How cells determine or lose their identity is a key question in the study of development and carcinogenesis. Differentiation is regulated by a variety of factors that interact in networks. Intriguingly, some complex differentiation decisions involving many possible outcomes are not easily reduced to a defined series of binary fate decisions. I have approached the mechanism by which cells make such complex decisions in two ways.

First, in a "top-down" approach, I asked which classes of simple network designs could provide sufficiently rich behavior to account for differentiation decisions. Competitive heterodimerization networks, which are present in numerous developmental and physiological contexts, stand out as being particularly flexible. Modeling of these networks suggests unforeseen biological functions.

Second, in a "bottom-up" approach, I started to address a tractable, "real-world" experimental model of a complex differentiation decision. I chose a three-way differentiation decision made in the C. elegans germline, which provides a genetic network that has been extensively characterized. I showed experimentally that differentiation is controlled by positional and timing mechanisms. These results lay the groundwork for a systems biology analysis of differentiation in the C. elegans germline.

An important challenge for the future is to comprehensively characterize a given experimental model, by building on the understanding of simple networks that are amenable to mathematical study.