Measurement error is a complicating factor that could reduce or distort the information contained in an experiment. This problem becomes even more serious in the context of experiments to measure single-cell gene expression heterogeneity, in which important quantities such as RNA and protein copy numbers are themselves subjected to the inherent randomness of biochemical reactions.
Yet, it is not clear how measurement noise should be managed, in addition to other experiment design variables such as sampling size and frequency, in order to ensure that the collected data provides useful insights on the gene expression mechanism of interest.
To address these experiment design challenges, we propose a model-centric framework that makes explicit use of measurement error modeling and Fisher Information Matrix-based criteria to decide between experimental methods. This unified approach not only allows us to see how different noise characteristics affect uncertainty in parameter estimation, but also enables a systematic approach to designing hybrid experiments that combine different measurement methods.