Residual demand bias across regulatory layers in Escherichia coli

Classical demand theory predicts that genes used often should tend toward positive control, whereas genes used rarely should tend toward negative control. We test a weaker version of this idea in Escherichia coli : effective demand may bias, but not determine, regulatory sign. Effective demand was estimated as the fraction of transcriptome conditions in which a regulatory iModulon was active, and signed regulation was summarized at module, regulator, and regulator–gene edge levels. The clearest association appeared at the regulator-weighted layer, especially for local regulons with at most 50 targets, whereas broader and edge-level summaries were positive but less precise. External refitting preserved the positive direction, while sign-shuffle null models did not clearly separate the observed slopes from random sign allocation. Taken together, the analyses are consistent with a modest residual demand-associated tendency that is filtered by regulatory architecture, physiology, and dependence among targets.