Parse Biosciences and Graph Therapeutics formed a strategic partnership to generate a large, functional immune perturbation atlas by combining Graph’s patient‑derived perturbation assays with Parse’s GigLab single‑cell profiling. The program will profile hundreds of millions of cells across disease contexts to map immune responses and support AI‑first target discovery. Graph will run iterative, disease‑contextual perturbations while Parse’s Evercode technology generates high‑throughput whole transcriptome single‑cell datasets. Graph aims to use active learning to prioritize conditions and perturbations that translate to clinical reality. The resource targets a key translational gap: the mismatch between simplified discovery models and patient heterogeneity. Large, standardized single‑cell perturbation datasets can power machine learning models used to predict therapeutic effects and improve target selection.