Parse Biosciences and Tahoe Therapeutics announced a collaboration to generate a 300 million single-cell perturbation atlas to train AI models that predict therapeutic outcomes. Parse will use its GigaLab services to produce single-cell transcriptomes across tissues, disease states and thousands of genetic and chemical perturbations; Tahoe will supply Mosaic-perturbed samples. The deal expands on Tahoe’s prior 100M dataset and is tailored to create foundational training material for virtual cell models. Tahoe CSO Johnny Yu framed the project as critical infrastructure for building AI that understands human biology and predicting patient responses. Financial terms were not disclosed; the collaboration follows Parse’s announced acquisition by Qiagen and underscores continued investment in large-scale single-cell datasets as inputs for generative and predictive biology platforms.