Figure 1

The modeling concept and framework. (a) A feedback link from metabolite 1 to regulator 1 is established (black dash arc). Two perturbation responses are explained by the model. The up regulation of gene 2 by supplying metabolite 1 is explained by the path marked by red arrows (metabolite 1 → regulator 1 → gene 2). The decrease of flux 3 by deleting enzyme (gene) 1 is explained by the path marked by green arrows (enzyme 1 → flux 1 → metabolite 2 → flux 2 → metabolite 3 → flux 3). Edges (metabolite 1, regulator 1) and (regulator 1, gene2) must be both positive or both negative in order to explain the first response. (b) The framework of the joint model. A factor graph model is constructed from metabolic and regulatory networks and perturbation data. A learning algorithm is applied to find the identities and functions of feedback links and active paths. These information are used to predict each perturbation response. By comparing with the observed data, each response is explained or contradicted with the prediction.