Fig. 1. AML state-transition model and state-space.

(A) Our state-transition model represents the miRNA transcriptome as a particle undergoing Brownian motion in a double-well quasi-potential. The time evolution of the miRNA transcriptome is represented as movement in the potential. In normal control mice in the absence of an oncogenic event, there exists a large energy barrier between a state of normal hematopoiesis and AML, corresponding to a low probability of state-transition (left). An oncogenic event, such as Cbfb-MYH11 (CM) fusion resulting from inv(16), alters the quasi-potential, reducing the energy barrier and increasing the probability of state-transition (right). The states are characterized by local maxima and minima, labeled c₁, c₂, and c₃, corresponding to perturbed hematopoiesis, an unstable transition, and AML states, respectively. (B) The first two PCs (PC1 and PC2) of a time-series study of miRNA expression in a murine model of CM-driven AML reveals state-transition trajectories that are mapped to critical points in the model.