The Stochastic Systems Identification Toolkit (SSIT) to Model, Fit, Predict, and Design Single-Cell Experiments

Quantitative analysis of single-cell measurements requires tools to treat cellular variability as part of the signal rather than as nuisance. The Stochastic System Identification Toolkit (SSIT) is an open-source MATLAB package to build, fit, and optimize stochastic models of biochemical reaction networks. The SSIT supports multiple solution schemes—deterministic ODEs, the Stochastic Simulation Algorithm (SSA), direct Chemical Master Equation solutions via Finite State Projections (FSP), basic moment closures, and hybrid model reductions—allowing users to trade accuracy and cost within a common interface. The SSIT handles diverse data (e.g., smFISH, flow cytometry, time lapse-trajectories, or single-cell sequencing), includes utilities to process and model data distortions, and offers likelihood-based and Bayesian parameter inference and diagnostics. Sensitivity analysis and Fisher-information tools quantify identifiability and guide experiments designs for measurement times and perturbations. We show that the SSIT recovers parameters and predictive distributions on benchmarks of experimental gene-expression data in temporally varying environments. The package (https://github.com/MunskyGroup/SSIT) includes command-line and graphical interfaces, examples, and documentation to facilitate community use and extension for reproducible, data-driven modeling of stochastic cellular systems.