Cellarity published a toxicogenomics framework and an AI model (ToxPredictor) built on DILImap, a large transcriptomics resource, to predict dose‑related drug‑induced liver injury (DILI). The group reported validation sensitivity of 88% at 100% specificity and identification of Phase III safety failures that animal studies missed. The resource and model are open‑source for noncommercial use. Industry observers noted the potential to integrate transcriptomics‑driven safety screening earlier in discovery to reduce late‑stage attrition and to inform dose selection for IND‑enabling studies.