The immgenT consortium establishes a comprehensive molecular reference framework for mouse T cells by profiling ~800,000 cells across virtually all organs and disease states, revealing through AI-powered integration that T cell diversity occupies a finite landscape of 110 reproducible molecular states organized into eight lineages that are reused and redistributed across tissues and immune perturbations rather than generating context-specific identities, with this framework demonstrating near-saturation when validated against external datasets and providing superior cell identification compared to traditional marker- or signature-based approaches.

The immgenT collaborative project generated a comprehensive molecular atlas of T cells spanning virtually all mouse organs and disease states, profiling ∼800,000 cells from 750 samples with RNA, 128-plex surface protein, and αβTCR sequence. Applying a deep generative model to joint RNA and protein data defined a finite landscape of T-cell states organized into eight lineages and 110 robust clusters, integrating identical cells from different contexts, and resolving prior nomenclatures. Analysis of effector molecules, transcription factors and modules showed that both immunological functions and regulatory programs are shared across cell states. This framework provides a stable, reusable reference, demonstrated by computationally integrating 16 external datasets from diverse biological contexts. A set of public web tools supports browsing of these data, allows mapping of any dataset onto the immgenT framework. These results propose a molecular classification of T cells organized around a set of shared states reused across immunological contexts.