Figure 1 from Cell State Transitions Drive the Evolution of Disease Progression in B-Lymphoblastic Leukemia

Parameterization process for the Markov chain model. A, Schematic of cell states characterized by CD34 and CD38 and the 16 possible transitions between four states. B, Markov transition rate matrix which represents the transition probability from the state given by the row into the state given by the column. C, Newton–Mason search algorithm identifies Markov matrix parameter values in three steps: initial guess, iteration, and ensemble. D, Validation of algorithm convergence. Initial guess has a high residual error, which converges logarithmically toward the target vector until acceptable tolerance is reached (dashed line). Error is measured between steady-state vector v (corresponding to transition matrix M) and the flow cytometry measurements for each patient sample.