Accurate modeling of gene regulation remains a central challenge for whole-cell simulations, where transcriptional control must be represented in a way that is both mechanistic and computationally tractable. Gillespie-based parameter fitting has been widely used to estimate promoter-specific kinetics, but this approach is time-consuming and difficult to adapt when new experimental conditions or promoter types are introduced. Here, we present a Hidden Markov Model (HMM)-based framework that captures transcriptional regulation as a sequence of discrete states—such as unbound DNA, transcription factor binding, and co-activator engagement—linked by experimentally constrained transition probabilities. Fitting is performed directly on RNA polymerase II dwell-time distributions from ChEC-seq data, providing a clear connection between experimental observations and model parameters. This Markovian approach is faster, more stable, and more flexible than Gillespie-based methods. New regulatory elements can be incorporated by expanding the state space, allowing straightforward adjustment to diverse promoter architectures. Applied to budding yeast, the model recapitulates the distinct transcriptional signatures of simple housekeeping promoters, transcription factor-dominated loci, and co-activator-rich ribosomal protein promoters. By providing a modular and data-driven representation of promoter dynamics, this framework improves both the efficiency and accuracy of gene regulatory layers in whole-cell simulations.
Wu et al. (Sun,) studied this question.