Unifying Non-Equilibrium Information Thermodynamics and Genome Engine Dynamics: Maxwell's Demon Control of Cancer Cell Fates

Cell-fate determination demands genome-wide coordination, yet a physical mechanism that orchestrates coherent reorganization across thousands of genes remains unresolved. Here, we establish a data-driven biophysical framework that models genome dynamics as an open, non-equilibrium genome engine by unifying information thermodynamics with genome-engine mechanics, yielding testable control principles for cell-fate change. Time-series transcriptomes from MCF-7 and HL-60 cells exhibit coherent stochastic behavior (CSB), revealing genomic self-organized criticality (SOC). Although single-gene trajectories are noisy, large gene ensembles show reproducible center-of-mass dynamics. Within this SOC regime, a dynamic critical point (CP), a bimodal singular gene ensemble, acts as an internal Maxwells demon (MD) that reciprocally controls the remainder of the genome (peripheral expression system; PES) via information-thermodynamic feedback. This control converts stochastic fluctuations into directed global order. The CP fulfills two coupled roles. Thermodynamically, it modulates entropy production through exchange with the cell environment and tunes CP-PES synchronization. Dynamically, it functions as an SOC-based genome engine controller that coordinates genome-wide reorganization during fate transition. These roles unify in an MD cycle in which the CP converts information-thermodynamic flux into genome engine work, establishing a dissipative, effectively irreversible arrow of time in fate control. Consistent with this directionality, fate commitment occurs in HRG-stimulated MCF-7 and atRA- or DMSO-stimulated HL-60 cells, but not in EGF-stimulated MCF-7 cells. Using a CSB-based mean-field formulation, we unify genome engine mechanics and open information thermodynamics within a single measurable system that yields decision variables and time-gated rules. These rules predict commitment, separate fate from non-fate trajectories, distinguish committed cancer outcomes, and motivate three intervention classes for dynamic cancer control.