Abstract
Our conception of gene regulation specificity has undergone profound changes over the last 20 years. Previously, regulators were considered to control few genes, recognized with exquisite specificity by a ``lock and key'' mechanism. But recently, genome-wide exploration of regulator binding site occupancy (whether on DNA or RNA targets) revealed extensive lists of molecular targets for every studied regulator. Such poor biochemical specificity suggested that each regulator controls many genes, collectively contributing to biological phenotypes. Here I propose a third model, whereby regulators' biological specificity is only partially due to ``lock and key'' biochemistry. Rather, regulators affect many genes at the microscopic scale, but biological consequences for most interactions are attenuated at the mesoscopic scale: only a few regulatory events propagate from microscopic to macroscopic scale, others are made inconsequential by homeostatic mechanisms. That model is well supported by the microRNA literature, and data suggest it extends to other regulators. It reconciles contradicting observations from biochemistry and comparative genomics on one hand, and in vivo genetics on the other hand, but this conceptual unification is obscured by common misconceptions and counter-intuitive modes of graphical display. Profound understanding of gene regulation requires conceptual clarification, and better suited statistical analyses and graphical representation.