Noisy Perturbation Models Distinguish Network Specific from Embedding Variability

Recently, measurement technologies allowing to determine the abundance of tens signaling proteins in thousands of single cells became available. The interpretation of this high dimensional end-point time course data is often difficult, because sources of cell-to-cell abundance variation in measured species are hard to determine. Here I present an analytic tool to tackle this problem. By using a recently developed chemical signal generator to manipulate input noise of biochemical networks, measurement of state variables and modeling of input noise propagation, pathway-specific variability can be distinguished from environmental variability caused by network embedding. By employing different sources of natural input noise, changes in the output variability were quantified, indicating that also synthetic noisy perturbations are biologically feasible. The presented analytic tool shows how signal generators can improve our understanding of the origin of cellular variability and help to interpret multiplexed single cell information.