UC Berkeley Library

Computational models are useful for quantitative elucidation of the dynamical behavior of biological systems. Oftentimes, several competing models (i.e., hypotheses) are proposed to describe the underlying molecular mechanisms of a biological system. Selecting the most representative model is imperative for obtaining meaningful quantitative insights into the dynamics of the system of interest. However, discrimination between competing models poses a significant challenge due to heterogeneity that is intrinsic to biological systems. This paper demonstrates the effectiveness of a probabilistic approach to optimal experiment design for model discrimination in the presence of time-invariant, cell-to-cell differences within a cell population. The JAK2/STAT5 signaling pathway, which is involved in proliferation and differentiation of hematopoietic stem cells, is used as a case study.