“…For example, CellRank builds on RNA velocity and trajectory inference models (such as pseudotime) to predict fate potentials given the stochastic nature of fate decisions ( Lange et al, 2022 ). CellRank has been used to predict fate probabilities and reprogramming outcomes in several developmental systems ( Hersbach et al, 2022 ; Lange et al, 2022 ; Van Bruggen et al, 2022 ; Bono et al, 2023 ; Matsushita et al, 2023 ), whereas applications to cancer systems have mainly focused on trajectory inference for the immune compartment rather than cancer cells ( Xue et al, 2022 ; Friedrich et al, 2023 ; Jainarayanan et al, 2023 ). DeepVelo uses neural networks to learn transcriptomic dynamics, building a model that predicts trajectories, driver genes, and the effect of in silico perturbation on fate decisions ( Chen et al, 2022 ); however, this approach has not yet been applied to cancer systems.…”