Resource-aware cell factory design through multi-scale metabolic models with expression and proteome constraints

The rational design of microbial cell factories for high bioproduction remains a key challenge in metabolic engineering. While advanced modelling frameworks incorporating protein resource allocation, such as enzyme-constrained models (ecGEMs) and Expression and Thermodynamic Flux (ETFL), provide superior predictive power, they lack practical workflows for strain design. Here, we present StrainOptimizer, a computational platform for rational strain design that integrates resource allocation principles across multiple modelling scales, including expression–metabolism coupling, subcellular compartmentalization, and enzyme capacity. Our benchmarking demonstrates that each principle offers distinct advantages: models coupling metabolism and expression (like ETFL) enable the identification of non-metabolic targets, organelle-level proteomic constraints improve precision for high–protein-cost products, and protein-usage-based objectives consistently outperformed traditional flux-based approaches. To demonstrate its practical utility, we applied StrainOptimizer to an engineered sclareol-overproducing Saccharomyces cerevisiae strain. The platform identified novel bottlenecks in precursor availability that traditional methods missed. Experimental implementation of these targets achieved a 55% success rate, increasing titers by up to 26% and productivity by 45%. StrainOptimizer bridges the gap between resource allocation theory and applied engineering, providing a robust framework currently validated in yeast systems while demonstrating significant potential for expansion to diverse species and substrates.