Abstract
Drug discovery has traditionally been driven by strategies aimed at identifying a single molecular target for a given disease, an approach inspired by Paul Ehrlich's "magic bullet" concept. While enormously productive, this reductionist paradigm has progressively given way to more comprehensive views, first through polypharmacology and multi-target drug design, then through network medicine. This article traces this conceptual evolution through a simple graph-based formalism, arguing that the natural endpoint of this trajectory is a fully systems-level paradigm, here termed System-Based Drug Discovery, in which the organism itself is modeled as a complex, multiscale, and multilayer network. Grounded in complexity science, this framework does not replace but integrates the traditional top-down and bottom-up discovery strategies within a unifying representation. The growing availability of biological data and computing power, together with recent advances in molecular generative models, is already turning this conceptual shift into practice, extending network-based reasoning from target identification and drug repositioning to the actual design of new molecular entities.