Title: Calibrating a Parameterized Stochastic Boolean Network Model

 

Abstract: Boolean networks are widely used to model gene regulatory dynamics, but calibrating them from limited data is challenging. This work introduces a stochastic Boolean network framework with activation and degradation probabilities, ensuring a steady state through ergodicity. To overcome exponential complexity, the model is reduced to a lower-dimensional system capturing marginal gene activity. Parameters are estimated from steady-state gene expression data using a regularized optimization approach and simulation-based methods. Experiments on random networks show accurate parameter recovery, demonstrating that the approach can effectively infer biologically meaningful dynamics from sparse data.