perturbational transcriptomic dataset with over 1,700 samples and 1.26 million single cells, enabling cross-cell-type drug response analysis and benchmarking of perturbation prediction models. The second is a single-cell multi-omic hematopoiesis atlas combining transcriptomics, surface receptor expression, and chromatin accessibility, providing a detailed view of blood cell development. The third dataset tracks the timeline of megakaryocyte differentiation under chemical perturbation, allowing researchers to study maturation trajectories and time-resolved drug effects. These datasets are expected to fuel new research into cellular dynamics, support the development of next-generation AI models, and accelerate drug discovery across the industry. By sharing its data and methodology, Cellarity aims to foster collaboration and innovation in the pursuit of therapies for complex diseases. Founded in 2019 by Flagship Pioneering, Cellarity is advancing a new paradigm in drug development. Its platform is also being applied to other indications in hematology and immunology, with an active collaboration with Novo Nordisk targeting metabolic dysfunction-associated steatohepatitis (MASH). The Science publication not only validates Cellarity’s approach but also sets a new standard for integrating AI and multi-omics in the quest to cure complex diseases.