Pop-Corn directly predicts perturbation-driven compositional shifts, outperforming expression-mediated pipelines on held-out perturbations
The work presents Pop-Corn, a method that directly predicts how a perturbation reshapes cell-type and cell-state composition without reconstructing gene expression; the authors find that even models accurately predicting perturbation-induced changes in average gene expression perform poorly at forecasting compositional shifts, while in the primary T-cell benchmark Pop-Corn predicted the overall cell-state composition of held-out perturbations more accurately than the evaluated expression-prediction pipelines and better preserved the diversity of observed cell states; the authors further extend it to intact tissue, predicting perturbation-induced cell-type proportion changes in local cellular neighborhoods and using attention patterns to generate hypotheses about context-dependent cellular