Subpopulation networks: a scaling principle for engineering living systems

Cell populations underpin applications ranging from bioproduction and biotherapeutics to microbiome engineering and biocomputing. However, treating them as bulk quantities rather than as structured systems limits our understanding of population-level stability, productivity, and robustness. Cellular populations self-organize into coexisting subpopulations (SPs) representing phenotypically distinct states that emerge from the interplay of biological noise and selective pressures. We propose that networks of interacting SPs constitute a fundamental organizational layer for engineering stable cell population dynamics. Surveying published single-cell studies across bacterial, yeast, and mammalian systems, we show that populations consistently self-organize into a limited number of SPs. This dimensional reduction can be captured experimentally by clustering single-cell data and gives rise to scaling laws that link population structure to collective performance.