Left: TFs bind to enhancer and promoter regions depending on the chromatin state. All molecules are drawn at a scale of 1:10⁶. Center left: A simplified gene regulatory network consisting of TFs (nodes of the network graph) and their interactions (edges of the network graph). Center right: A gene regulatory network can be modeled by a dynamical system, which can be represented by a potential energy landscape. The position in the landscape is determined by TF abundance. The shape of the landscape depends on the TF interactions. Cells follow the path of steepest descent to stable states, which correspond to cell types. Right: The collection of trajectories in the potential landscape form a cell-type decision tree, which highlights the hierarchical nature of differentiation.