Epigenetic changes such as DNA methylation and histone modifications do not alter the DNA sequence, but they do affect gene expression. They are bistable: cells choose discrete phenotypes separated by barriers, a picture formalized in Waddington's epigenetic landscape. They also show memory and hysteresis: an epigenetic mark set in response to a stimulus can persist after the stimulus is gone, and erasing it requires crossing a different threshold. These features are hallmarks of dissipative, rate-independent evolution. The work stands somewhat apart: nearby in the archive sit studies of evolution, horizontal gene transfer and cellular processes, but nobody had applied the variational structure of rate-independent systems to epigenetics.
The authors build the model on a triple (Q, E, Ψ): a state space Q (a closed convex subset of $R^n$), a stored-energy functional E that depends on the state and on an external loading S(t), and a dissipative potential Ψ that is convex and positively homogeneous of degree 1 (this property is what makes the evolution rate-independent). They postulate an energy principle: the state minimizes the sum of the energy and the dissipative distance from the previous state. From it they derive the equations: an energy formulation (global stability plus energy balance) and a Biot-type subdifferential inclusion. For the numerical solution they use a variational time integrator based on incremental minimization, which in the scalar case reduces to a two-sided return map.
Existence of energy solutions is established under natural coercivity and continuity assumptions. Uniqueness is proved when the energy landscape is convex, and for nonconvex (bistable) cases the authors introduce balanced-viscosity solutions, which select the physically correct path during epigenetic switching; in the scalar case their uniqueness is proved under the condition that the number of local minima is finite. A variational time integrator is constructed, its convergence to energy solutions is proved, and global energy consistency of first order is shown. On three benchmark problems (a linear play operator, a bistable double well, and a nonlinear convex example) the numerical method reproduces the exact solutions and demonstrates the predicted accuracy.
The work carries the theory of energy and balanced-viscosity solutions from mechanics into epigenetics, giving a thermodynamically consistent framework for modeling epigenetic memory. Unlike neighboring work on evolution and cellular processes, it is the first to use the variational structure of rate-independent systems here, which allows the evolution equations to be derived systematically, without ad-hoc assumptions. This removes a typical difficulty of quantitative epigenetics, namely the arbitrary choice of evolution laws, and guarantees that the laws of thermodynamics hold for any RIE model.
The framework can be extended beyond the rate-independent idealization: adding a quadratic term to the dissipative potential turns the model viscous, which makes it possible to describe relaxation, dependence on stimulus frequency and damage accumulation. Vector-valued state variables (for several coupled epigenetic marks) and nonsymmetric dissipative potentials (to distinguish methylation from demethylation) are natural next steps. It is also promising to treat RIE as the first order of a coarse-grained theory, where the potentials are determined from microscopic data.
The model can be applied to quantitative descriptions of epigenetic switching, for example the epithelial-mesenchymal transition, and to calibrating activation thresholds against experimental data on hypoxia. It also offers a basis for designing therapeutic strategies that target epigenetic marks.
The next step is to apply RIE to a scalar bistable model, where the healthy and damaged phenotypes are two wells, and to calibrate the analytical thresholds and residual "epigenetic scars" against experimental data. It is also necessary to study the finite-viscosity regime and to identify the viscosity parameter from frequency-resolved experiments.
In epigenetics it remains an open question when a quasistatic energy description is legitimate, and how the effective energy relates to the quasipotential landscape obtained from stochastic dynamics. RIE offers a thermodynamically consistent alternative, but the connection between these approaches needs clarification. Also open is the problem of identifiability of effective potentials from macroscopic observables.
In the bistable model the transition between the two wells happens as a jump at a loading equal to the dissipation threshold, regardless of the barrier height; this is an analogue of Maxwell's rule in phase transitions.
The idea of an epigenetic landscape echoes Conrad Waddington's "Epigenetic Landscape," often visualized as hilly terrain down which a ball rolls, standing for a cell.