Populations of genetically identical cells display heterogeneity. The noise that arises due to stochastic processes in a system has implications for many cellular processes, including signal transduction and cell-fate decisions. Understanding the mechanisms behind these processes therefore requires an understanding of the generation and processing of noise in a system. We identify, characterise and measure heterogeneity through the variance seen in protein (or other macromolecule) concentrations. In this work, we present a model of variance that has two component parts: intrinsic noise, which is inherent to the system under study, and extrinsic noise, which arises due to factors external to the system. Intrinsic noise results from the stochasticity of reactions and is modelled with the linear noise approximation. Extrinsic noise results from random environmental conditions yielding random rate parameters and is modelled using the unscented transform. Both components can be decomposed into individual reactions, allowing us to attribute the variability seen in an output to the processes that give it rise.
