Supporting Online Text for

Differences in perspectives about regulatory structures are also related to differences in concept and representation of dynamical processes. Describing individual gene effects begins with identifying individual genes, mechanisms of gene interactions, and pathways of gene products. Describing attractors involves characterizing the convergence of transcriptome wide cell states where majorities of genes determine behavior rather than any one. The difference between low dimensional (few variable) and high dimensional (many variable) collective dynamics is central. Characterizing cellular regulatory networks more generally as distributed control systems where individual genes can exert strong influence requires bridging these two views. We describe a framework in which individual genes and collective states can be considered together to evaluate their mutual influence. The difficulty we overcome is the contrast in the quantities needed to describe the two different pictures. What is needed are analogs of control coefficients, which have been used to study the impact of individual catalysts on system metabolic flows (S1). Identifying such coefficients requires a measure of differences of collective states. Whil