Frankhouser, D. E.; Zhao, D.; Fu, Y.-H.; Dey, A.; Chen, Z.; Irizarry, J.; Rangel Ambriz, J.; Branciamore, S.; O'Meally, D.; Ghoda, L.; Trent, J.; Forman, S.; MacLean, A. L.; Kuo, Y.-H.; Zhang, B.; Rockne, R. C.; Marcucci, G.

Single cell RNA sequencing (scRNA-seq) has revolutionized our understanding of cancer, yet identifying meaningful disease states from single cell data remains challenging. Here, we systematically explore the cancer specific information content encoded in single cell versus bulk transcriptomics to resolve this paradox and clarify how discrete disease-defining states emerge from inherently noisy single cell data. Using chronic myeloid leukemia (CML) progression as a model, we demonstrate that, while single cell transcriptomes exist along continuous transcriptional microstates, clinically relevant leukemia phenotypes clearly manifest only at the pseudobulk (macrostate) level. By leveraging state-transition theory, we reveal how robust disease phenotype state-transitions are governed by cell type specific contributions. Our results establish a theoretical framework explaining why discrete disease phenotypes remain hidden at the single cell scale but emerge clearly at the aggregated macrostate level, enabling previously inaccessible biological insights into leukemia evolution. Broadly applicable across cancers and other complex diseases, our approach fundamentally advances single cell analysis by clarifying how microscopic transcriptional variation collectively shapes macroscopic disease dynamics.